#  @theaustinlyons Austin Lyons Austin Lyons posts on X about $nvda, ai, $amd, $intc the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours. ### Engagements: [------] [#](/creator/twitter::9015062/interactions)  - [--] Week [------] +196% - [--] Month [-------] -26% - [--] Months [---------] +400% - [--] Year [---------] +7,617% ### Mentions: [--] [#](/creator/twitter::9015062/posts_active)  - [--] Week [--] +200% - [--] Month [--] -1.60% - [--] Months [---] +150% - [--] Year [---] +3,233% ### Followers: [-----] [#](/creator/twitter::9015062/followers)  - [--] Week [-----] +1.80% - [--] Month [-----] +6.20% - [--] Months [-----] +172% - [--] Year [-----] +431% ### CreatorRank: [-------] [#](/creator/twitter::9015062/influencer_rank)  ### Social Influence **Social category influence** [stocks](/list/stocks) [technology brands](/list/technology-brands) [finance](/list/finance) [automotive brands](/list/automotive-brands) [countries](/list/countries) [fashion brands](/list/fashion-brands) [social networks](/list/social-networks) [cryptocurrencies](/list/cryptocurrencies) [celebrities](/list/celebrities) [events](/list/events) **Social topic influence** [$nvda](/topic/$nvda), [ai](/topic/ai), [$amd](/topic/$amd), [$intc](/topic/$intc), [$crdo](/topic/$crdo), [meta](/topic/meta), [open ai](/topic/open-ai), [$mrvl](/topic/$mrvl), [$avgo](/topic/$avgo), [6969](/topic/6969) **Top accounts mentioned or mentioned by** [@nvidia](/creator/undefined) [@intel](/creator/undefined) [@amd](/creator/undefined) [@qualcomm](/creator/undefined) [@vikramskr](/creator/undefined) [@microntech](/creator/undefined) [@schwabnetwork](/creator/undefined) [@semidoped](/creator/undefined) [@openai](/creator/undefined) [@meta](/creator/undefined) [@benbajarin](/creator/undefined) [@arm](/creator/undefined) [@synopsys](/creator/undefined) [@cadence](/creator/undefined) [@apple](/creator/undefined) [@nvidias](/creator/undefined) [@microsoft](/creator/undefined) [@patrickmoorhead](/creator/undefined) [@chipstrat](/creator/undefined) [@bmw](/creator/undefined) **Top assets mentioned** [NVIDIA Corp. (NVDA)](/topic/$nvda) [Advanced Micro Devices (AMD)](/topic/$amd) [Intel Corporation (INTC)](/topic/$intc) [Credo Technology Group Holding Ltd (CRDO)](/topic/$crdo) [Marvell Technology Inc (MRVL)](/topic/$mrvl) [Broadcom, Inc. (AVGO)](/topic/$avgo) [QUALCOMM, Inc. (QCOM)](/topic/$qcom) [Alphabet Inc Class A (GOOGL)](/topic/$googl) [Amazon.com, Inc. (AMZN)](/topic/$amzn) [Microsoft Corp. (MSFT)](/topic/$msft) [Micron Technology, Inc. (MU)](/topic/$mu) [Metadium (META)](/topic/$meta) [Arista Networks Inc (ANET)](/topic/$anet) [Rivian Automotive, Inc. (RIVN)](/topic/$rivn) [Synopsys, Inc. (SNPS)](/topic/$snps) [Arm Holdings plc (ARM)](/topic/$arm) [Goldman Sachs (GS)](/topic/goldman-sachs) [CoreWeave, Inc (CRWV)](/topic/$crwv) [Cisco Systems Inc (CSCO)](/topic/$csco) [Dell Technologies, Inc. (DELL)](/topic/dell) [Apple, Inc. (AAPL)](/topic/$aapl) [Oracle Corporation (ORCL)](/topic/$orcl) [Taiwan Semiconductor (TSM)](/topic/$tsm) [Astera Labs, Inc. (ALAB)](/topic/$alab) [Tesla, Inc. (TSLA)](/topic/$tsla) [DeepSeek (DEEPSEEK)](/topic/deepseek) [Cadence Design Systems Inc (CDNS)](/topic/$cdns) ### Top Social Posts Top posts by engagements in the last [--] hours "Intel announced that it will hold an investor webinar at 1:30 p.m. PDT on Apr [--] [----] to present the vision and financial framework for the Intel Foundry business including the recast financials and new segment reporting structure" [X Link](https://x.com/theaustinlyons/status/1765513886210412552) 2024-03-06T23:04Z [---] followers, [---] engagements "Look dude I was in the room in [----] when you Jensen outlined the general purpose strategy to me and it made a ton of sense. Its brilliant And now youre like nah its not general purpose anymore @BenBajarin on Nvidias framing shift Episode 59: Discussing All Things AI Accelerators With Austin Lyons @benbajarin and @jaygoldberg and joined by @theaustinlyons to discuss all things AI accelerators and real world implementations. Apple Podcasts: https://t.co/LdoiJJ7MOw Spotify: https://t.co/IIqUBpCttY Episode 59: Discussing All Things AI Accelerators With Austin Lyons @benbajarin and @jaygoldberg" [X Link](https://x.com/theaustinlyons/status/1769914387521827302) 2024-03-19T02:30Z [---] followers, [---] engagements "Nvidia CUDAs moat is surmountable & how AMD can overcome their software woes and compete. Read more 👇🏻 $AMD $NVDA Can AMD Bridge Nvidias Software Moat Thoughts On Transient Moats Betting on Generative AI AMD's Path Forward https://t.co/3mrunB8SOZ Can AMD Bridge Nvidias Software Moat Thoughts On Transient Moats Betting on Generative AI AMD's Path Forward https://t.co/3mrunB8SOZ" [X Link](https://x.com/theaustinlyons/status/1774989106864328790) 2024-04-02T02:35Z [---] followers, [---] engagements "Intel Made Inside America. Its time for an @intel Foundry slogan. Appeal directly to consumers during the Super Bowl March Madness and The Masters. Im no graphic designer or brand expert but desperate times call for desperate measures. Make the PC refresh cycle happen" [X Link](https://x.com/theaustinlyons/status/1837479890972516358) 2024-09-21T13:11Z [---] followers, [--] engagements "@pmarca I asked NotebookLM to generate a podcast covering the Techno Optimism Manifesto. The result was good "He wants us to think about the cost of fear. If we're always focused on what could go wrong are we missing out on what could go right" https://notebooklm.google.com/notebook/5485a54e-29cc-4db2-8223-69c5ae2cb1a1/audio https://notebooklm.google.com/notebook/5485a54e-29cc-4db2-8223-69c5ae2cb1a1/audio" [X Link](https://x.com/theaustinlyons/status/1841493751597789260) 2024-10-02T15:01Z [---] followers, [--] engagements "@magicsilicon @BenBajarin Let go $INTC foundry" [X Link](https://x.com/theaustinlyons/status/1874246430141137107) 2025-01-01T00:09Z [---] followers, [---] engagements "Its not generosityits strategy. ASML is leaving money on the table on purpose fostering win-win relationships with customers while keeping regulatory pressure and future competition away. And yes someday there will be competition be it China or next-gen technologies. But competitors wont have decades of trust built up like ASML. Its the TSMC playbook or the antithesis to everyone hates working with Intel. ASML can easily make gross margins in the [--] - 65% range if they want to theyre just far too generous. https://t.co/g20RWVgMiO ASML can easily make gross margins in the [--] - 65% range if" [X Link](https://x.com/theaustinlyons/status/1875987253534572881) 2025-01-05T19:26Z [---] followers, [----] engagements "Consumer-scale Gemini inference workloads incoming Gemini-powered voice control is coming to Google TV targeting 270M MAUs across Google/Android TV OS. Great for Google's TPU utilizationif its actually useful. Litmus test: Restart that Netflix documentary we didnt finish last night. If Gemini integrates deeply it's game-changing. If it just opens Netflix Useful but less compelling. Still even shaving seconds off adds up with millions of users and could drive meaningful inference demand. From CES: Google is giving third party TVs running Google TV a Gemini and always-on upgrade starting at the" [X Link](https://x.com/theaustinlyons/status/1876313966898507942) 2025-01-06T17:04Z [---] followers, [--] engagements "An underappreciated benefit of on-premises AI is that inference is a fixed cost instead of the variable cost of using a cloud API (e.g. OpenAI's API). @Qualcomm also shows a nice form factor here reminding us that on-premises AI can still be small and sleek. AI moves to the #edge at #CES2025. Our new #AI on-premises solution brings lower costs and more privacy to businesses: https://t.co/eb2nkJyRwB https://t.co/W9ls80tX4x AI moves to the #edge at #CES2025. Our new #AI on-premises solution brings lower costs and more privacy to businesses: https://t.co/eb2nkJyRwB https://t.co/W9ls80tX4x" [X Link](https://x.com/theaustinlyons/status/1876318296129388887) 2025-01-06T17:21Z [---] followers, [----] engagements "Having a hard time keeping up with all the changes Ill listen to this pod on my run tomorrow morning. The final salvo (I'm told for real this time) of the Biden administration's AI/Semiconductor export controls has just been fired. Without hyperbole it is another very big very important move. Jordan and I unpack the new rule in this great convo. The final salvo (I'm told for real this time) of the Biden administration's AI/Semiconductor export controls has just been fired. Without hyperbole it is another very big very important move. Jordan and I unpack the new rule in this great convo" [X Link](https://x.com/theaustinlyons/status/1879717073821249760) 2025-01-16T02:27Z [---] followers, [---] engagements "Startups must invest time into building products; debugging things that "just work" on @nvidia is the antithesis of that. Hence startups are happy to pay a premium to ensure their product development velocity. They're paying for a "turn-key" platform. Yet some startups might have more time than money and thus be willing to take on some development velocity risk. At the same time Hopper rental prices are dropping dramatically and will continue as Blackwell becomes more available so cash-strapped startups might still opt for previous generation Nvidia as the alternative to current gen" [X Link](https://x.com/theaustinlyons/status/1880678184397488424) 2025-01-18T18:06Z [---] followers, [---] engagements "Yes. $QCOM well positioned for edge inference. Phone XR auto laptops and dabbling in on-premise edge servers Open-source distilled models that dont need crazy hardware (aka deepseek 7b) that are still capable has blown the doors wide open for edge inference apple and Qualcomm are major players that will stand to benefit due to their existing hardware advantage Open-source distilled models that dont need crazy hardware (aka deepseek 7b) that are still capable has blown the doors wide open for edge inference apple and Qualcomm are major players that will stand to benefit due to their existing" [X Link](https://x.com/theaustinlyons/status/1884040014532862379) 2025-01-28T00:45Z [---] followers, [---] engagements "Do algorithmic improvements like DeepSeeks mean that margins will shift from hardware vendors to labs Hard to see why. Algorithmic improvements are quickly copied from one lab to another making it hard for them to maintain technological differentiation. Hardware improvements take much longer to copy. [--]. Breakdown of DeepSeek V3 efficiency vs Llama 3: - Better: 11x fewer FLOPs per token thanks to MoE 37B vs 405B activated params - Better: 2x faster numerics fp8 vs bf16 training - Worse: 0.5x flops utilization 16% vs 33% end-to-end MFU* - Neutral: similar hardware [--]. Breakdown of DeepSeek V3" [X Link](https://x.com/theaustinlyons/status/1884061587901743116) 2025-01-28T02:11Z [---] followers, [---] engagements "Intel Products holding Foundry to high standards is excellent. And it must go both ways: Foundry needs Product to be a good customer (minimize hot lots etc) @intel $INTC Another strong point from @MJHolthaus that shows us how she views foundry partner with the lens on best foundry part for @Intel product. This is the right mindset to bring the best product to market regardless of foundry. Best process wins https://t.co/wUQWgf0pSt Another strong point from @MJHolthaus that shows us how she views foundry partner with the lens on best foundry part for @Intel product. This is the right mindset to" [X Link](https://x.com/theaustinlyons/status/1885098846599954935) 2025-01-30T22:52Z [---] followers, [----] engagements "I read the DeepSeek R1 (reasoning model) paper and explored the implications for the semiconductor industry: Inference-time scaling is AIs newest demand vector. More reasoning per query means more compute per query. DeepSeeks open approach opens the door to competition while its distilled models bring reasoning to the edge. Where does this all lead https://t.co/2si2hp5Uaf Inference-time scaling is AIs newest demand vector. More reasoning per query means more compute per query. DeepSeeks open approach opens the door to competition while its distilled models bring reasoning to the edge. Where" [X Link](https://x.com/theaustinlyons/status/1888974752208851277) 2025-02-10T15:34Z [---] followers, [----] engagements "Im looking forward to a future with reasoning agents at my desk and multimodal GenAI in my glasses 😎 #DeepSeek R1 caused ripples. Newly released #AI models plus smaller model sizes and edge inference will lead to innovation in commercial applications in devices powered by @Snapdragon: https://t.co/7qjIf0UG5r #DeepSeek R1 caused ripples. Newly released #AI models plus smaller model sizes and edge inference will lead to innovation in commercial applications in devices powered by @Snapdragon: https://t.co/7qjIf0UG5r" [X Link](https://x.com/theaustinlyons/status/1889374687450304910) 2025-02-11T18:03Z [---] followers, [---] engagements "Nvidias fast roadmap cadence gives it an edge over ASIC rivals says Jensen. Compute networking memory software all evolve rapidly. #Nvidia #Tech $NVDA https://www.chipstrat.com/p/nvidia-4q25-earnings https://www.chipstrat.com/p/nvidia-4q25-earnings" [X Link](https://x.com/theaustinlyons/status/1897682082568208574) 2025-03-06T16:14Z [---] followers, [----] engagements "Awesome first person video views from @Meta Ray Bans At Mobile World Congress I caught up with @ziad_asghar SVP and GM of XR at @Qualcomm for an insightful conversation about on-device AI and its transformative potential for wearable technologies like smart glasses. Partially recorded through our Ray-Ban @Meta glasses our https://t.co/DYvzE8RnGh At Mobile World Congress I caught up with @ziad_asghar SVP and GM of XR at @Qualcomm for an insightful conversation about on-device AI and its transformative potential for wearable technologies like smart glasses. Partially recorded through our" [X Link](https://x.com/theaustinlyons/status/1899217335413542955) 2025-03-10T21:54Z [---] followers, [---] engagements "Tencent dropped a hybrid Transformer+Mamba MoE model 🚀 Introducing Hunyuan-TurboS the first ultra-large Hybrid-Transformer-Mamba MoE model Traditional pure Transformer models struggle with long-text training and inference due to O(N) complexity and KV-Cache issues. Hunyuan-TurboS combines: ✅ Mamba's efficient long-sequence https://t.co/QoCXkS8gUQ 🚀 Introducing Hunyuan-TurboS the first ultra-large Hybrid-Transformer-Mamba MoE model Traditional pure Transformer models struggle with long-text training and inference due to O(N) complexity and KV-Cache issues. Hunyuan-TurboS combines: ✅ Mamba's" [X Link](https://x.com/theaustinlyons/status/1899636681868792283) 2025-03-12T01:40Z [---] followers, [---] engagements "Projecting $1T of annual data center CapEx in [----] $NVDA" [X Link](https://x.com/theaustinlyons/status/1902049983156617316) 2025-03-18T17:30Z [---] followers, [----] engagements "Jensen said this is his favorite slide $NVDA and talking cuLITHO "in another five years time every mask every single lithography will be processed on NVIDIA CUDA"" [X Link](https://x.com/theaustinlyons/status/1902051095364419990) 2025-03-18T17:34Z [---] followers, [----] engagements "What a beautiful flex love this slide @nvidia $NVDA" [X Link](https://x.com/theaustinlyons/status/1902056645573636250) 2025-03-18T17:56Z [---] followers, [----] engagements "Professor Huang explaining scale-up innovations from air-cooled integrated NVLink to disintegrated and liquid-cooled @nvidia $NVDA" [X Link](https://x.com/theaustinlyons/status/1902057768506552568) 2025-03-18T18:01Z [---] followers, [---] engagements ""Tiny 50W CPU" @Arm well done @nvidia $NVDA $ARM" [X Link](https://x.com/theaustinlyons/status/1902066747223818345) 2025-03-18T18:37Z [---] followers, [----] engagements "Introducing Nvidia Photonics @nvidia's first copackaged option silicon photonic system and the world's first [---] terabyte per second CPO $NVDA" [X Link](https://x.com/theaustinlyons/status/1902068975753896370) 2025-03-18T18:45Z [---] followers, [----] engagements "Introducing Nvidia Newton an open-source physics engine in development with Google DeepMind and Disney Research purpose-built for developing robots. @nvidia @GoogleDeepMind $NVDA" [X Link](https://x.com/theaustinlyons/status/1902076915995017667) 2025-03-18T19:17Z [---] followers, [---] engagements "As Jensen said AI datacenters are power-constrained. Every building blocks efficiency counts and memory is no exception. Microns HBM3E 8H cuts power by 30% vs. competitors. That adds up meaningfully when you're running 100K+ GPU clusters. @MicronTech $mu The rise of large language models has challenged us to reimagine how computational systems process information. #HBM3E has emerged as a breakthrough memory technology helping AI models achieve new levels of reasoning and intelligence. Discover how the higher bandwidth and https://t.co/mR1zFifzTe The rise of large language models has" [X Link](https://x.com/theaustinlyons/status/1903145433419166099) 2025-03-21T18:03Z [---] followers, [----] engagements "SK Hynix isnt keeping Samsung in checkits Micron theyre targeting. Lately theres been talk among engineers that Micron is said to be in top form and is pursuing them with alarming intensity. Report: These Days SK Hynix Is More Focused on Restraining Micron Than Samsung SK Hynix isnt keeping Samsung in checkits Micron theyre targeting. Lately theres been talk among engineers that Micron is said to be in top form and is pursuing them with alarming Report: These Days SK Hynix Is More Focused on Restraining Micron Than Samsung SK Hynix isnt keeping Samsung in checkits Micron theyre targeting." [X Link](https://x.com/theaustinlyons/status/1903416730333901024) 2025-03-22T12:01Z [---] followers, [----] engagements "I think the interesting questions here: why did Apple use TPUs for training and why switch to Nvidia now If the GB300s are also for training maybe the TPUs (or the surrounding software ecosystem) were slowing Apple down. If the GB300s are for inference maybe its simply a TCO play - cheaper at scale inference with GB300s and Dynamo" [X Link](https://x.com/theaustinlyons/status/1904527313523322971) 2025-03-25T13:34Z [---] followers, [---] engagements "2nm AMD 🔬 + x86 made in the USA Next-generation @AMD EPYC processor codenamed Venice is the first HPC product in the industry to be taped out and brought up on the TSMC advanced 2nm (N2) process technology. AMD also announced the successful bring up and validation of its 5th Gen AMD EPYC CPU products at TSMCs new fabrication facility in Arizona underscoring its commitment to U.S. manufacturing" [X Link](https://x.com/theaustinlyons/status/1911917075209019619) 2025-04-14T22:58Z [---] followers, [----] engagements "Micron Announces Business Unit Reorganization to Capitalize on AI Growth Across All Market Segments @MicronTech $MU https://investors.micron.com/news-releases/news-release-details/micron-announces-business-unit-reorganization-capitalize-ai https://investors.micron.com/news-releases/news-release-details/micron-announces-business-unit-reorganization-capitalize-ai" [X Link](https://x.com/theaustinlyons/status/1912856144583799084) 2025-04-17T13:10Z [---] followers, [----] engagements "I visited @Qualcomm HQ for the first time and enjoyed the presentations demos and ride-alongs. One idea that got my wheels spinning (pun intended) is the use of GenAI entertainment for kids during road trips. My kids love the surprise and creativity GenAI enables. I can see them drawing on a car screen and watching it come to life via Unreal Engine connecting to Suno and generating music on demand or even creating a storybook with illustrations to be read aloud. Zonal audio allows each child to hear their own personalized experience as well. The possibilities are endless Parental controls are" [X Link](https://x.com/theaustinlyons/status/1912871480397832388) 2025-04-17T14:11Z [---] followers, [----] engagements "Synopsys CEO Sassine Ghazi on stage with LBT to discuss @Synopsys enabling @intel 18A 14A" [X Link](https://x.com/theaustinlyons/status/1917251516588867702) 2025-04-29T16:15Z [---] followers, [---] engagements "@Synopsys @intel Cadence CEO Anirudh Devan says Intel Foundry is a national asset and has great transistor and packaging technology. Encourages LBT to work as one team with the ecosystem @Cadence" [X Link](https://x.com/theaustinlyons/status/1917253369888149510) 2025-04-29T16:23Z [---] followers, [---] engagements "@Synopsys @intel @Cadence Siemens EDA CEO out to discuss 18A 18AP 14AE plus collaboration with Advanced Packaging @Siemens" [X Link](https://x.com/theaustinlyons/status/1917254232702689579) 2025-04-29T16:26Z [---] followers, [---] engagements "@Synopsys @intel @Cadence @Siemens Naga up looking forward to hearing his ideas regarding winning customers for Intel Foundry" [X Link](https://x.com/theaustinlyons/status/1917258709929189676) 2025-04-29T16:44Z [---] followers, [---] engagements "Jensen doing a good job of articulating @nvidia's move from from chip designer to AI infrastructure (system) designer #COMPUTEX" [X Link](https://x.com/theaustinlyons/status/1924302005595668799) 2025-05-19T03:12Z [---] followers, [----] engagements "Elon making a virtual appearance at MS Build to launch @grok on Azure" [X Link](https://x.com/theaustinlyons/status/1924505569706312072) 2025-05-19T16:40Z [---] followers, [----] engagements "500M WAUs for ChatGPT run entirely on Azure @Microsoft @OpenAI" [X Link](https://x.com/theaustinlyons/status/1924886351658549557) 2025-05-20T17:54Z [---] followers, [----] engagements "Will @micron appear in foldable @Apple iPhones "Industrys thinnest LPDDR5X package paving the way for exciting new smartphone designs." $MU $AAPL Micron is thrilled to announce the world's first [--] (1-gamma) node 16GB #LPDDR5X setting a new standard for flagship smartphones. What makesMicron [--] LPDDR5Xa game-changer Smaller: the industrys thinnest package at just 0.61mm perfect for ultra-thin or foldable https://t.co/eEeLuqSycz Micron is thrilled to announce the world's first [--] (1-gamma) node 16GB #LPDDR5X setting a new standard for flagship smartphones. What makesMicron [--] LPDDR5Xa game-changer" [X Link](https://x.com/theaustinlyons/status/1929915385626714560) 2025-06-03T14:57Z [---] followers, [----] engagements "Love the look of the R3X. Every other carmaker copies Tesla designs. This is fresh. Nice work @Rivian R3X at Unreal Fest in Orlando. Excited for more Unreal Engine updates across all Rivian vehicles. https://t.co/TwVySI3Rk3 R3X at Unreal Fest in Orlando. Excited for more Unreal Engine updates across all Rivian vehicles. https://t.co/TwVySI3Rk3" [X Link](https://x.com/theaustinlyons/status/1929931744234316197) 2025-06-03T16:02Z [---] followers, [---] engagements "The AI Factory is here. @nvidia defines the AI factory vision. @Dell turns it into a product. Hardware services financing included. Chatting with @jaygoldberg on @T_h_e_Circuit 🎙👇" [X Link](https://x.com/theaustinlyons/status/1930677355225268546) 2025-06-05T17:25Z [---] followers, [----] engagements "The importance of high-performance memory has never been greater fueled by its crucial role in supporting the growing demands of AI training and inference workloads in data centers. @MicronTech $MU today announced the shipment of HBM4 36GB 12-high samples to multiple key customers. Micron is now sampling our HBM4 36GB 12-high memory to key customers engineered to meet the growing demands of AI training and inference in #datacenter and #cloud environments. Designed for seamless integration with next-gen AI platforms this milestone offers customers a https://t.co/w4lFn3G7yX Micron is now" [X Link](https://x.com/theaustinlyons/status/1932433463203795238) 2025-06-10T13:43Z [----] followers, [----] engagements "Snapdragon for robotics on the horizon With continued auto downturn ADAS companies repositioning to address Robotics while power semis going after DC. Qualcomm sees itself as one of the top [--] players in robotics as many of its technologies like wireless camera AI low power etc. relevant to robotics. (comments at https://t.co/EtDRhOvgQo With continued auto downturn ADAS companies repositioning to address Robotics while power semis going after DC. Qualcomm sees itself as one of the top [--] players in robotics as many of its technologies like wireless camera AI low power etc. relevant to robotics." [X Link](https://x.com/theaustinlyons/status/1932767763808034912) 2025-06-11T11:51Z [----] followers, [---] engagements "Wow @LisaSu and @sama at @AMD Advancing AI" [X Link](https://x.com/theaustinlyons/status/1933230755007869348) 2025-06-12T18:31Z [----] followers, [----] engagements "Apple doesnt need to win the AI race. It needs to win the developer race. The best AI-native experiences wont come from Cupertino. Theyll come from third-party devs building for real workflows unique hobbies and vertical niches Apple will never touch. @Apple's job Make it simple and worthwhile to build those apps. @WWDC25 was a good start: on-device models clean APIs Swift-native frameworks. But the mission isnt done. Dev loyalty is the ultimate moat. Want AI-native personal computing to thrive Give developers great tooling Cut App Store fees Offer GenAI cloud credits Feature indie builders" [X Link](https://x.com/theaustinlyons/status/1934988607716016509) 2025-06-17T14:56Z [----] followers, [----] engagements "Hands-free AI. Oakley x Meta brings AI assistance voice commands and 3K video and looks cool. https://www.theverge.com/news/690133/meta-oakley-hstn-ai-glasses-price-date https://www.theverge.com/news/690133/meta-oakley-hstn-ai-glasses-price-date" [X Link](https://x.com/theaustinlyons/status/1936055207575621915) 2025-06-20T13:35Z [----] followers, [---] engagements "We are increasingly seeing performance use cases with the Ray-Bans like people wearing them on roller coasters cycling and being around water so were trying to lean into that says Alex Himmel the companys vice president in charge of wearables in an interview. The Meta Ray-Bans were supposed to be Metas last non-AR/display glasses. Then they got popular. Now Meta is going upmarket and rolling out Oakley smart glasses. Details here https://t.co/yOdLnQ1gq3 The Meta Ray-Bans were supposed to be Metas last non-AR/display glasses. Then they got popular. Now Meta is going upmarket and rolling out" [X Link](https://x.com/theaustinlyons/status/1936060066307662256) 2025-06-20T13:54Z [----] followers, [---] engagements "🔥 $AMD % 💰 @sama said When you told me the specs for MI450 I thought no way. MI350 delivers: massive memory native FP6 rack density OCP-compliant datacenter-ready MI450 brings: rack-scale coherence (2026) @xai @Meta @OpenAI @cohere are already running workloads and contributing to the Instinct ecosystem🚀 ROCm has real momentum 🛠 Read more 📈 $AMD: FP6 acceleration rack density HBM capacity AI NIC MI350 MI450 roadmap UALink ROCm [--] traction customer wins (@Meta @OpenAI @xai @OracleCloud @cohere) full-stack systems TCO advantage. Read the full @AMD breakdown 👇 https://t.co/V7hzwy46Wm " [X Link](https://x.com/theaustinlyons/status/1937620669648314770) 2025-06-24T21:15Z [----] followers, [----] engagements "🚨 Record quarter for $MU Revenue: $9.3B (+15% QoQ) HBM up 50% QoQ 🔥 Data center revenue doubled YoY Memory matters in this AI era Has it ever mattered more Today Micron reported our Q3 FY25 financial results. Learn more: https://t.co/1fmBBXLw0y https://t.co/iWlla4F36m Today Micron reported our Q3 FY25 financial results. Learn more: https://t.co/1fmBBXLw0y https://t.co/iWlla4F36m" [X Link](https://x.com/theaustinlyons/status/1937967275614822452) 2025-06-25T20:12Z [----] followers, [----] engagements "I spoke with @Heineken at @Microsoft Build. They are very bullish on agentic AI for intelligent automation. They see it as a revenue driver not just cost savings. They've thought deeply about how to adapt company culture to be AI-forward. Also they gave me a beer for asking good questions 🍻😁 From retail and government services to healthcare AI innovation is accelerating across every industry. Incredible to see the breadth of progress across the Netherlands during my recent visit. https://t.co/smSvyIUuXk From retail and government services to healthcare AI innovation is accelerating across" [X Link](https://x.com/theaustinlyons/status/1938299057136713772) 2025-06-26T18:11Z [----] followers, [----] engagements "Enjoyed talking @AMD with @DianeKingHall on @SchwabNetwork this morning Video link in comments. Main points: AMD sees the AI chip market topping $500B by [----] with most growth in inference (running models). AMD is offering a credible alternative to Nvidia focused on value and performance. Shift from training to inference plays to AMDs strengths; big customers already include OpenAI Meta xAI. AMD's technical bets include large memory capacity to support larger models and reduce TCO. Historically strong in hardware AMD was weak in software but theyre catching up via acquisitions new talent and" [X Link](https://x.com/theaustinlyons/status/1939732300230488569) 2025-06-30T17:06Z [----] followers, [----] engagements "Between Modular's world class engineering & SFC's ruthless price optimizations we're committed to creating the highest performance best priced inference in the world. We run a bespoke process with each customer to optimize for your specific requirements and only roll out when the price & performance are hitting your expectations. We've partnered with Modular to create Large Scale Inference (LSI) a new OpenAI-compatible inference service. It's up to 85% cheaper than other offerings & can handle trillion-token scale. We originally created it at the request of a major AI lab to do large scale" [X Link](https://x.com/theaustinlyons/status/1940599936698966275) 2025-07-03T02:34Z [----] followers, [---] engagements "@PatrickMoorhead @nvidia @Cadence @Synopsys Agree that the lifting of EDA restrictions increases the chances of an Nvidia Blackwell chip (B30) shipping to China. Good take" [X Link](https://x.com/theaustinlyons/status/1940808479360668034) 2025-07-03T16:22Z [----] followers, [---] engagements "🧵Nvidia's next frontier is AI in the physical world. Why Visual-Language-Action (VLA) models enable generalist robots that can see understand and act. $NVDA @NVIDIARobotics @nvidia https://t.co/pmvY6E1VJJ https://t.co/pmvY6E1VJJ" [X Link](https://x.com/theaustinlyons/status/1940813527188754598) 2025-07-03T16:42Z [----] followers, [----] engagements "VLAs unlock true real-world flexibility. Forget hardcoded factory bots. Imagine robots that learn new tasks on the fly adapt to chaotic spaces and follow natural language commands. Nvidias full-stack approach makes it possible" [X Link](https://x.com/theaustinlyons/status/1940813529193631758) 2025-07-03T16:42Z [----] followers, [--] engagements "Nvidias three computers strategy: ⚡ Train (DGX) ⚡ Simulate (Omniverse and Cosmos) ⚡ Deploy (Thor) A continuous loop that keeps robots learning and improving. Nvidia controls this loop completely. No one else is close" [X Link](https://x.com/theaustinlyons/status/1940813531320143933) 2025-07-03T16:42Z [----] followers, [--] engagements "This is not just about GPUs. Robots will need on-device transformers lots of memory and advanced sensors in addition to strong local compute. Each robot is like a mobile mini data center. A huge new semiconductor market is emerging and Nvidia is in the lead from an end-to-end systems perspective" [X Link](https://x.com/theaustinlyons/status/1940813532859187591) 2025-07-03T16:42Z [----] followers, [---] engagements "There isn't a customer in the world that would not leap at a viable alternative to $NVDA. I think people underestimate this $AMD might be starting to catch up to $NVDA I HIGHLY recommend this interview with a current high-ranking employee at $DELL talking about $AMD vs $NVDA and the client demand: [--]. According to him $AMD's main problem is ROCm which is immature compared to $NVDA's CUDA and the https://t.co/vcWpuEchqc $AMD might be starting to catch up to $NVDA I HIGHLY recommend this interview with a current high-ranking employee at $DELL talking about $AMD vs $NVDA and the client demand: 1." [X Link](https://x.com/theaustinlyons/status/1940910544606974341) 2025-07-03T23:08Z [----] followers, [----] engagements "+1 for an EV minivan from @Rivian Help This is what it looks like when you try to put [--] kids in an EV. And I have [--] @elonmusk @Rivian let's solve this problem We will be the first in line for an EV passenger van and have lots of future drivers who will need cars 🤩 https://t.co/RLSzYe7oas Help This is what it looks like when you try to put [--] kids in an EV. And I have [--] @elonmusk @Rivian let's solve this problem We will be the first in line for an EV passenger van and have lots of future drivers who will need cars 🤩 https://t.co/RLSzYe7oas" [X Link](https://x.com/theaustinlyons/status/1940923158061896076) 2025-07-03T23:58Z [----] followers, [---] engagements "2026 looks strong for HBM demand. Reasoning-era inference with ever larger context windows for agentic AI will push capacity and bandwidth needs to new highs. Nvidia Rubin & AMD MI450 will drive a massive HBM4 ramp likely joined by TPU v8 Trainium [--] and Intel Jaguar Shores. Chinese hyperscalers (ByteDance Alibaba Baidu) likely scaling up too. Micron holds 5% HBM share today targeting 20%+ in the next few years. HBM is margin accretive and a key growth driver. $MU @MicronTech Micron said last week that it is now shipping HBM to four customers in high volume spanning GPUs and ASIC platforms" [X Link](https://x.com/theaustinlyons/status/1941924071165624457) 2025-07-06T18:15Z [----] followers, [---] engagements "Forget cars. New consumer products for @Apple should be robotics and AR. Robotics is 5-10yr horizon and offers far more growth potential than EVs while AR is 1-3yr horizon still nascent and now supercharged by GenAI. Chinese smartphone maker Xiaomi has pulled off something Apple failed to do despite a decade-long attempt and thats to make a car. @Lindadalew looks at what helped Xiaomi overtake Apples EV ambitions https://t.co/486n2qPKrG https://t.co/th2cEuQrZ5 Chinese smartphone maker Xiaomi has pulled off something Apple failed to do despite a decade-long attempt and thats to make a car." [X Link](https://x.com/theaustinlyons/status/1942040370549866693) 2025-07-07T01:58Z [----] followers, [---] engagements "How long will @xai exist An LLM that can reference X posts and has a different personality is useful but is it a sustainable business xAI raises $10B in debt and equity TechCrunch https://t.co/0QIVsN8uxD xAI raises $10B in debt and equity TechCrunch https://t.co/0QIVsN8uxD" [X Link](https://x.com/theaustinlyons/status/1942184054540927163) 2025-07-07T11:28Z [----] followers, [---] engagements "@asu_PH 😂 true that. Started as bitcoin miner ended up as $9B acquisition. Well played" [X Link](https://x.com/theaustinlyons/status/1942259296927875582) 2025-07-07T16:27Z [----] followers, [--] engagements "Metas debt plan is public and straightforward. Stargate funding is still opaque for me. Eg Oracles $30B/yr OpenAI compute deal who pays (SoftBank) and what happens if OAI demand or SoftBank funding falls short Meta is reportedly looking to raise $29 billion from private capital to help build AI data centers in the US with this consisting of $3 billion in equity and $26 billion in debt. $META Meta is reportedly looking to raise $29 billion from private capital to help build AI data centers in the US with this consisting of $3 billion in equity and $26 billion in debt. $META" [X Link](https://x.com/theaustinlyons/status/1942624331550581060) 2025-07-08T16:38Z [----] followers, [---] engagements "Worlds best devices company is considering putting hardware engineer in charge. Makes sense. Hope he can articulate and stand firm on the AI strategy for said devices inspire talent and push into robotics and glasses. As I reported a year ago hardware engineering chief John Ternus is primed to be the next CEO of Apple when Tim Cook eventually retires. https://t.co/tu7oJnDiwv As I reported a year ago hardware engineering chief John Ternus is primed to be the next CEO of Apple when Tim Cook eventually retires. https://t.co/tu7oJnDiwv" [X Link](https://x.com/theaustinlyons/status/1942706733861986329) 2025-07-08T22:05Z [----] followers, [---] engagements "Bullish on glasses + GenAI After wearing Ray-Ban Meta Wayfarer glasses for a few weeks . I feel kind of naked wearing regular sunglasses. I've found three use-cases that are hard to roll back: * Spontaneous photos of my kids when we're out and about. Any cool pose that has a half-life of [--] seconds I After wearing Ray-Ban Meta Wayfarer glasses for a few weeks . I feel kind of naked wearing regular sunglasses. I've found three use-cases that are hard to roll back: * Spontaneous photos of my kids when we're out and about. Any cool pose that has a half-life of [--] seconds I" [X Link](https://x.com/theaustinlyons/status/1942913528836374966) 2025-07-09T11:47Z [----] followers, [---] engagements "Announced today same day @perplexity_ai shipped their browser. AI-native browser is ground zero for agentic AI will also aim to replace Google as default search Reuters is reporting that OpenAI is about to release an AI-powered web browser to directly compete with Chrome that will 'fundamentally change how consumers browse the web'. https://t.co/ikWyfpt6HZ Reuters is reporting that OpenAI is about to release an AI-powered web browser to directly compete with Chrome that will 'fundamentally change how consumers browse the web'. https://t.co/ikWyfpt6HZ" [X Link](https://x.com/theaustinlyons/status/1943009958548902101) 2025-07-09T18:10Z [----] followers, [---] engagements "@__tinygrad__ @AnushElangovan Genuine question: How important is minimizing idle power when designing a datacenter AI accelerator Wouldn't customers want to limit idle time e.g. a race car designer would care more about active efficiency rather than idle fuel burn" [X Link](https://x.com/theaustinlyons/status/1943013534201483621) 2025-07-09T18:25Z [----] followers, [---] engagements "Unilever uses GenAI to generate marketing assets. Love this story. Enterprises have real use cases for that infrastructure buildout. And it requires storage too. 📈 In my latest for @WSJ I investigate @Unilever's plans to turbocharge its influencer marketing program and make boring toiletries like soap go viral on social media (with a little help from AI). Read here https://t.co/z4LKAE2Rrj In my latest for @WSJ I investigate @Unilever's plans to turbocharge its influencer marketing program and make boring toiletries like soap go viral on social media (with a little help from AI). Read here" [X Link](https://x.com/theaustinlyons/status/1943164716597285148) 2025-07-10T04:25Z [----] followers, [---] engagements "Makes sense. RealSense cameras are widely used robotics is taking off and $INTC is focusing on core biz. Best bet is to spin out and go for it like a startup would. The company realized it may have a better chance keeping up with demand and scaling itself if it spun out of Intel and raised its own capital Orbach said. The company realized it may have a better chance keeping up with demand and scaling itself if it spun out of Intel and raised its own capital Orbach said" [X Link](https://x.com/theaustinlyons/status/1943643574869860392) 2025-07-11T12:08Z [----] followers, [---] engagements "18A is the culmination of Intels [--] nodes in [--] years sprint. A rumored write-off doesnt erase the mission: get back in the game prove RibbonFET + PowerVia in real silicon and set the stage for 14A to truly compete with TSMC. We unpack the strategy risks and next moves. 👇 https://t.co/qdxD4Qx5eZ https://t.co/qdxD4Qx5eZ" [X Link](https://x.com/theaustinlyons/status/1943662616410308907) 2025-07-11T13:24Z [----] followers, [----] engagements "@Beth_Kindig Makes sense. RealSense cameras are widely used robotics is taking off and Intel is tightening the belt to focus on core businesses. It's a good move to spin this out so RealSense can go for it like a startup would" [X Link](https://x.com/theaustinlyons/status/1943663817101074477) 2025-07-11T13:29Z [----] followers, [---] engagements "🚕💥 Goldman Sachs forecasts North Americas AV rideshare market to grow at a 90% CAGR from [--------]. Gross margins for vertically integrated robotaxi operators could reach 4050% driving $3.5B in US AV gross profit by 2030" [X Link](https://x.com/theaustinlyons/status/1943748947022914035) 2025-07-11T19:07Z [----] followers, [----] engagements "We cant let one company control the entire AI infrastructure stack. Ive been hearing the same. Not bearish for Nvidia as there is plenty to go around. But supplier diversity + inference era is good for AMD" [X Link](https://x.com/theaustinlyons/status/1944576383440801795) 2025-07-14T01:55Z [----] followers, [----] engagements "Enjoyed talking @nvidia on @SchwabNetwork with @NPetallides 🏁 $5T race: Nvidia vs Microsoft vs Apple who gets there first 🤖 Robotaxis & humanoid robots push Nvidia beyond datacenters. ✅ Nvidia demand outstrips supply; Nvidia can sell every chip it makes. We also talked @Meta 🕶 Metas GenAI-driven advertising & AR glasses push a new frontier. Video in comments" [X Link](https://x.com/theaustinlyons/status/1944853997095149809) 2025-07-14T20:18Z [----] followers, [----] engagements "Rapid-fire Q&A on these TV spots is hard and Ive got a long way to go. But Ive been using ChatGPTs voice mode to practice and its very helpful. I feed it my talking points and tell it to act as the host firing questions at me. Its the right UI (voice) and infinitely patient. Surprisingly useful Thanks @OpenAI and @ChatGPTapp. Enjoyed talking @nvidia on @SchwabNetwork with @NPetallides 🏁 $5T race: Nvidia vs Microsoft vs Apple who gets there first 🤖 Robotaxis & humanoid robots push Nvidia beyond datacenters. ✅ Nvidia demand outstrips supply; Nvidia can sell every chip it makes. We also talked" [X Link](https://x.com/theaustinlyons/status/1944859140683083832) 2025-07-14T20:38Z [----] followers, [---] engagements "@Nvidia's next waves are autonomy and humanoid robotics. Goldman Sachs sees the North American robotaxi market growing at 90% CAGR to [----]. Morgan Stanley estimates a $5 trillion market for humanoid robots by [----]. Both need massive compute for training simulation and onboard intelligence and Nvidia uniquely covers all three. We see a multi-year possibly decade-long AI infrastructure buildout ahead. Still constrained by supply not demand. As new cycles like robotaxis and robots kick in Nvidia is positioned to benefit across each layer of the stack from edge to cloud. Francisco Bido says" [X Link](https://x.com/theaustinlyons/status/1944870737178734845) 2025-07-14T21:24Z [----] followers, [--] engagements "Is MI308 back on Will there be an @AMD Instinct roadmap for China like H20/B30 $AMD" [X Link](https://x.com/theaustinlyons/status/1944956656460423278) 2025-07-15T03:06Z [----] followers, [---] engagements "@PatrickMoorhead Also $ALAB 🫡" [X Link](https://x.com/theaustinlyons/status/1945119653635727770) 2025-07-15T13:53Z [----] followers, [---] engagements "Big for @nvidia. Nuro is a customer so is Lucid so is Uber. We continue to believe robotaxis / autonomy are huge growth opportunities for Nvidia; DGX for training RTX Pro and Cosmos for simulation Drive Thor for onboard compute. Competition in this market is good MORE: Lucid is also partnering with Nuro and Uber on ROBOTAXIS. The plan is to deploy at least [-----] robotaxis over six years. Uber will purchase and operate Lucid Gravity SUVs outfitted with Nuro Driver technology on its ride-sharing network and is making separate MORE: Lucid is also partnering with Nuro and Uber on ROBOTAXIS. The" [X Link](https://x.com/theaustinlyons/status/1945850095154483326) 2025-07-17T14:16Z [----] followers, [----] engagements "Or @NVIDIARobotics could just buy an analog company and build out a suite of compute + sensors for robotics. Let analog companies be commoditized. Build the high value solution offering: AI compute + analog sensor + software (CUDA)" [X Link](https://x.com/theaustinlyons/status/1945932374291673546) 2025-07-17T19:43Z [----] followers, [---] engagements "$AMAT going Zuck-mode but in Asia: For talent aggressive recruitment is underway with no limitations on scale. Multiple industry insiders stated in unison "Applied Materials is contacting the entire industry to secure high-level talent in advanced packaging. They are gathering so many people that it's as if they are 'sucking up' packaging personnel." In particular it has been identified that the expansion of personnel is centered around the Advanced Packaging Technology (APT) team in Taiwan. Applied Materials' Taiwan APT team is currently the organization leading major equipment development" [X Link](https://x.com/theaustinlyons/status/1946020295434756401) 2025-07-18T01:32Z [----] followers, [----] engagements "Check out this footage of ASML's 2nm EUV tool in the Rapidus 2nm fab: Rapidus Achieves Significant Milestone at its State-of-the-Art Foundry with Prototyping of Leading-Edge 2nm GAA Transistors: First prototypes successfully demonstrate electrical characteristics as the company continues to hit targets leading up to [----] mass-production TOKYO Rapidus Achieves Significant Milestone at its State-of-the-Art Foundry with Prototyping of Leading-Edge 2nm GAA Transistors: First prototypes successfully demonstrate electrical characteristics as the company continues to hit targets leading up to 2027" [X Link](https://x.com/theaustinlyons/status/1946312946641956898) 2025-07-18T20:55Z [----] followers, 29.4K engagements "Why local AI image generation beats @midjourney: ✅ One-time hardware cost zero monthly fees ⚡ Instant results. No queues no Relax Mode 🔒 Private by default (no $60 paywall for stealth) 💼 Unlimited commercial use no fine print 🧠 Fully offline. No Discord no cloud dependency Midjourney charges up to $100+/month for features your laptop can now do for free. AMDs nice demos show whats coming. We'll need continued innovation on local models to make them better faster and easier to use. But if we assume the future will be better than the present we can see where this is going. Nice work @AMD" [X Link](https://x.com/theaustinlyons/status/1947358566748459285) 2025-07-21T18:10Z [----] followers, [----] engagements "Now they want to scale beyond the $500B Stargate commitment. Love the ambition but skeptical about the capital flow. "we are planning to significantly expand the ambitions of stargate past the $500 billion commitment we announced in january" we have signed a deal for an additional [---] gigawatts of capacity with oracle as part of stargate. easy to throw around numbers but this is a _gigantic_ infrastructure project. some progress photos from abilene: https://t.co/JfutuoYvn9 we have signed a deal for an additional [---] gigawatts of capacity with oracle as part of stargate. easy to throw around" [X Link](https://x.com/theaustinlyons/status/1947641690795479198) 2025-07-22T12:55Z [----] followers, [---] engagements "AMD's AI hardware is competitive. Software is catching up. The next hurdle for enterprises: integration. Nvidia offers a turnkey enterprise AI stack. AMD by contrast offers a toolkit. Some assembly required. They seem to have partner resources like webinars and training courses to help. Would like to hear more from enterprises using Instinct. Maybe Silo AI helps with that enterprise sales engineering too "Silo AI brings a team of world-class AI scientists and engineers to AMD experienced in developing cutting-edge AI models platforms and solutions for large enterprise customers including" [X Link](https://x.com/theaustinlyons/status/1947662514390491139) 2025-07-22T14:18Z [----] followers, [----] engagements "@levie Agree. Plus contributions from Agentic AI autonomy robotics. https://www.chipstrat.com/p/nvidias-climb-to-4t-and-the-peaks https://www.chipstrat.com/p/nvidias-climb-to-4t-and-the-peaks" [X Link](https://x.com/theaustinlyons/status/1947722513628135441) 2025-07-22T18:16Z [----] followers, [----] engagements "Nvidia is worth $4T. But this isnt the top. A 🧵on how @nvidia got here and whats fueling the next climb. $NVDA 👇 Inference is exploding. Autonomy & robotics are trillion-dollar markets. Annual cadence + constrained supply preserve growth. China is an opportunity for autonomy and robotics but a risk too. The $4T climb is just another waypoint for @nvidia $NVDA 👇 https://t.co/nN2cPgTiQR Inference is exploding. Autonomy & robotics are trillion-dollar markets. Annual cadence + constrained supply preserve growth. China is an opportunity for autonomy and robotics but a risk too. The $4T climb is" [X Link](https://x.com/theaustinlyons/status/1947748742578192678) 2025-07-22T20:01Z [----] followers, [----] engagements "1. Nvidia made the GenAI wave possible. @nvidia was the only company ready to scale when ChatGPT went viral. Training clusters All Nvidia. But wouldnt rivals move quickly to challenge its dominance in large-scale training They couldnt. Nvidias scale-up and scale-out interconnect hardware is only half the battle. Efficient training at scale requires ultra-low-latency lossless communication across accelerators. Thats where software comes in. Nvidias NCCL makes frontier-scale training even possible. So when the GenAI wave hit Nvidia was the only supplier with a production-ready massively" [X Link](https://x.com/theaustinlyons/status/1947748744835043549) 2025-07-22T20:01Z [----] followers, [--] engagements "3. The Optimistic View (which played out) Alternatively when Nvidia hit $1T in [----] optimists believed ROI would accrue quickly as models were deployed and CapEx would continue to grow. Something like this (see example below). Fortunately for @nvidia this is the path were on Inference is here and while ROI is hard to quantify its definitely compounding. Were off the shaky ground of "training scaling hope" and into the inference era. 👇 $NVDA" [X Link](https://x.com/theaustinlyons/status/1947748751109722252) 2025-07-22T20:01Z [----] followers, [--] engagements "4. The game has shifted. Inference is the engine of the AI economy now and the name of the game. Training matters too but inference is 📈 ChatGPT alone handles 2.5B queries per day. Enterprises are deploying GenAI across workflows and they want ROI not benchmarks and research papers. Heres an eye-opening anecdote from @IsabelleBiscuit at the Wall Street Journal. Accounting firm RSM is finding enough value in generative AI to justify a fivefold increase in its GenAI investment Why 👇 @nvidia $NVDA" [X Link](https://x.com/theaustinlyons/status/1947748753726689587) 2025-07-22T20:01Z [----] followers, [--] engagements "5. Enter agentic AI and reasoning models. These models plan use tools and take actions across multi-step workflows. They generate 510x more tokens per query. They make GenAI more useful. Agentic AI was the core unlock for RSM. They connected multiple tasks into automated workflows powered by agents that can take action or make decisions. This shift increased productivity significantly. When AI can reason @nvidia GPUs generate more value across more use cases. But theres an even more important implementation detail. Agentic thinking increases the amount of tokens generated per query often by 5" [X Link](https://x.com/theaustinlyons/status/1947748755786371532) 2025-07-22T20:01Z [----] followers, [--] engagements "6. Constrained supply and annual upgrades are secretly brilliant. One underappreciated driver of Nvidias sustained revenue growth is the persistent constraint on supply. The AI infrastructure buildout is unfolding more slowly than demand suggests largely due to TSMCs limited CoWoS advanced packaging capacity and the time-intensive process of building liquid-cooled datacenters with sufficient power access. (Unless you're @elonmusk building Colossus [--] and 2). This is actually good for Nvidia It keeps prices high stretches demand across multiple quarters and gives customers time to realize" [X Link](https://x.com/theaustinlyons/status/1947748757787062391) 2025-07-22T20:01Z [----] followers, [---] engagements "9. More Growth Ahead But heres why I think the future still looks compelling and why remarkably the worlds largest company still has room to grow. 👇 @nvidia $NVDA" [X Link](https://x.com/theaustinlyons/status/1947748764774547644) 2025-07-22T20:01Z [----] followers, [--] engagements "10. Autonomy Robotics China Nvidias next growth wave spans autonomy robotics and Chinaeach a massive market in its own right. Goldman sees 90% CAGR in one segment. Morgan Stanley sees trillions in another. And China is key to Nvidias success but adds risk. At the center of it all is Nvidias Three-Computer thesis: train simulate deploy. And its already taking off. 👇" [X Link](https://x.com/theaustinlyons/status/1947748766397960596) 2025-07-22T20:01Z [----] followers, [---] engagements "@IntTradeNetwork @nvidia Yes 💯lack of focus on the strategic points that matter" [X Link](https://x.com/theaustinlyons/status/1947789213157474401) 2025-07-22T22:41Z [----] followers, [--] engagements "A good reminder for folks outside the tech bubble that ChatGPT isnt as pervasive as it seems; yet this growth is a sign it someday will be. Takes time. Note that this asks if theyve *ever* used it not regularly use it Pew have updated their ChatGPT polling from a year ago. Usage has roughly doubled since summer [----]. The share of employed adults who use ChatGPT for work has roughly tripled over the same period to 28%. Two thirds of US adults have still not yet used ChatGPT. https://t.co/bY0Gghv7Tg Pew have updated their ChatGPT polling from a year ago. Usage has roughly doubled since summer" [X Link](https://x.com/theaustinlyons/status/1948019144265220443) 2025-07-23T13:55Z [----] followers, [---] engagements "Generative video is becoming popular So many of you are loving turning your photos into short videos in the @Geminiapp and theGemini API. Next up well be rolling this feature out to @YouTube Shorts and @GooglePhotos.And soonRemixyour Google Photos into comics sketches + 3D animations. https://t.co/ct8AYvJpA7 So many of you are loving turning your photos into short videos in the @Geminiapp and theGemini API. Next up well be rolling this feature out to @YouTube Shorts and @GooglePhotos.And soonRemixyour Google Photos into comics sketches + 3D animations. https://t.co/ct8AYvJpA7" [X Link](https://x.com/theaustinlyons/status/1948057331214028917) 2025-07-23T16:27Z [----] followers, [---] engagements "Yes agree the aggressive financing is necessary. just that it might not all come through depending on the customer. Meta can self-fund so Zucks buildout claims are credible. OpenAI depends on several patrons each with their own funding sources and commitments . if Im Oracle I temper expectations. (They didnt even seem to post this announcement to their newsroom so I think they recognize this) xAI sits in the middle. Elon has Tesla/X/SpaceX revenue (though unclear how often he can tap it) plus unconventional financing. If he says hes building 550K GPU Colossus [--] odds are it happens" [X Link](https://x.com/theaustinlyons/status/1948282667876729064) 2025-07-24T07:22Z [----] followers, [--] engagements "Ben Thompson absolutely nails it here. The insight that the CFO is typically the bottleneck for disruptive innovation yet in Googles case that dam has seemingly broken is sharp well-argued and seems right. 7-24-2025 ($) Google Earnings Google Flips the Switch on Cloud Search Notes https://t.co/69vJUThVop 7-24-2025 ($) Google Earnings Google Flips the Switch on Cloud Search Notes https://t.co/69vJUThVop" [X Link](https://x.com/theaustinlyons/status/1948358530055966739) 2025-07-24T12:24Z [----] followers, [----] engagements "Cadence Earnings Q2 2025: Rev: $1.275B (+20% y/y) EPS: $1.65 (non-GAAP) $0.59 (GAAP incl. $140M legal charge) FY25 guide raised Segment growth: EDA +16% (71% of rev) IP +25% (13%) System Design & Analysis +35% (16%) @Cadence $CDNS" [X Link](https://x.com/theaustinlyons/status/1949942847546748986) 2025-07-28T21:19Z [----] followers, [----] engagements "Chiplets demand new tools. 3D-IC and heterogeneous integration break traditional flows. Chiplets need full-system floorplanning thermal + interconnect optimization. Cadence is betting on agentic AI to handle the complexity" [X Link](https://x.com/theaustinlyons/status/1949948804871942398) 2025-07-28T21:43Z [----] followers, [---] engagements "Agentic AI scales design Cadence says Cerebrus AI Studio can now manage 30-50M instance SoC designs up from 3-5M. Thats a 10x leap in scale just as chiplets' integration push limits of human-led flows" [X Link](https://x.com/theaustinlyons/status/1949949085646983379) 2025-07-28T21:44Z [----] followers, [--] engagements "Big gain for Nvidia (big loss for Groq) Hosted my first hackathon @lifebypixels is going to change the game at @NVIDIAAIDev and how we engage with the community Hosted my first hackathon @lifebypixels is going to change the game at @NVIDIAAIDev and how we engage with the community" [X Link](https://x.com/theaustinlyons/status/1949977144378929648) 2025-07-28T23:35Z [----] followers, [---] engagements "Everywhere theres AI theres Micron. Today Micron announced a new portfolio of SSDs purpose-built for AI workloads with three industry firsts that raise the bar for performance capacity and efficiency in AI data centers. 🧵 $MU @MicronTech" [X Link](https://x.com/theaustinlyons/status/1950208746111553833) 2025-07-29T14:56Z [----] followers, [----] engagements "Why it matters: AI models are bigger context windows longer and token throughput demands higher. Storage must scale in performance and density while cutting energy use. Microns G9 NAND plus vertical integration delivers speed efficiency and security at every stage of the AI data pipeline. $MU @MicronTech" [X Link](https://x.com/theaustinlyons/status/1950208755364184479) 2025-07-29T14:56Z [----] followers, [---] engagements "From data lakes transformation training inference Micron now has an SSD for every stage: [----] ION: massive capacity for capture & ingest 7600: low-latency transformation 9650: PCIe Gen6 throughput for training/inference Everywhere theres AI theres Micron. $MU @MicronTech" [X Link](https://x.com/theaustinlyons/status/1950208756941226210) 2025-07-29T14:56Z [----] followers, [---] engagements "Local GenAI at frontier scale is real. Heres the proof 🧵 @AMD runs @AIatMetas 109Bparameter Llama4Scout with toolcalling locally on a Ryzen AI MAX+ laptop. Enabled by: Large memory in consumer silicon MCP dev tooling Quantized models with huge context windows" [X Link](https://x.com/theaustinlyons/status/1950315797198622923) 2025-07-29T22:01Z [----] followers, [----] engagements "Yesterday's cloud models are here at the edge. Models at this scale usually need multiGPU datacenter rigs with 80GB+ VRAM each. @AIatAMD pulled it off with: 96GB Variable Graphics Memory Optimized llama.cpp via Vulkan 4bit quantization CPU + GPU + NPU acceleration in Ryzen AI MAX+" [X Link](https://x.com/theaustinlyons/status/1950315799698428365) 2025-07-29T22:01Z [----] followers, [--] engagements "AMD Variable Graphics Memory lets Ryzen AI 300series (and newer) CPUs turn a chunk of system RAM into dedicated VRAM for the iGPU" [X Link](https://x.com/theaustinlyons/status/1950315801585865017) 2025-07-29T22:01Z [----] followers, [--] engagements "DISCLAIMER: The video is sped up 5X. So the demo is useful and on a laptop but it is still time-consuming. Yet there are often use cases like Deep Research where it's fine to fire off a query and wait for it to run in the background. On the other hand painfully long wait times are an argument for more powerful AI workstations or even little desktop AI servers. Keep it local but much faster" [X Link](https://x.com/theaustinlyons/status/1950315804580335666) 2025-07-29T22:01Z [----] followers, [--] engagements "The other local breakthrough: long context length. Many local LLMs cap at 4K32K tokens. This ran at 256K tokens (Flash Attention + KV cache Q8). Thats enough to keep an entire SEC filing research paper or large codebase in memory while reasoning" [X Link](https://x.com/theaustinlyons/status/1950315806266380568) 2025-07-29T22:01Z [----] followers, [--] engagements "MCP + longcontext local inference changes whats possible. MCP lets an LLM call APIs browse and automate apps. Long context means it can remember all intermediate results and instructions in one session chaining actions together without losing state" [X Link](https://x.com/theaustinlyons/status/1950315807772189131) 2025-07-29T22:01Z [----] followers, [--] engagements "Of course it's private too. Lots of work family and personal use cases where I might as well keep information photos and videos private if I can" [X Link](https://x.com/theaustinlyons/status/1950315809416581200) 2025-07-29T22:01Z [----] followers, [---] engagements "MCP + long context + big local memory = private capable agentic AI that works across tools and steps. No cloud required. AI PCs might be a real pullforward in [----] at this rate not just hype. Need developers to make that come true though Thats it for today. $AMD @AMD" [X Link](https://x.com/theaustinlyons/status/1950315810880442398) 2025-07-29T22:01Z [----] followers, [---] engagements "@zephyr_z9 @boztank The caveat is that OAI doesn't have a sustainable business (yet). @Meta does. Zuck can afford the CapEx and OpEx; he's for sure in this game in [--] years. OpenAI depends on Microsoft Softbank Oracle benevolence. Less clear how that plays out in [--] years" [X Link](https://x.com/theaustinlyons/status/1950555586472669430) 2025-07-30T13:54Z [----] followers, [---] engagements "No Apple No problem says Qualcomm. "CEO Cristiano Amon: We're forecasting fiscal twenty twenty five to be the second consecutive year of greater than 15% YoY growth in total QCT non-Apple revenues" $QCOM @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1950661531647332526) 2025-07-30T20:55Z [----] followers, [----] engagements "Qualcomm re-enters the AI datacenter market. Strategy: ARM Orion CPUs + NPUs optimized for inference efficiency AlphaWave interconnect IP and NVIDIA NVLink Fusion compatibility. In talks with a hyperscaler; targeting revenue in FY28. $QCOM @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1950662688909955466) 2025-07-30T21:00Z [----] followers, [----] engagements "Qualcomm is formally entering the robotics game. Target: $1T TAM in the next decade. Snapdragons heterogeneous compute (CPU+GPU+NPU) lowlatency connectivity and sensor fusion built for automotive maps well to humanoids & industrial robots. $QCOM @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1950664461691584540) 2025-07-30T21:07Z [----] followers, [----] engagements "Qualcomms AI datacenter push: Orion ARM CPUs for hyperscalers & inference head nodes AI accelerator cards + full racks focused on tokens/$ & tokens/W AlphaWave IP for leading scaleout connectivity In advanced talks with a major customer; aiming for a "halo effect" to validate platform $QCOM @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1950665468983046624) 2025-07-30T21:11Z [----] followers, [---] engagements "$ARM Q1 FY26 revenue hit $1.05B. Best Q1 ever 2nd highest quarter in company history. Royalty revenue $585M (+25% YoY). Licensing $468M. AI demand is driving record performance. @Arm" [X Link](https://x.com/theaustinlyons/status/1950666693925056858) 2025-07-30T21:15Z [----] followers, [---] engagements "Microsoft CEO Satya Nadella: "It was a very strong close to what was a record fiscal year for us. All up Microsoft Cloud surpassed $168B in annual revenue up 23%. The rate of innovation and the speed of diffusion is unlike anything we have seen. To that end we are building the most comprehensive suite of AI products and tech stack at massive scale." @Microsoft $MSFT" [X Link](https://x.com/theaustinlyons/status/1950671544448885165) 2025-07-30T21:35Z [----] followers, [----] engagements "$ARM says demand for its Compute Subsystems (CSS) exceeded expectations [--] customers in market; royalties [--] Arm v9 [--] new CSS licenses this quarter (2 DC [--] PC) Xiaomi X Ring [--] & Samsung Galaxy Flip [--] ship on CSS-based Exynos [----] New Zena CSS targets AI-driven automotive Major OEM commits to multi-gen Arm GPU for flagship phones @Arm" [X Link](https://x.com/theaustinlyons/status/1950672933698867707) 2025-07-30T21:40Z [----] followers, [----] engagements "$ARM CEO says @Nvidia is giving them a massive boost: from nearly zero share in Hopper (x86 head nodes) to almost total share in Blackwell (Grace Blackwell). Combined with steady wins in generalpurpose data centers Arm says hyperscaler CPU share is now nearing 50%. @Arm" [X Link](https://x.com/theaustinlyons/status/1950675759426572567) 2025-07-30T21:51Z [----] followers, [----] engagements "The [----] narrative: 💀 AI will kill Google Search 💀 Search ads will collapse 💀 Google is doomed [----] reality: 📈 Search revenue is growing 📈 Ads are thriving 📈 GenAI is additive What happened Read it all here👇 Why GenAI is Good for Google. Also: Google Search vs. ChatGPT Google vs Meta Risks. 👇 https://t.co/Xn5DdetSdq https://t.co/wKQOFq1Pf7 Why GenAI is Good for Google. Also: Google Search vs. ChatGPT Google vs Meta Risks. 👇 https://t.co/Xn5DdetSdq https://t.co/wKQOFq1Pf7" [X Link](https://x.com/theaustinlyons/status/1950965722114367846) 2025-07-31T17:04Z [----] followers, [----] engagements "https://www.chipstrat.com/p/why-genai-is-good-for-google https://www.chipstrat.com/p/why-genai-is-good-for-google" [X Link](https://x.com/theaustinlyons/status/1950965723607535700) 2025-07-31T17:04Z [----] followers, [---] engagements "🔥🔥🔥 "The winners may not be the ones with the best models but the ones with the best financing terms. When the debt comes due having the smartest chatbot wont matter if you cant pay the bills." This is an underappreciated but important angle The AI arms race isn't just about chips it's about who can finance fastest. $CRWV flooding markets w/ debt $NVDA powering trillion-dollar capex OpenAI's Stargate backed by banks xAI doubling GPUs. The real race is in credit markets. Full breakdown: https://t.co/k8AtbG4JWY The AI arms race isn't just about chips it's about who can finance fastest. $CRWV" [X Link](https://x.com/theaustinlyons/status/1951022737482195153) 2025-07-31T20:50Z [----] followers, [----] engagements "Strong growth but AWS is late to GenAI and still supply constrained. The risk is that Azure and GCP scale faster and lock in share before AWS can close the gap. From Andy Jassy on the AWS earnings call: "In the rapidly evolving world of generative AI AWS continues to build a large fast growing triple digit year over year percentage multibillion dollar business with more demand than we have supply for at the moment." $AMZN $GOOG $MSFT" [X Link](https://x.com/theaustinlyons/status/1951031568245366948) 2025-07-31T21:25Z [----] followers, [----] engagements "Jassy's rebuttal to "youre growing slower than Azure/GCP: Growth % can mislead when youre already much bigger. Azure is 65% the size of AWS so even with lower % growth it can add considerable revenue in dollars. Also calls on other value props of AWS like performance reliability etc $AMZN $MSFT $GOOG" [X Link](https://x.com/theaustinlyons/status/1951035453295173853) 2025-07-31T21:41Z [----] followers, [---] engagements "A better Siri is nice but the real unlock is empowering 3rdparty devs including local GenAI via Foundation Models Framework. Tim Cook says Apple is putting all of our energy behind AI. Hopefully that extends to the dev ecosystem. Some AFM talent has already been poached $AAPL @Apple" [X Link](https://x.com/theaustinlyons/status/1951038324237598983) 2025-07-31T21:52Z [----] followers, [---] engagements "GenAI will transform productivity apps. Will the incumbents reinvent their own software or will startups beat them to it We are working on productivity (apps) now because that is where the most value is created. You can make a lot of money doing this and build your team and tech." Fundamental Research Labs nabs $30M to build AI agents across verticals TechCrunch https://t.co/Q2OI2uVFCC Fundamental Research Labs nabs $30M to build AI agents across verticals TechCrunch https://t.co/Q2OI2uVFCC" [X Link](https://x.com/theaustinlyons/status/1951302511333200316) 2025-08-01T15:22Z [----] followers, [---] engagements "Google Search is dead long live ChatGPT right Not so fast. Revenue: UP 📈 Search: UP 🔍 Earnings: Strong 💪 GenAI isnt killing Google; its boosting Google Heres why. 👇 $GOOG" [X Link](https://x.com/theaustinlyons/status/1952749248140706173) 2025-08-05T15:11Z [----] followers, [----] engagements "https://www.chipstrat.com/p/why-genai-is-good-for-google https://www.chipstrat.com/p/why-genai-is-good-for-google" [X Link](https://x.com/theaustinlyons/status/1952749250736984131) 2025-08-05T15:11Z [----] followers, [---] engagements "gpt-oss-120b for production general purpose high reasoning use cases that fits into a single H100 GPU (117B parameters with 5.1B active parameters) gpt-oss-20b for lower latency and local or specialized use cases (21B parameters with 3.6B active parameters) Our open models are here. Both of them. https://t.co/9tFxefOXcg Our open models are here. Both of them. https://t.co/9tFxefOXcg" [X Link](https://x.com/theaustinlyons/status/1952777218213724244) 2025-08-05T17:02Z [----] followers, [----] engagements "AMD announces its 33rd consecutive quarter of YoY CPU server share gains 🔥 Thats over [--] years of uninterrupted progress an incredible streak and a huge testament to the teams making it happen. $AMD @AMD" [X Link](https://x.com/theaustinlyons/status/1952838869839216884) 2025-08-05T21:07Z [----] followers, [---] engagements "🚨 AMD MI350 Series entered volume production ahead of schedule in June. 📈 Steep ramp in 2H [----] to power multiple large-scale deployments including Oracles 27000+ node AI cluster with MI355X accelerators. $AMD @AMD" [X Link](https://x.com/theaustinlyons/status/1952840060824694969) 2025-08-05T21:12Z [----] followers, [----] engagements "Its the inference era where atscale inference workloads are the name of the game. $AMD says MI355 delivers up to 40% more tokens per dollar for said workloads. Investors will be watching for this to show up as higher Instinct sales and market share gains. Hope to hear more in the commentary. @AMD" [X Link](https://x.com/theaustinlyons/status/1952840665228034312) 2025-08-05T21:14Z [----] followers, [----] engagements "The MI350 ramp is the big Instinct story for Q3 driven by deployments at multiple customers. So far it doesn't sound like we'll see much of that quite yet in Q2. If execution stays on track Q3 could be the start of a small share uptick $AMD @AMD" [X Link](https://x.com/theaustinlyons/status/1952842206458618049) 2025-08-05T21:20Z [----] followers, [----] engagements "AMDs AI roadmap is aimed at gigawattscale deployments the kind needed to reach tens of billions in annual Instinct revenue. Strong MI350 traction + positive MI400 feedback + relationships with right customers could make that ramp a reality. $AMD @AMD" [X Link](https://x.com/theaustinlyons/status/1952851748005269702) 2025-08-05T21:58Z [----] followers, [---] engagements "Inference is becoming more complex and tokenintensive as reasoning models and distributed inference make rackscale architectures important for inference as well. $AMD aims to be a fullscale solution provider not just a chip vendor for large deployments and says customers are validating that approach. @AMD" [X Link](https://x.com/theaustinlyons/status/1952852917767242183) 2025-08-05T22:03Z [----] followers, [---] engagements "As I said on @chipstrat back in January CPU TAM is increasing thanks to AI. $AMD is seeing the same" [X Link](https://x.com/theaustinlyons/status/1952853698344042625) 2025-08-05T22:06Z [----] followers, [----] engagements "AWS is losing share just as GenAI inference demand explodes (reasoning & agents) $AMZN cant afford to fall behind because half of Amazons profit comes from AWS Time to stop coasting and start buying GPUs. If Nvidia is slow to allocate get some Instinct 355X too. Fast. Trainium2/3 is a long-term TCO play. Gotta shore up GPUs ASAP. From Q2 call: Jassy: we could be doing more revenue. we're working hard on changing that and how much capacity we have. the single biggest constraint is power. Get creative to overcome those constraints $AMZN needs to convince investors it has a plan here. $AMZN" [X Link](https://x.com/theaustinlyons/status/1953191720486879690) 2025-08-06T20:29Z [----] followers, [----] engagements "Read more here: https://www.chipstrat.com/p/aws-is-missing-the-ai-infra-moment https://www.chipstrat.com/p/aws-is-missing-the-ai-infra-moment" [X Link](https://x.com/theaustinlyons/status/1953191723326144861) 2025-08-06T20:29Z [----] followers, [---] engagements "SkyWater increased American capacity for foundational silicon significantly. If you haven't heard of $SKYT I wrote a short intro below. https://t.co/yDolBAutll https://t.co/yDolBAutll" [X Link](https://x.com/theaustinlyons/status/1953215988188758186) 2025-08-06T22:05Z [----] followers, [----] engagements "Watching closely. $AMZN relies heavily on AWS: 18% of revenue but 53% of operating profit. Yet AWS grew just 17% last quarter while Google Cloud grew 32% and Azure 39%. Amazon cant afford to coast during the GenAI wave Amazon needs to regain AI infra momentum fast" [X Link](https://x.com/theaustinlyons/status/1953480094493950122) 2025-08-07T15:35Z [----] followers, [--] engagements "@munster_gene Definitely agree that Google's search user experience can and will be improved. The good news for Google is that LLMs make everything about search better. And Google has something that OpenAI doesn't. a profitable business model" [X Link](https://x.com/theaustinlyons/status/1953838335614505164) 2025-08-08T15:18Z [----] followers, [----] engagements "@handleym99 1/ OAI will make money in two ways A) commercial B) consumer A: The commercial biz will make money from the API (supporting "wrapper" companies) plus selling subscriptions (e.g. $200/mo). But that won't be enough to support widespread free consumer inference" [X Link](https://x.com/theaustinlyons/status/1953878524537766009) 2025-08-08T17:58Z [----] followers, [--] engagements "Joined @IvanaSpear & @NPetallides on @SchwabNetwork to talk Mag [--]. Nvidia: High ROIC leading. Azure & GCP: Winning AI cloud; AWS short on GPUs & power. Google: GenAI boosts ads. Apple: On-device AI still forming. $NVDA $MSFT $GOOG $AAPL $META $AMZN $TSLA https://www.youtube.com/watchv=Q6Chi6JXAbQ https://www.youtube.com/watchv=Q6Chi6JXAbQ" [X Link](https://x.com/theaustinlyons/status/1955002250256060874) 2025-08-11T20:23Z [----] followers, [----] engagements "Demand steady pricing power doing the heavy lifting for $MU as it raises Q4 guide: Rev: $10.7B $11.2B GM: 42% 44.5% EPS: $2.50 $2.85 Sumit Sadana: This is primarily about pricing pricing has been strong across end markets" [X Link](https://x.com/theaustinlyons/status/1955271343986946114) 2025-08-12T14:13Z [----] followers, [---] engagements "$MU isnt the commodity memory maker of old. We have either very robust share or a trajectory of share gains in every single high-margin portion of our business. Mix shift + pricing + share gains are driving results leaving them really well positioned for FY26" [X Link](https://x.com/theaustinlyons/status/1955272721308188786) 2025-08-12T14:18Z [----] followers, [---] engagements "AI boom = double whammy for $MU. HBM demand is surging and with 3:1 HBM wafers to DDR5 fab capacity is being diverted to HBM. Thats tightening DDR5 supply even as demand stays solid driving higher pricing for both HBM and DDR5" [X Link](https://x.com/theaustinlyons/status/1955273898028925375) 2025-08-12T14:23Z [----] followers, [----] engagements "$MU says HBM4E will be even less of a commodity as it becomes part of the compute architecture. Customers plan to embed GPU logic into the HBM base die offloading functions and creating custom memory tailored to their specs" [X Link](https://x.com/theaustinlyons/status/1955275178734207406) 2025-08-12T14:28Z [----] followers, [----] engagements "Continuing the "not a commodity" HBM4e explanation $MU says embedding customer IP in the base die is so costly that most will stick to only one or two suppliers" [X Link](https://x.com/theaustinlyons/status/1955275924292735028) 2025-08-12T14:31Z [----] followers, [---] engagements "@IntTradeNetwork Fair point Maybe custom AI ASIC makers too like Broadcom" [X Link](https://x.com/theaustinlyons/status/1955278331819036878) 2025-08-12T14:40Z [----] followers, [--] engagements "Yes @awscloud is missing the AI infra moment. The cloud wars are heating up and the leader of the last decade is now playing catch-up. AWS was the first major cloud provider with a commanding lead for years. But in the GenAI era the game has shifted. Heres the twist: AWS is only 18% of Amazons revenue yet it drives over half the companys profit. That makes it not just important but existential Since the large language model boom AWS has been losing net new revenue share to Microsoft Azure and Google Cloud. Why - Microsoft bought GPUs early and at scale. - AWS hesitated and now it is supply" [X Link](https://x.com/theaustinlyons/status/1955340757046530288) 2025-08-12T18:48Z [----] followers, [---] engagements "At the @Oppenheimer Tech Conference @Qualcomm's Nakul Duggal says $QCOM will commercialize it's first automotive driving stack with @BMW with more details coming in September at IAA Mobility in Munich. Big milestone for Snapdragon Ride as Qualcomm moves deeper into ADAS" [X Link](https://x.com/theaustinlyons/status/1955360573576626460) 2025-08-12T20:07Z [----] followers, [----] engagements "CoreWeave: increased backlog ($30B) yet questions about how they fund that buildout. Good point from @anurag_rana4 that the biggest CSPs have very profitable primary business to fund the land grab buildout right now. $CRWV $NBIS $MSFT $GOOG $AMZN $ORCL CoreWeave reported adjusted earnings per share loss of [--] cents for the June quarter compared to Wall Street's consensus estimate for a [--] cent loss. @anurag_rana4 breaks down the results on "Bloomberg The Close" https://t.co/FtdaKiMS7Q https://t.co/q5Ou4GB6fk CoreWeave reported adjusted earnings per share loss of [--] cents for the June quarter" [X Link](https://x.com/theaustinlyons/status/1955375122510160070) 2025-08-12T21:05Z [----] followers, [----] engagements "With AI shifting from training to inference especially reasoning its good to see $CRWV positioning as the best purpose-built AI cloud for both. Customer diversification signals confidence in CoreWeaves inference capabilities" [X Link](https://x.com/theaustinlyons/status/1955378385833644389) 2025-08-12T21:18Z [----] followers, [---] engagements "$CRWV seeing a "massive increase" in inference workloads as validated by watching the power meter lol" [X Link](https://x.com/theaustinlyons/status/1955387760845656340) 2025-08-12T21:55Z [----] followers, [----] engagements "WOW update your financial models $CRWV implies a 7-year useful life for GPUs. H100s & A100s still earn $ after training contracts end repurposed for inference. To be fair Jensen has been saying this all along :) $CRWV $NVDA @CoreWeave @nvidia" [X Link](https://x.com/theaustinlyons/status/1955389989245898807) 2025-08-12T22:04Z [----] followers, 26.7K engagements "$QCOM's ADAS play has three flavors - Chip + QCOM software stack (BMW global launch) - Chip + 3rd-party stack (common in China) - Chip + OEMs own stack BMW is the first global rollout of QCOMs co-developed stack opening the door to a verified non-China turn-key merchant stack for other OEMs. From the JP Morgan Hardware & Semis Forum. @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1955730253349085216) 2025-08-13T20:36Z [----] followers, [---] engagements "$QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and centralized compute. The same stack maps directly to robotics drones and humanoids and thanks to the deal with BMW QCOM can resell it globally. @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1955731417503343093) 2025-08-13T20:41Z [----] followers, [---] engagements "$QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip redundancy in mid-tier and swap in higher ADAS silicon for premium. @Qualcomm" [X Link](https://x.com/theaustinlyons/status/1955732341755928936) 2025-08-13T20:45Z [----] followers, [----] engagements "@BMW @Qualcomm https://x.com/theaustinlyons/status/1955732341755928936 $QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip redundancy in mid-tier and swap in higher ADAS silicon for premium. @Qualcomm https://t.co/MspJ9c5U8E https://x.com/theaustinlyons/status/1955732341755928936 $QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip" [X Link](https://x.com/theaustinlyons/status/1955742942448038028) 2025-08-13T21:27Z [----] followers, [---] engagements "@BMW @Qualcomm https://x.com/theaustinlyons/status/1955731417503343093 $QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and centralized compute. The same stack maps directly to robotics drones and humanoids and thanks to https://t.co/h0LEk08K07 https://x.com/theaustinlyons/status/1955731417503343093 $QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and" [X Link](https://x.com/theaustinlyons/status/1955743092398543227) 2025-08-13T21:27Z [----] followers, [---] engagements "Fun 10-min run with @AlexCoffeyCS on mega tech + semis $AAPL Why partnering for models makes sense $TSLA EV business strain = funding risk for future bets (robotaxis robotics) $MSFT Infra + apps both winning from AI $INTC How it fell behind whats next $AMD How its catching up whats holding it back $MU Why memory matters and Microns position in HBM @SchwabNetwork @Apple @Tesla @Microsoft @intel @AMD @MicronTech https://www.youtube.com/watchv=Qj3XCKVmfUQ https://www.youtube.com/watchv=Qj3XCKVmfUQ" [X Link](https://x.com/theaustinlyons/status/1956069638456140103) 2025-08-14T19:05Z [----] followers, [----] engagements "Watch here https://www.youtube.com/watchv=Qj3XCKVmfUQ https://www.youtube.com/watchv=Qj3XCKVmfUQ" [X Link](https://x.com/theaustinlyons/status/1956069641383797195) 2025-08-14T19:05Z [----] followers, [---] engagements "$AMAT down 10% after hours on softer Q4 guide. China digestion + fab timing weighing near term. Record Q3 rev/EPS beat prior guide with strength in DRAM etch metal dep & OLED display margins. Long-term AI chipmaking build-out HBM/packaging share gains still on track" [X Link](https://x.com/theaustinlyons/status/1956087623656800556) 2025-08-14T20:16Z [----] followers, [----] engagements "Intel can stand on its own the White House just wants to bridge the gap and help find anchor tenants get to full fabs profitability and back to business. I had the pleasure of being on @BBCWorld with @stevelai last night to talk about reports of the White House seeking an equity stake in @intel" [X Link](https://x.com/theaustinlyons/status/1956325006369054803) 2025-08-15T12:00Z [----] followers, [--] engagements "Intel can stand on its own the White House just wants to bridge the gap and help find anchor tenants get to full fabs profitability and back to business. I had the pleasure of being on @BBCWorld with @stevelai last night to talk about reports of the White House seeking an equity stake in @intel $INTC" [X Link](https://x.com/theaustinlyons/status/1956328005703180481) 2025-08-15T12:11Z [----] followers, [----] engagements "Thinking about writing on $NVDA scale-up and scale-out networking but want your feedback first. What are yall seeing from NVIDIA that you like or dont like Pros cons questions hot takes" [X Link](https://x.com/theaustinlyons/status/1956845100983795960) 2025-08-16T22:26Z [----] followers, [----] engagements "Yes risk of monopsony. See $AMATs earnings call; beyond China they are softening the guide due to a leading edge foundry pulling back. @Mojo_flyin @rwang07 @OmerCheeema Ive said it before and Ill say it again; ASMLs risk is monopsony only one buyer of its leading machines. Someone else has to emerge opposite of TSMC. @Mojo_flyin @rwang07 @OmerCheeema Ive said it before and Ill say it again; ASMLs risk is monopsony only one buyer of its leading machines. Someone else has to emerge opposite of TSMC" [X Link](https://x.com/theaustinlyons/status/1957059262015742035) 2025-08-17T12:37Z [----] followers, [----] engagements "@DasTechnocrat @business The equity stake wouldnt be net new money so not enough to stand up Ohio plant. Also the key question is how to force fabless to use Intel Foundry ie get fabless companies to overcome switching costs. Hence our expectation of tarrifs" [X Link](https://x.com/theaustinlyons/status/1957501653344743562) 2025-08-18T17:55Z [----] followers, [---] engagements "@chilling_dc @DasTechnocrat @business @chilling_dc You're saying the administration could say to AMD or Nvidia for example "commit to making something at Intel Foundry or no China market access for you"" [X Link](https://x.com/theaustinlyons/status/1957559898557825342) 2025-08-18T21:47Z [----] followers, [--] engagements "@chilling_dc @DasTechnocrat @business Problem with threatening *all* chips is that's a huge tax on the US tech ecosystem. It would probably be better to tariff only leading-edge chips (to incentivize Intel 18A/14A) or something quite targeted. Not sure Xilinx's FPGAs or AMD's Xbox chips etc should be tariffed" [X Link](https://x.com/theaustinlyons/status/1957562683848270146) 2025-08-18T21:58Z [----] followers, [---] engagements "SoftBanks $2B bet on $INTC is fantastic. For Intel a vote of confidence and a bit of clean capital. Speculatively some potential volume from Arm or related silicon. Meanwhile Masa Son bought in ahead b4 the biggest sovereign backstop gets in. Still need customers and more capital but a good day. @intel SoftBank Group and Intel Corporation Sign $2B Investment Agreement https://t.co/jTn8kYzQWh SoftBank Group and Intel Corporation Sign $2B Investment Agreement https://t.co/jTn8kYzQWh" [X Link](https://x.com/theaustinlyons/status/1957632294992162864) 2025-08-19T02:34Z [----] followers, [----] engagements "$INTC's wild few weeks Jul [--] Earnings: slows Ohio 14A fab build-out Aug [--] Trump calls CEO Lip-Bu Tan highly conflicted and urges him to quit Aug [--] Tan meets Trump at WH; Trump reverses tone calls his story amazing Aug [--] Leak: WH exploring equity stake in Intel to back U.S. fabs Aug [--] AM Report: plan could convert CHIPS grants into 10% govt stake Aug [--] PM SoftBank invests $2B at $23/share Whew. What's next Details on WH equity stake" [X Link](https://x.com/theaustinlyons/status/1957801285475819541) 2025-08-19T13:46Z [----] followers, [----] engagements "See the background slide for $CDNS projection of $1.2T semiconductor TAM in [----] "The acceleration of silicon" - Anirudh Devgan President and CEO of Cadence Learn more: https://t.co/te3CZWD5l6 https://t.co/xeGgFK8Iow "The acceleration of silicon" - Anirudh Devgan President and CEO of Cadence Learn more: https://t.co/te3CZWD5l6 https://t.co/xeGgFK8Iow" [X Link](https://x.com/theaustinlyons/status/1957860492162031894) 2025-08-19T17:41Z [----] followers, [----] engagements "Teslas auto business is nearly worthless but the robotaxi business will be huge in [----] 🤔😅 We expect Tesla to win 35% market share versus competitors Waymo at 15% Uber at 38% and Lyft at 13% generating almost $250 billion in revenue in [----]. After discounting the robotaxi EBITDA of $145 billion at 8.5% discount rate we estimate an implied value of Teslas robotaxi business at $298.61 per share energy business at $30.73 per share and auto business at $28.09 per share totaling an implied fair value of $357.43 per share. William Blair went to Austin to compare $TSLA robotaxi and $GOOGL Waymo:" [X Link](https://x.com/theaustinlyons/status/1958115647147110510) 2025-08-20T10:35Z [----] followers, [---] engagements "Not saying restricted just pointing out everyone wants negotiation power and supply chain diversification but that requires actually doing business with Intel Foundry. And if US government is truly bought into IF for national security it makes sense that the govt would encourage fabless designers to give IF a chance e.g. with a small product line" [X Link](https://x.com/theaustinlyons/status/1958262346293318038) 2025-08-20T20:18Z [----] followers, [--] engagements "@rwang07 $9B in a single month for networking or is that for the quarter" [X Link](https://x.com/theaustinlyons/status/1958489716304630267) 2025-08-21T11:21Z [----] followers, [---] engagements "$NVDA BYOC (bring your own CPU). Integrate NVLink C2C IP into your silicon and get [---] GB/s CPU-to-accelerator bandwidth. Think of it as replacing Grace/Vera on a Bianca board with your own silicon. @nvidia @NVIDIANetworking $NVDA" [X Link](https://x.com/theaustinlyons/status/1958633627702763559) 2025-08-21T20:53Z [----] followers, [----] engagements "15 years ago Andy Grove wrote Losing semiconductor manufacturing means losing the industrial commonsthe knowhow suppliers and ecosystems that drive future innovation. Quite prescient as the U.S. works to rebuild and restore its semi manufacturing base. $INTC How to make an American job. https://t.co/8U5CsvzJB6 https://t.co/x8yfyBb54r How to make an American job. https://t.co/8U5CsvzJB6 https://t.co/x8yfyBb54r" [X Link](https://x.com/theaustinlyons/status/1958869414016950706) 2025-08-22T12:30Z [----] followers, [----] engagements "Managing jitter across distributed datacenters 🤯 You have scale-up scale-out and now scale-across. @NVIDIA moat widens with Spectrum-XGS Ethernet that can tie GPU clusters together across distributed locations. GPU super clusters can now be tied between hyperscalers neoclouds etc. Wild. https://t.co/415uq919Wu https://t.co/ltNqZ0Vbrw You have scale-up scale-out and now scale-across. @NVIDIA moat widens with Spectrum-XGS Ethernet that can tie GPU clusters together across distributed locations. GPU super clusters can now be tied between hyperscalers neoclouds etc. Wild. https://t.co/415uq919Wu" [X Link](https://x.com/theaustinlyons/status/1958913499377410466) 2025-08-22T15:25Z [----] followers, [---] engagements "Looking forward to $INTC updates on: 1) new investors (incl. PE) 2) anchor 14A customers 3) 14A build-out restart Reminder: 20A/18A in AZ was 49% ($15B) co-funded by Brookfield under Intels Smart Capital program. BIG NEWS: The United States of America now owns 10% of Intel one of our great American technology companies. This historic agreement strengthens U.S. leadership in semiconductors which will both grow our economy and help secure Americas technological edge. Thanks to Intel https://t.co/AYMuX14Rgi BIG NEWS: The United States of America now owns 10% of Intel one of our great American" [X Link](https://x.com/theaustinlyons/status/1958993628669280379) 2025-08-22T20:44Z [----] followers, [----] engagements "Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky) US GOVT AGREES TO 433.3MLN SHARES OF INTEL COMMON STOCK AT 20.47/SHARE NOW [-----] US GOVT AGREES TO 433.3MLN SHARES OF INTEL COMMON STOCK AT 20.47/SHARE NOW 24.94" [X Link](https://x.com/theaustinlyons/status/1958996545669628043) 2025-08-22T20:55Z [----] followers, 11.6K engagements "Also US Gov't got in cheaper than Masa and Softbank ($20.47 vs $23) Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky) Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky)" [X Link](https://x.com/theaustinlyons/status/1958997202594144261) 2025-08-22T20:58Z [----] followers, [---] engagements "@PatrickMoorhead @howardlutnick @LipBuTan1 Looks like DJT had a post on Truth Social https://x.com/RapidResponse47/status/1958999176324555140 https://t.co/suLz96EODS https://x.com/RapidResponse47/status/1958999176324555140 https://t.co/suLz96EODS" [X Link](https://x.com/theaustinlyons/status/1959007141379154381) 2025-08-22T21:37Z [----] followers, [----] engagements "Some analysis and thoughts on the U.S. Governments investment in Intel Corporation per the Friday announcement. $INTC @intel" [X Link](https://x.com/theaustinlyons/status/1959617726680309816) 2025-08-24T14:04Z [----] followers, [---] engagements "More analysis and thoughts on the U.S. Governments investment in $INTC" [X Link](https://x.com/theaustinlyons/status/1959667569461272794) 2025-08-24T17:22Z [----] followers, [----] engagements "MI450 is when the competition really starts 👀 $AMD hits hard on third gen of their roadmap Frank AMD Sr Fellow presented today at hotchips conference explaining how scale up BW is 10x greater than scale out BW. This is the reason why MI355's scale up world size of [--] is not enough for frontier diagg prefill+wideEP inferencing and frontier training compared to GB200's https://t.co/bqrXnzQ8Uj Frank AMD Sr Fellow presented today at hotchips conference explaining how scale up BW is 10x greater than scale out BW. This is the reason why MI355's scale up world size of [--] is not enough for frontier" [X Link](https://x.com/theaustinlyons/status/1959743087074251156) 2025-08-24T22:22Z [----] followers, [----] engagements "My thoughts on @dMatrix_AI and 3D-stacked digital in-memory compute (3DIMC): d-Matrix is taking direct aim at the memory wall with 3DIMC vertically stacking DRAM with logic dies (see image below). The company claims a 10x boost in bandwidth and 10x better energy efficiency versus HBM4. These gains could reshape inference economics in an era where datacenter power is capped and agentic AI demands low-latency interactivity and high throughput at scale. Theres precedent here. Nearly [--] years ago AMD bet on stacked DRAM and co-developed HBM. The early pain of costly interposers and low yields" [X Link](https://x.com/theaustinlyons/status/1959980073949270105) 2025-08-25T14:03Z [----] followers, [----] engagements "At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its real or I wouldnt be here."" [X Link](https://x.com/theaustinlyons/status/1960058405223014430) 2025-08-25T19:15Z [----] followers, 34.3K engagements "Don also mentioned Clearwater Forest's 17% IPC uplift from Sierra Forest $INTC At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its real or I wouldnt be here." https://t.co/PBiTyghJxG At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its" [X Link](https://x.com/theaustinlyons/status/1960059173019816135) 2025-08-25T19:18Z [----] followers, [----] engagements "Rapidus will hit 2nm in [----] a node or two behind $TSM & $INTC. Hence the pitch is time-to-silicon not node leadership. Who values that Curious to see. AI chip startups hyperscalers refining in-house silicon (i.e to catch up to TPU v7) But thats low-volume R&D value prop too. Can it scale to HVM CEO says 25K WPM by [----]. @hotchipsorg" [X Link](https://x.com/theaustinlyons/status/1960457857259753867) 2025-08-26T21:42Z [----] followers, [----] engagements "$NVDA Spectrum-X Ethernet already doing ARR of $10B Plus they still sell Infiniband :) It's all about AI networking these days. $NVDA CFO: "escalating demands of AI compute clusters necessitate high efficiency and low latency networking" It's all about AI networking these days. $NVDA CFO: "escalating demands of AI compute clusters necessitate high efficiency and low latency networking"" [X Link](https://x.com/theaustinlyons/status/1960813595140481260) 2025-08-27T21:16Z [----] followers, [----] engagements "$NVDA earnings shout out to @hotchipsorg and $CRWV. "At Hotchips we introduced Spectrum-XGS Ethernet a technology designed to unify disparate data centers into gigascale AI super factories. CoreWeave is an initial adopter of the solution which is projected to double GPU-to-GPU communication speed."" [X Link](https://x.com/theaustinlyons/status/1960814136297734644) 2025-08-27T21:18Z [----] followers, [----] engagements "My takeaways from $NVDA earnings: Networking = huge bright spot China = upside optionality Thor = underappreciated future play Nearly $50B in a single quarter 🤯 Air-cooled RTX Pro servers = nice incremental add End-to-end AI system = Nvidias unbeatable edge @nvidia @NVIDIANetworkng" [X Link](https://x.com/theaustinlyons/status/1960840314941530363) 2025-08-27T23:02Z [----] followers, 12.1K engagements Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing
@theaustinlyons Austin Lyons
Austin Lyons posts on X about $nvda, ai, $amd, $intc the most. They currently have [-----] followers and [---] posts still getting attention that total [------] engagements in the last [--] hours.
Engagements: [------] #
- [--] Week [------] +196%
- [--] Month [-------] -26%
- [--] Months [---------] +400%
- [--] Year [---------] +7,617%
Mentions: [--] #
- [--] Week [--] +200%
- [--] Month [--] -1.60%
- [--] Months [---] +150%
- [--] Year [---] +3,233%
Followers: [-----] #
- [--] Week [-----] +1.80%
- [--] Month [-----] +6.20%
- [--] Months [-----] +172%
- [--] Year [-----] +431%
CreatorRank: [-------] #
Social Influence
Social category influence stocks technology brands finance automotive brands countries fashion brands social networks cryptocurrencies celebrities events
Social topic influence $nvda, ai, $amd, $intc, $crdo, meta, open ai, $mrvl, $avgo, 6969
Top accounts mentioned or mentioned by @nvidia @intel @amd @qualcomm @vikramskr @microntech @schwabnetwork @semidoped @openai @meta @benbajarin @arm @synopsys @cadence @apple @nvidias @microsoft @patrickmoorhead @chipstrat @bmw
Top assets mentioned NVIDIA Corp. (NVDA) Advanced Micro Devices (AMD) Intel Corporation (INTC) Credo Technology Group Holding Ltd (CRDO) Marvell Technology Inc (MRVL) Broadcom, Inc. (AVGO) QUALCOMM, Inc. (QCOM) Alphabet Inc Class A (GOOGL) Amazon.com, Inc. (AMZN) Microsoft Corp. (MSFT) Micron Technology, Inc. (MU) Metadium (META) Arista Networks Inc (ANET) Rivian Automotive, Inc. (RIVN) Synopsys, Inc. (SNPS) Arm Holdings plc (ARM) Goldman Sachs (GS) CoreWeave, Inc (CRWV) Cisco Systems Inc (CSCO) Dell Technologies, Inc. (DELL) Apple, Inc. (AAPL) Oracle Corporation (ORCL) Taiwan Semiconductor (TSM) Astera Labs, Inc. (ALAB) Tesla, Inc. (TSLA) DeepSeek (DEEPSEEK) Cadence Design Systems Inc (CDNS)
Top Social Posts
Top posts by engagements in the last [--] hours
"Intel announced that it will hold an investor webinar at 1:30 p.m. PDT on Apr [--] [----] to present the vision and financial framework for the Intel Foundry business including the recast financials and new segment reporting structure"
X Link 2024-03-06T23:04Z [---] followers, [---] engagements
"Look dude I was in the room in [----] when you Jensen outlined the general purpose strategy to me and it made a ton of sense. Its brilliant And now youre like nah its not general purpose anymore @BenBajarin on Nvidias framing shift Episode 59: Discussing All Things AI Accelerators With Austin Lyons @benbajarin and @jaygoldberg and joined by @theaustinlyons to discuss all things AI accelerators and real world implementations. Apple Podcasts: https://t.co/LdoiJJ7MOw Spotify: https://t.co/IIqUBpCttY Episode 59: Discussing All Things AI Accelerators With Austin Lyons @benbajarin and @jaygoldberg"
X Link 2024-03-19T02:30Z [---] followers, [---] engagements
"Nvidia CUDAs moat is surmountable & how AMD can overcome their software woes and compete. Read more 👇🏻 $AMD $NVDA Can AMD Bridge Nvidias Software Moat Thoughts On Transient Moats Betting on Generative AI AMD's Path Forward https://t.co/3mrunB8SOZ Can AMD Bridge Nvidias Software Moat Thoughts On Transient Moats Betting on Generative AI AMD's Path Forward https://t.co/3mrunB8SOZ"
X Link 2024-04-02T02:35Z [---] followers, [---] engagements
"Intel Made Inside America. Its time for an @intel Foundry slogan. Appeal directly to consumers during the Super Bowl March Madness and The Masters. Im no graphic designer or brand expert but desperate times call for desperate measures. Make the PC refresh cycle happen"
X Link 2024-09-21T13:11Z [---] followers, [--] engagements
"@pmarca I asked NotebookLM to generate a podcast covering the Techno Optimism Manifesto. The result was good "He wants us to think about the cost of fear. If we're always focused on what could go wrong are we missing out on what could go right" https://notebooklm.google.com/notebook/5485a54e-29cc-4db2-8223-69c5ae2cb1a1/audio https://notebooklm.google.com/notebook/5485a54e-29cc-4db2-8223-69c5ae2cb1a1/audio"
X Link 2024-10-02T15:01Z [---] followers, [--] engagements
"@magicsilicon @BenBajarin Let go $INTC foundry"
X Link 2025-01-01T00:09Z [---] followers, [---] engagements
"Its not generosityits strategy. ASML is leaving money on the table on purpose fostering win-win relationships with customers while keeping regulatory pressure and future competition away. And yes someday there will be competition be it China or next-gen technologies. But competitors wont have decades of trust built up like ASML. Its the TSMC playbook or the antithesis to everyone hates working with Intel. ASML can easily make gross margins in the [--] - 65% range if they want to theyre just far too generous. https://t.co/g20RWVgMiO ASML can easily make gross margins in the [--] - 65% range if"
X Link 2025-01-05T19:26Z [---] followers, [----] engagements
"Consumer-scale Gemini inference workloads incoming Gemini-powered voice control is coming to Google TV targeting 270M MAUs across Google/Android TV OS. Great for Google's TPU utilizationif its actually useful. Litmus test: Restart that Netflix documentary we didnt finish last night. If Gemini integrates deeply it's game-changing. If it just opens Netflix Useful but less compelling. Still even shaving seconds off adds up with millions of users and could drive meaningful inference demand. From CES: Google is giving third party TVs running Google TV a Gemini and always-on upgrade starting at the"
X Link 2025-01-06T17:04Z [---] followers, [--] engagements
"An underappreciated benefit of on-premises AI is that inference is a fixed cost instead of the variable cost of using a cloud API (e.g. OpenAI's API). @Qualcomm also shows a nice form factor here reminding us that on-premises AI can still be small and sleek. AI moves to the #edge at #CES2025. Our new #AI on-premises solution brings lower costs and more privacy to businesses: https://t.co/eb2nkJyRwB https://t.co/W9ls80tX4x AI moves to the #edge at #CES2025. Our new #AI on-premises solution brings lower costs and more privacy to businesses: https://t.co/eb2nkJyRwB https://t.co/W9ls80tX4x"
X Link 2025-01-06T17:21Z [---] followers, [----] engagements
"Having a hard time keeping up with all the changes Ill listen to this pod on my run tomorrow morning. The final salvo (I'm told for real this time) of the Biden administration's AI/Semiconductor export controls has just been fired. Without hyperbole it is another very big very important move. Jordan and I unpack the new rule in this great convo. The final salvo (I'm told for real this time) of the Biden administration's AI/Semiconductor export controls has just been fired. Without hyperbole it is another very big very important move. Jordan and I unpack the new rule in this great convo"
X Link 2025-01-16T02:27Z [---] followers, [---] engagements
"Startups must invest time into building products; debugging things that "just work" on @nvidia is the antithesis of that. Hence startups are happy to pay a premium to ensure their product development velocity. They're paying for a "turn-key" platform. Yet some startups might have more time than money and thus be willing to take on some development velocity risk. At the same time Hopper rental prices are dropping dramatically and will continue as Blackwell becomes more available so cash-strapped startups might still opt for previous generation Nvidia as the alternative to current gen"
X Link 2025-01-18T18:06Z [---] followers, [---] engagements
"Yes. $QCOM well positioned for edge inference. Phone XR auto laptops and dabbling in on-premise edge servers Open-source distilled models that dont need crazy hardware (aka deepseek 7b) that are still capable has blown the doors wide open for edge inference apple and Qualcomm are major players that will stand to benefit due to their existing hardware advantage Open-source distilled models that dont need crazy hardware (aka deepseek 7b) that are still capable has blown the doors wide open for edge inference apple and Qualcomm are major players that will stand to benefit due to their existing"
X Link 2025-01-28T00:45Z [---] followers, [---] engagements
"Do algorithmic improvements like DeepSeeks mean that margins will shift from hardware vendors to labs Hard to see why. Algorithmic improvements are quickly copied from one lab to another making it hard for them to maintain technological differentiation. Hardware improvements take much longer to copy. [--]. Breakdown of DeepSeek V3 efficiency vs Llama 3: - Better: 11x fewer FLOPs per token thanks to MoE 37B vs 405B activated params - Better: 2x faster numerics fp8 vs bf16 training - Worse: 0.5x flops utilization 16% vs 33% end-to-end MFU* - Neutral: similar hardware [--]. Breakdown of DeepSeek V3"
X Link 2025-01-28T02:11Z [---] followers, [---] engagements
"Intel Products holding Foundry to high standards is excellent. And it must go both ways: Foundry needs Product to be a good customer (minimize hot lots etc) @intel $INTC Another strong point from @MJHolthaus that shows us how she views foundry partner with the lens on best foundry part for @Intel product. This is the right mindset to bring the best product to market regardless of foundry. Best process wins https://t.co/wUQWgf0pSt Another strong point from @MJHolthaus that shows us how she views foundry partner with the lens on best foundry part for @Intel product. This is the right mindset to"
X Link 2025-01-30T22:52Z [---] followers, [----] engagements
"I read the DeepSeek R1 (reasoning model) paper and explored the implications for the semiconductor industry: Inference-time scaling is AIs newest demand vector. More reasoning per query means more compute per query. DeepSeeks open approach opens the door to competition while its distilled models bring reasoning to the edge. Where does this all lead https://t.co/2si2hp5Uaf Inference-time scaling is AIs newest demand vector. More reasoning per query means more compute per query. DeepSeeks open approach opens the door to competition while its distilled models bring reasoning to the edge. Where"
X Link 2025-02-10T15:34Z [---] followers, [----] engagements
"Im looking forward to a future with reasoning agents at my desk and multimodal GenAI in my glasses 😎 #DeepSeek R1 caused ripples. Newly released #AI models plus smaller model sizes and edge inference will lead to innovation in commercial applications in devices powered by @Snapdragon: https://t.co/7qjIf0UG5r #DeepSeek R1 caused ripples. Newly released #AI models plus smaller model sizes and edge inference will lead to innovation in commercial applications in devices powered by @Snapdragon: https://t.co/7qjIf0UG5r"
X Link 2025-02-11T18:03Z [---] followers, [---] engagements
"Nvidias fast roadmap cadence gives it an edge over ASIC rivals says Jensen. Compute networking memory software all evolve rapidly. #Nvidia #Tech $NVDA https://www.chipstrat.com/p/nvidia-4q25-earnings https://www.chipstrat.com/p/nvidia-4q25-earnings"
X Link 2025-03-06T16:14Z [---] followers, [----] engagements
"Awesome first person video views from @Meta Ray Bans At Mobile World Congress I caught up with @ziad_asghar SVP and GM of XR at @Qualcomm for an insightful conversation about on-device AI and its transformative potential for wearable technologies like smart glasses. Partially recorded through our Ray-Ban @Meta glasses our https://t.co/DYvzE8RnGh At Mobile World Congress I caught up with @ziad_asghar SVP and GM of XR at @Qualcomm for an insightful conversation about on-device AI and its transformative potential for wearable technologies like smart glasses. Partially recorded through our"
X Link 2025-03-10T21:54Z [---] followers, [---] engagements
"Tencent dropped a hybrid Transformer+Mamba MoE model 🚀 Introducing Hunyuan-TurboS the first ultra-large Hybrid-Transformer-Mamba MoE model Traditional pure Transformer models struggle with long-text training and inference due to O(N) complexity and KV-Cache issues. Hunyuan-TurboS combines: ✅ Mamba's efficient long-sequence https://t.co/QoCXkS8gUQ 🚀 Introducing Hunyuan-TurboS the first ultra-large Hybrid-Transformer-Mamba MoE model Traditional pure Transformer models struggle with long-text training and inference due to O(N) complexity and KV-Cache issues. Hunyuan-TurboS combines: ✅ Mamba's"
X Link 2025-03-12T01:40Z [---] followers, [---] engagements
"Projecting $1T of annual data center CapEx in [----] $NVDA"
X Link 2025-03-18T17:30Z [---] followers, [----] engagements
"Jensen said this is his favorite slide $NVDA and talking cuLITHO "in another five years time every mask every single lithography will be processed on NVIDIA CUDA""
X Link 2025-03-18T17:34Z [---] followers, [----] engagements
"What a beautiful flex love this slide @nvidia $NVDA"
X Link 2025-03-18T17:56Z [---] followers, [----] engagements
"Professor Huang explaining scale-up innovations from air-cooled integrated NVLink to disintegrated and liquid-cooled @nvidia $NVDA"
X Link 2025-03-18T18:01Z [---] followers, [---] engagements
""Tiny 50W CPU" @Arm well done @nvidia $NVDA $ARM"
X Link 2025-03-18T18:37Z [---] followers, [----] engagements
"Introducing Nvidia Photonics @nvidia's first copackaged option silicon photonic system and the world's first [---] terabyte per second CPO $NVDA"
X Link 2025-03-18T18:45Z [---] followers, [----] engagements
"Introducing Nvidia Newton an open-source physics engine in development with Google DeepMind and Disney Research purpose-built for developing robots. @nvidia @GoogleDeepMind $NVDA"
X Link 2025-03-18T19:17Z [---] followers, [---] engagements
"As Jensen said AI datacenters are power-constrained. Every building blocks efficiency counts and memory is no exception. Microns HBM3E 8H cuts power by 30% vs. competitors. That adds up meaningfully when you're running 100K+ GPU clusters. @MicronTech $mu The rise of large language models has challenged us to reimagine how computational systems process information. #HBM3E has emerged as a breakthrough memory technology helping AI models achieve new levels of reasoning and intelligence. Discover how the higher bandwidth and https://t.co/mR1zFifzTe The rise of large language models has"
X Link 2025-03-21T18:03Z [---] followers, [----] engagements
"SK Hynix isnt keeping Samsung in checkits Micron theyre targeting. Lately theres been talk among engineers that Micron is said to be in top form and is pursuing them with alarming intensity. Report: These Days SK Hynix Is More Focused on Restraining Micron Than Samsung SK Hynix isnt keeping Samsung in checkits Micron theyre targeting. Lately theres been talk among engineers that Micron is said to be in top form and is pursuing them with alarming Report: These Days SK Hynix Is More Focused on Restraining Micron Than Samsung SK Hynix isnt keeping Samsung in checkits Micron theyre targeting."
X Link 2025-03-22T12:01Z [---] followers, [----] engagements
"I think the interesting questions here: why did Apple use TPUs for training and why switch to Nvidia now If the GB300s are also for training maybe the TPUs (or the surrounding software ecosystem) were slowing Apple down. If the GB300s are for inference maybe its simply a TCO play - cheaper at scale inference with GB300s and Dynamo"
X Link 2025-03-25T13:34Z [---] followers, [---] engagements
"2nm AMD 🔬 + x86 made in the USA Next-generation @AMD EPYC processor codenamed Venice is the first HPC product in the industry to be taped out and brought up on the TSMC advanced 2nm (N2) process technology. AMD also announced the successful bring up and validation of its 5th Gen AMD EPYC CPU products at TSMCs new fabrication facility in Arizona underscoring its commitment to U.S. manufacturing"
X Link 2025-04-14T22:58Z [---] followers, [----] engagements
"Micron Announces Business Unit Reorganization to Capitalize on AI Growth Across All Market Segments @MicronTech $MU https://investors.micron.com/news-releases/news-release-details/micron-announces-business-unit-reorganization-capitalize-ai https://investors.micron.com/news-releases/news-release-details/micron-announces-business-unit-reorganization-capitalize-ai"
X Link 2025-04-17T13:10Z [---] followers, [----] engagements
"I visited @Qualcomm HQ for the first time and enjoyed the presentations demos and ride-alongs. One idea that got my wheels spinning (pun intended) is the use of GenAI entertainment for kids during road trips. My kids love the surprise and creativity GenAI enables. I can see them drawing on a car screen and watching it come to life via Unreal Engine connecting to Suno and generating music on demand or even creating a storybook with illustrations to be read aloud. Zonal audio allows each child to hear their own personalized experience as well. The possibilities are endless Parental controls are"
X Link 2025-04-17T14:11Z [---] followers, [----] engagements
"Synopsys CEO Sassine Ghazi on stage with LBT to discuss @Synopsys enabling @intel 18A 14A"
X Link 2025-04-29T16:15Z [---] followers, [---] engagements
"@Synopsys @intel Cadence CEO Anirudh Devan says Intel Foundry is a national asset and has great transistor and packaging technology. Encourages LBT to work as one team with the ecosystem @Cadence"
X Link 2025-04-29T16:23Z [---] followers, [---] engagements
"@Synopsys @intel @Cadence Siemens EDA CEO out to discuss 18A 18AP 14AE plus collaboration with Advanced Packaging @Siemens"
X Link 2025-04-29T16:26Z [---] followers, [---] engagements
"@Synopsys @intel @Cadence @Siemens Naga up looking forward to hearing his ideas regarding winning customers for Intel Foundry"
X Link 2025-04-29T16:44Z [---] followers, [---] engagements
"Jensen doing a good job of articulating @nvidia's move from from chip designer to AI infrastructure (system) designer #COMPUTEX"
X Link 2025-05-19T03:12Z [---] followers, [----] engagements
"Elon making a virtual appearance at MS Build to launch @grok on Azure"
X Link 2025-05-19T16:40Z [---] followers, [----] engagements
"500M WAUs for ChatGPT run entirely on Azure @Microsoft @OpenAI"
X Link 2025-05-20T17:54Z [---] followers, [----] engagements
"Will @micron appear in foldable @Apple iPhones "Industrys thinnest LPDDR5X package paving the way for exciting new smartphone designs." $MU $AAPL Micron is thrilled to announce the world's first [--] (1-gamma) node 16GB #LPDDR5X setting a new standard for flagship smartphones. What makesMicron [--] LPDDR5Xa game-changer Smaller: the industrys thinnest package at just 0.61mm perfect for ultra-thin or foldable https://t.co/eEeLuqSycz Micron is thrilled to announce the world's first [--] (1-gamma) node 16GB #LPDDR5X setting a new standard for flagship smartphones. What makesMicron [--] LPDDR5Xa game-changer"
X Link 2025-06-03T14:57Z [---] followers, [----] engagements
"Love the look of the R3X. Every other carmaker copies Tesla designs. This is fresh. Nice work @Rivian R3X at Unreal Fest in Orlando. Excited for more Unreal Engine updates across all Rivian vehicles. https://t.co/TwVySI3Rk3 R3X at Unreal Fest in Orlando. Excited for more Unreal Engine updates across all Rivian vehicles. https://t.co/TwVySI3Rk3"
X Link 2025-06-03T16:02Z [---] followers, [---] engagements
"The AI Factory is here. @nvidia defines the AI factory vision. @Dell turns it into a product. Hardware services financing included. Chatting with @jaygoldberg on @T_h_e_Circuit 🎙👇"
X Link 2025-06-05T17:25Z [---] followers, [----] engagements
"The importance of high-performance memory has never been greater fueled by its crucial role in supporting the growing demands of AI training and inference workloads in data centers. @MicronTech $MU today announced the shipment of HBM4 36GB 12-high samples to multiple key customers. Micron is now sampling our HBM4 36GB 12-high memory to key customers engineered to meet the growing demands of AI training and inference in #datacenter and #cloud environments. Designed for seamless integration with next-gen AI platforms this milestone offers customers a https://t.co/w4lFn3G7yX Micron is now"
X Link 2025-06-10T13:43Z [----] followers, [----] engagements
"Snapdragon for robotics on the horizon With continued auto downturn ADAS companies repositioning to address Robotics while power semis going after DC. Qualcomm sees itself as one of the top [--] players in robotics as many of its technologies like wireless camera AI low power etc. relevant to robotics. (comments at https://t.co/EtDRhOvgQo With continued auto downturn ADAS companies repositioning to address Robotics while power semis going after DC. Qualcomm sees itself as one of the top [--] players in robotics as many of its technologies like wireless camera AI low power etc. relevant to robotics."
X Link 2025-06-11T11:51Z [----] followers, [---] engagements
"Wow @LisaSu and @sama at @AMD Advancing AI"
X Link 2025-06-12T18:31Z [----] followers, [----] engagements
"Apple doesnt need to win the AI race. It needs to win the developer race. The best AI-native experiences wont come from Cupertino. Theyll come from third-party devs building for real workflows unique hobbies and vertical niches Apple will never touch. @Apple's job Make it simple and worthwhile to build those apps. @WWDC25 was a good start: on-device models clean APIs Swift-native frameworks. But the mission isnt done. Dev loyalty is the ultimate moat. Want AI-native personal computing to thrive Give developers great tooling Cut App Store fees Offer GenAI cloud credits Feature indie builders"
X Link 2025-06-17T14:56Z [----] followers, [----] engagements
"Hands-free AI. Oakley x Meta brings AI assistance voice commands and 3K video and looks cool. https://www.theverge.com/news/690133/meta-oakley-hstn-ai-glasses-price-date https://www.theverge.com/news/690133/meta-oakley-hstn-ai-glasses-price-date"
X Link 2025-06-20T13:35Z [----] followers, [---] engagements
"We are increasingly seeing performance use cases with the Ray-Bans like people wearing them on roller coasters cycling and being around water so were trying to lean into that says Alex Himmel the companys vice president in charge of wearables in an interview. The Meta Ray-Bans were supposed to be Metas last non-AR/display glasses. Then they got popular. Now Meta is going upmarket and rolling out Oakley smart glasses. Details here https://t.co/yOdLnQ1gq3 The Meta Ray-Bans were supposed to be Metas last non-AR/display glasses. Then they got popular. Now Meta is going upmarket and rolling out"
X Link 2025-06-20T13:54Z [----] followers, [---] engagements
"🔥 $AMD % 💰 @sama said When you told me the specs for MI450 I thought no way. MI350 delivers: massive memory native FP6 rack density OCP-compliant datacenter-ready MI450 brings: rack-scale coherence (2026) @xai @Meta @OpenAI @cohere are already running workloads and contributing to the Instinct ecosystem🚀 ROCm has real momentum 🛠 Read more 📈 $AMD: FP6 acceleration rack density HBM capacity AI NIC MI350 MI450 roadmap UALink ROCm [--] traction customer wins (@Meta @OpenAI @xai @OracleCloud @cohere) full-stack systems TCO advantage. Read the full @AMD breakdown 👇 https://t.co/V7hzwy46Wm "
X Link 2025-06-24T21:15Z [----] followers, [----] engagements
"🚨 Record quarter for $MU Revenue: $9.3B (+15% QoQ) HBM up 50% QoQ 🔥 Data center revenue doubled YoY Memory matters in this AI era Has it ever mattered more Today Micron reported our Q3 FY25 financial results. Learn more: https://t.co/1fmBBXLw0y https://t.co/iWlla4F36m Today Micron reported our Q3 FY25 financial results. Learn more: https://t.co/1fmBBXLw0y https://t.co/iWlla4F36m"
X Link 2025-06-25T20:12Z [----] followers, [----] engagements
"I spoke with @Heineken at @Microsoft Build. They are very bullish on agentic AI for intelligent automation. They see it as a revenue driver not just cost savings. They've thought deeply about how to adapt company culture to be AI-forward. Also they gave me a beer for asking good questions 🍻😁 From retail and government services to healthcare AI innovation is accelerating across every industry. Incredible to see the breadth of progress across the Netherlands during my recent visit. https://t.co/smSvyIUuXk From retail and government services to healthcare AI innovation is accelerating across"
X Link 2025-06-26T18:11Z [----] followers, [----] engagements
"Enjoyed talking @AMD with @DianeKingHall on @SchwabNetwork this morning Video link in comments. Main points: AMD sees the AI chip market topping $500B by [----] with most growth in inference (running models). AMD is offering a credible alternative to Nvidia focused on value and performance. Shift from training to inference plays to AMDs strengths; big customers already include OpenAI Meta xAI. AMD's technical bets include large memory capacity to support larger models and reduce TCO. Historically strong in hardware AMD was weak in software but theyre catching up via acquisitions new talent and"
X Link 2025-06-30T17:06Z [----] followers, [----] engagements
"Between Modular's world class engineering & SFC's ruthless price optimizations we're committed to creating the highest performance best priced inference in the world. We run a bespoke process with each customer to optimize for your specific requirements and only roll out when the price & performance are hitting your expectations. We've partnered with Modular to create Large Scale Inference (LSI) a new OpenAI-compatible inference service. It's up to 85% cheaper than other offerings & can handle trillion-token scale. We originally created it at the request of a major AI lab to do large scale"
X Link 2025-07-03T02:34Z [----] followers, [---] engagements
"@PatrickMoorhead @nvidia @Cadence @Synopsys Agree that the lifting of EDA restrictions increases the chances of an Nvidia Blackwell chip (B30) shipping to China. Good take"
X Link 2025-07-03T16:22Z [----] followers, [---] engagements
"🧵Nvidia's next frontier is AI in the physical world. Why Visual-Language-Action (VLA) models enable generalist robots that can see understand and act. $NVDA @NVIDIARobotics @nvidia https://t.co/pmvY6E1VJJ https://t.co/pmvY6E1VJJ"
X Link 2025-07-03T16:42Z [----] followers, [----] engagements
"VLAs unlock true real-world flexibility. Forget hardcoded factory bots. Imagine robots that learn new tasks on the fly adapt to chaotic spaces and follow natural language commands. Nvidias full-stack approach makes it possible"
X Link 2025-07-03T16:42Z [----] followers, [--] engagements
"Nvidias three computers strategy: ⚡ Train (DGX) ⚡ Simulate (Omniverse and Cosmos) ⚡ Deploy (Thor) A continuous loop that keeps robots learning and improving. Nvidia controls this loop completely. No one else is close"
X Link 2025-07-03T16:42Z [----] followers, [--] engagements
"This is not just about GPUs. Robots will need on-device transformers lots of memory and advanced sensors in addition to strong local compute. Each robot is like a mobile mini data center. A huge new semiconductor market is emerging and Nvidia is in the lead from an end-to-end systems perspective"
X Link 2025-07-03T16:42Z [----] followers, [---] engagements
"There isn't a customer in the world that would not leap at a viable alternative to $NVDA. I think people underestimate this $AMD might be starting to catch up to $NVDA I HIGHLY recommend this interview with a current high-ranking employee at $DELL talking about $AMD vs $NVDA and the client demand: [--]. According to him $AMD's main problem is ROCm which is immature compared to $NVDA's CUDA and the https://t.co/vcWpuEchqc $AMD might be starting to catch up to $NVDA I HIGHLY recommend this interview with a current high-ranking employee at $DELL talking about $AMD vs $NVDA and the client demand: 1."
X Link 2025-07-03T23:08Z [----] followers, [----] engagements
"+1 for an EV minivan from @Rivian Help This is what it looks like when you try to put [--] kids in an EV. And I have [--] @elonmusk @Rivian let's solve this problem We will be the first in line for an EV passenger van and have lots of future drivers who will need cars 🤩 https://t.co/RLSzYe7oas Help This is what it looks like when you try to put [--] kids in an EV. And I have [--] @elonmusk @Rivian let's solve this problem We will be the first in line for an EV passenger van and have lots of future drivers who will need cars 🤩 https://t.co/RLSzYe7oas"
X Link 2025-07-03T23:58Z [----] followers, [---] engagements
"2026 looks strong for HBM demand. Reasoning-era inference with ever larger context windows for agentic AI will push capacity and bandwidth needs to new highs. Nvidia Rubin & AMD MI450 will drive a massive HBM4 ramp likely joined by TPU v8 Trainium [--] and Intel Jaguar Shores. Chinese hyperscalers (ByteDance Alibaba Baidu) likely scaling up too. Micron holds 5% HBM share today targeting 20%+ in the next few years. HBM is margin accretive and a key growth driver. $MU @MicronTech Micron said last week that it is now shipping HBM to four customers in high volume spanning GPUs and ASIC platforms"
X Link 2025-07-06T18:15Z [----] followers, [---] engagements
"Forget cars. New consumer products for @Apple should be robotics and AR. Robotics is 5-10yr horizon and offers far more growth potential than EVs while AR is 1-3yr horizon still nascent and now supercharged by GenAI. Chinese smartphone maker Xiaomi has pulled off something Apple failed to do despite a decade-long attempt and thats to make a car. @Lindadalew looks at what helped Xiaomi overtake Apples EV ambitions https://t.co/486n2qPKrG https://t.co/th2cEuQrZ5 Chinese smartphone maker Xiaomi has pulled off something Apple failed to do despite a decade-long attempt and thats to make a car."
X Link 2025-07-07T01:58Z [----] followers, [---] engagements
"How long will @xai exist An LLM that can reference X posts and has a different personality is useful but is it a sustainable business xAI raises $10B in debt and equity TechCrunch https://t.co/0QIVsN8uxD xAI raises $10B in debt and equity TechCrunch https://t.co/0QIVsN8uxD"
X Link 2025-07-07T11:28Z [----] followers, [---] engagements
"@asu_PH 😂 true that. Started as bitcoin miner ended up as $9B acquisition. Well played"
X Link 2025-07-07T16:27Z [----] followers, [--] engagements
"Metas debt plan is public and straightforward. Stargate funding is still opaque for me. Eg Oracles $30B/yr OpenAI compute deal who pays (SoftBank) and what happens if OAI demand or SoftBank funding falls short Meta is reportedly looking to raise $29 billion from private capital to help build AI data centers in the US with this consisting of $3 billion in equity and $26 billion in debt. $META Meta is reportedly looking to raise $29 billion from private capital to help build AI data centers in the US with this consisting of $3 billion in equity and $26 billion in debt. $META"
X Link 2025-07-08T16:38Z [----] followers, [---] engagements
"Worlds best devices company is considering putting hardware engineer in charge. Makes sense. Hope he can articulate and stand firm on the AI strategy for said devices inspire talent and push into robotics and glasses. As I reported a year ago hardware engineering chief John Ternus is primed to be the next CEO of Apple when Tim Cook eventually retires. https://t.co/tu7oJnDiwv As I reported a year ago hardware engineering chief John Ternus is primed to be the next CEO of Apple when Tim Cook eventually retires. https://t.co/tu7oJnDiwv"
X Link 2025-07-08T22:05Z [----] followers, [---] engagements
"Bullish on glasses + GenAI After wearing Ray-Ban Meta Wayfarer glasses for a few weeks . I feel kind of naked wearing regular sunglasses. I've found three use-cases that are hard to roll back: * Spontaneous photos of my kids when we're out and about. Any cool pose that has a half-life of [--] seconds I After wearing Ray-Ban Meta Wayfarer glasses for a few weeks . I feel kind of naked wearing regular sunglasses. I've found three use-cases that are hard to roll back: * Spontaneous photos of my kids when we're out and about. Any cool pose that has a half-life of [--] seconds I"
X Link 2025-07-09T11:47Z [----] followers, [---] engagements
"Announced today same day @perplexity_ai shipped their browser. AI-native browser is ground zero for agentic AI will also aim to replace Google as default search Reuters is reporting that OpenAI is about to release an AI-powered web browser to directly compete with Chrome that will 'fundamentally change how consumers browse the web'. https://t.co/ikWyfpt6HZ Reuters is reporting that OpenAI is about to release an AI-powered web browser to directly compete with Chrome that will 'fundamentally change how consumers browse the web'. https://t.co/ikWyfpt6HZ"
X Link 2025-07-09T18:10Z [----] followers, [---] engagements
"@tinygrad @AnushElangovan Genuine question: How important is minimizing idle power when designing a datacenter AI accelerator Wouldn't customers want to limit idle time e.g. a race car designer would care more about active efficiency rather than idle fuel burn"
X Link 2025-07-09T18:25Z [----] followers, [---] engagements
"Unilever uses GenAI to generate marketing assets. Love this story. Enterprises have real use cases for that infrastructure buildout. And it requires storage too. 📈 In my latest for @WSJ I investigate @Unilever's plans to turbocharge its influencer marketing program and make boring toiletries like soap go viral on social media (with a little help from AI). Read here https://t.co/z4LKAE2Rrj In my latest for @WSJ I investigate @Unilever's plans to turbocharge its influencer marketing program and make boring toiletries like soap go viral on social media (with a little help from AI). Read here"
X Link 2025-07-10T04:25Z [----] followers, [---] engagements
"Makes sense. RealSense cameras are widely used robotics is taking off and $INTC is focusing on core biz. Best bet is to spin out and go for it like a startup would. The company realized it may have a better chance keeping up with demand and scaling itself if it spun out of Intel and raised its own capital Orbach said. The company realized it may have a better chance keeping up with demand and scaling itself if it spun out of Intel and raised its own capital Orbach said"
X Link 2025-07-11T12:08Z [----] followers, [---] engagements
"18A is the culmination of Intels [--] nodes in [--] years sprint. A rumored write-off doesnt erase the mission: get back in the game prove RibbonFET + PowerVia in real silicon and set the stage for 14A to truly compete with TSMC. We unpack the strategy risks and next moves. 👇 https://t.co/qdxD4Qx5eZ https://t.co/qdxD4Qx5eZ"
X Link 2025-07-11T13:24Z [----] followers, [----] engagements
"@Beth_Kindig Makes sense. RealSense cameras are widely used robotics is taking off and Intel is tightening the belt to focus on core businesses. It's a good move to spin this out so RealSense can go for it like a startup would"
X Link 2025-07-11T13:29Z [----] followers, [---] engagements
"🚕💥 Goldman Sachs forecasts North Americas AV rideshare market to grow at a 90% CAGR from [--------]. Gross margins for vertically integrated robotaxi operators could reach 4050% driving $3.5B in US AV gross profit by 2030"
X Link 2025-07-11T19:07Z [----] followers, [----] engagements
"We cant let one company control the entire AI infrastructure stack. Ive been hearing the same. Not bearish for Nvidia as there is plenty to go around. But supplier diversity + inference era is good for AMD"
X Link 2025-07-14T01:55Z [----] followers, [----] engagements
"Enjoyed talking @nvidia on @SchwabNetwork with @NPetallides 🏁 $5T race: Nvidia vs Microsoft vs Apple who gets there first 🤖 Robotaxis & humanoid robots push Nvidia beyond datacenters. ✅ Nvidia demand outstrips supply; Nvidia can sell every chip it makes. We also talked @Meta 🕶 Metas GenAI-driven advertising & AR glasses push a new frontier. Video in comments"
X Link 2025-07-14T20:18Z [----] followers, [----] engagements
"Rapid-fire Q&A on these TV spots is hard and Ive got a long way to go. But Ive been using ChatGPTs voice mode to practice and its very helpful. I feed it my talking points and tell it to act as the host firing questions at me. Its the right UI (voice) and infinitely patient. Surprisingly useful Thanks @OpenAI and @ChatGPTapp. Enjoyed talking @nvidia on @SchwabNetwork with @NPetallides 🏁 $5T race: Nvidia vs Microsoft vs Apple who gets there first 🤖 Robotaxis & humanoid robots push Nvidia beyond datacenters. ✅ Nvidia demand outstrips supply; Nvidia can sell every chip it makes. We also talked"
X Link 2025-07-14T20:38Z [----] followers, [---] engagements
"@Nvidia's next waves are autonomy and humanoid robotics. Goldman Sachs sees the North American robotaxi market growing at 90% CAGR to [----]. Morgan Stanley estimates a $5 trillion market for humanoid robots by [----]. Both need massive compute for training simulation and onboard intelligence and Nvidia uniquely covers all three. We see a multi-year possibly decade-long AI infrastructure buildout ahead. Still constrained by supply not demand. As new cycles like robotaxis and robots kick in Nvidia is positioned to benefit across each layer of the stack from edge to cloud. Francisco Bido says"
X Link 2025-07-14T21:24Z [----] followers, [--] engagements
"Is MI308 back on Will there be an @AMD Instinct roadmap for China like H20/B30 $AMD"
X Link 2025-07-15T03:06Z [----] followers, [---] engagements
"@PatrickMoorhead Also $ALAB 🫡"
X Link 2025-07-15T13:53Z [----] followers, [---] engagements
"Big for @nvidia. Nuro is a customer so is Lucid so is Uber. We continue to believe robotaxis / autonomy are huge growth opportunities for Nvidia; DGX for training RTX Pro and Cosmos for simulation Drive Thor for onboard compute. Competition in this market is good MORE: Lucid is also partnering with Nuro and Uber on ROBOTAXIS. The plan is to deploy at least [-----] robotaxis over six years. Uber will purchase and operate Lucid Gravity SUVs outfitted with Nuro Driver technology on its ride-sharing network and is making separate MORE: Lucid is also partnering with Nuro and Uber on ROBOTAXIS. The"
X Link 2025-07-17T14:16Z [----] followers, [----] engagements
"Or @NVIDIARobotics could just buy an analog company and build out a suite of compute + sensors for robotics. Let analog companies be commoditized. Build the high value solution offering: AI compute + analog sensor + software (CUDA)"
X Link 2025-07-17T19:43Z [----] followers, [---] engagements
"$AMAT going Zuck-mode but in Asia: For talent aggressive recruitment is underway with no limitations on scale. Multiple industry insiders stated in unison "Applied Materials is contacting the entire industry to secure high-level talent in advanced packaging. They are gathering so many people that it's as if they are 'sucking up' packaging personnel." In particular it has been identified that the expansion of personnel is centered around the Advanced Packaging Technology (APT) team in Taiwan. Applied Materials' Taiwan APT team is currently the organization leading major equipment development"
X Link 2025-07-18T01:32Z [----] followers, [----] engagements
"Check out this footage of ASML's 2nm EUV tool in the Rapidus 2nm fab: Rapidus Achieves Significant Milestone at its State-of-the-Art Foundry with Prototyping of Leading-Edge 2nm GAA Transistors: First prototypes successfully demonstrate electrical characteristics as the company continues to hit targets leading up to [----] mass-production TOKYO Rapidus Achieves Significant Milestone at its State-of-the-Art Foundry with Prototyping of Leading-Edge 2nm GAA Transistors: First prototypes successfully demonstrate electrical characteristics as the company continues to hit targets leading up to 2027"
X Link 2025-07-18T20:55Z [----] followers, 29.4K engagements
"Why local AI image generation beats @midjourney: ✅ One-time hardware cost zero monthly fees ⚡ Instant results. No queues no Relax Mode 🔒 Private by default (no $60 paywall for stealth) 💼 Unlimited commercial use no fine print 🧠 Fully offline. No Discord no cloud dependency Midjourney charges up to $100+/month for features your laptop can now do for free. AMDs nice demos show whats coming. We'll need continued innovation on local models to make them better faster and easier to use. But if we assume the future will be better than the present we can see where this is going. Nice work @AMD"
X Link 2025-07-21T18:10Z [----] followers, [----] engagements
"Now they want to scale beyond the $500B Stargate commitment. Love the ambition but skeptical about the capital flow. "we are planning to significantly expand the ambitions of stargate past the $500 billion commitment we announced in january" we have signed a deal for an additional [---] gigawatts of capacity with oracle as part of stargate. easy to throw around numbers but this is a gigantic infrastructure project. some progress photos from abilene: https://t.co/JfutuoYvn9 we have signed a deal for an additional [---] gigawatts of capacity with oracle as part of stargate. easy to throw around"
X Link 2025-07-22T12:55Z [----] followers, [---] engagements
"AMD's AI hardware is competitive. Software is catching up. The next hurdle for enterprises: integration. Nvidia offers a turnkey enterprise AI stack. AMD by contrast offers a toolkit. Some assembly required. They seem to have partner resources like webinars and training courses to help. Would like to hear more from enterprises using Instinct. Maybe Silo AI helps with that enterprise sales engineering too "Silo AI brings a team of world-class AI scientists and engineers to AMD experienced in developing cutting-edge AI models platforms and solutions for large enterprise customers including"
X Link 2025-07-22T14:18Z [----] followers, [----] engagements
"@levie Agree. Plus contributions from Agentic AI autonomy robotics. https://www.chipstrat.com/p/nvidias-climb-to-4t-and-the-peaks https://www.chipstrat.com/p/nvidias-climb-to-4t-and-the-peaks"
X Link 2025-07-22T18:16Z [----] followers, [----] engagements
"Nvidia is worth $4T. But this isnt the top. A 🧵on how @nvidia got here and whats fueling the next climb. $NVDA 👇 Inference is exploding. Autonomy & robotics are trillion-dollar markets. Annual cadence + constrained supply preserve growth. China is an opportunity for autonomy and robotics but a risk too. The $4T climb is just another waypoint for @nvidia $NVDA 👇 https://t.co/nN2cPgTiQR Inference is exploding. Autonomy & robotics are trillion-dollar markets. Annual cadence + constrained supply preserve growth. China is an opportunity for autonomy and robotics but a risk too. The $4T climb is"
X Link 2025-07-22T20:01Z [----] followers, [----] engagements
"1. Nvidia made the GenAI wave possible. @nvidia was the only company ready to scale when ChatGPT went viral. Training clusters All Nvidia. But wouldnt rivals move quickly to challenge its dominance in large-scale training They couldnt. Nvidias scale-up and scale-out interconnect hardware is only half the battle. Efficient training at scale requires ultra-low-latency lossless communication across accelerators. Thats where software comes in. Nvidias NCCL makes frontier-scale training even possible. So when the GenAI wave hit Nvidia was the only supplier with a production-ready massively"
X Link 2025-07-22T20:01Z [----] followers, [--] engagements
"3. The Optimistic View (which played out) Alternatively when Nvidia hit $1T in [----] optimists believed ROI would accrue quickly as models were deployed and CapEx would continue to grow. Something like this (see example below). Fortunately for @nvidia this is the path were on Inference is here and while ROI is hard to quantify its definitely compounding. Were off the shaky ground of "training scaling hope" and into the inference era. 👇 $NVDA"
X Link 2025-07-22T20:01Z [----] followers, [--] engagements
"4. The game has shifted. Inference is the engine of the AI economy now and the name of the game. Training matters too but inference is 📈 ChatGPT alone handles 2.5B queries per day. Enterprises are deploying GenAI across workflows and they want ROI not benchmarks and research papers. Heres an eye-opening anecdote from @IsabelleBiscuit at the Wall Street Journal. Accounting firm RSM is finding enough value in generative AI to justify a fivefold increase in its GenAI investment Why 👇 @nvidia $NVDA"
X Link 2025-07-22T20:01Z [----] followers, [--] engagements
"5. Enter agentic AI and reasoning models. These models plan use tools and take actions across multi-step workflows. They generate 510x more tokens per query. They make GenAI more useful. Agentic AI was the core unlock for RSM. They connected multiple tasks into automated workflows powered by agents that can take action or make decisions. This shift increased productivity significantly. When AI can reason @nvidia GPUs generate more value across more use cases. But theres an even more important implementation detail. Agentic thinking increases the amount of tokens generated per query often by 5"
X Link 2025-07-22T20:01Z [----] followers, [--] engagements
"6. Constrained supply and annual upgrades are secretly brilliant. One underappreciated driver of Nvidias sustained revenue growth is the persistent constraint on supply. The AI infrastructure buildout is unfolding more slowly than demand suggests largely due to TSMCs limited CoWoS advanced packaging capacity and the time-intensive process of building liquid-cooled datacenters with sufficient power access. (Unless you're @elonmusk building Colossus [--] and 2). This is actually good for Nvidia It keeps prices high stretches demand across multiple quarters and gives customers time to realize"
X Link 2025-07-22T20:01Z [----] followers, [---] engagements
"9. More Growth Ahead But heres why I think the future still looks compelling and why remarkably the worlds largest company still has room to grow. 👇 @nvidia $NVDA"
X Link 2025-07-22T20:01Z [----] followers, [--] engagements
"10. Autonomy Robotics China Nvidias next growth wave spans autonomy robotics and Chinaeach a massive market in its own right. Goldman sees 90% CAGR in one segment. Morgan Stanley sees trillions in another. And China is key to Nvidias success but adds risk. At the center of it all is Nvidias Three-Computer thesis: train simulate deploy. And its already taking off. 👇"
X Link 2025-07-22T20:01Z [----] followers, [---] engagements
"@IntTradeNetwork @nvidia Yes 💯lack of focus on the strategic points that matter"
X Link 2025-07-22T22:41Z [----] followers, [--] engagements
"A good reminder for folks outside the tech bubble that ChatGPT isnt as pervasive as it seems; yet this growth is a sign it someday will be. Takes time. Note that this asks if theyve ever used it not regularly use it Pew have updated their ChatGPT polling from a year ago. Usage has roughly doubled since summer [----]. The share of employed adults who use ChatGPT for work has roughly tripled over the same period to 28%. Two thirds of US adults have still not yet used ChatGPT. https://t.co/bY0Gghv7Tg Pew have updated their ChatGPT polling from a year ago. Usage has roughly doubled since summer"
X Link 2025-07-23T13:55Z [----] followers, [---] engagements
"Generative video is becoming popular So many of you are loving turning your photos into short videos in the @Geminiapp and theGemini API. Next up well be rolling this feature out to @YouTube Shorts and @GooglePhotos.And soonRemixyour Google Photos into comics sketches + 3D animations. https://t.co/ct8AYvJpA7 So many of you are loving turning your photos into short videos in the @Geminiapp and theGemini API. Next up well be rolling this feature out to @YouTube Shorts and @GooglePhotos.And soonRemixyour Google Photos into comics sketches + 3D animations. https://t.co/ct8AYvJpA7"
X Link 2025-07-23T16:27Z [----] followers, [---] engagements
"Yes agree the aggressive financing is necessary. just that it might not all come through depending on the customer. Meta can self-fund so Zucks buildout claims are credible. OpenAI depends on several patrons each with their own funding sources and commitments . if Im Oracle I temper expectations. (They didnt even seem to post this announcement to their newsroom so I think they recognize this) xAI sits in the middle. Elon has Tesla/X/SpaceX revenue (though unclear how often he can tap it) plus unconventional financing. If he says hes building 550K GPU Colossus [--] odds are it happens"
X Link 2025-07-24T07:22Z [----] followers, [--] engagements
"Ben Thompson absolutely nails it here. The insight that the CFO is typically the bottleneck for disruptive innovation yet in Googles case that dam has seemingly broken is sharp well-argued and seems right. 7-24-2025 ($) Google Earnings Google Flips the Switch on Cloud Search Notes https://t.co/69vJUThVop 7-24-2025 ($) Google Earnings Google Flips the Switch on Cloud Search Notes https://t.co/69vJUThVop"
X Link 2025-07-24T12:24Z [----] followers, [----] engagements
"Cadence Earnings Q2 2025: Rev: $1.275B (+20% y/y) EPS: $1.65 (non-GAAP) $0.59 (GAAP incl. $140M legal charge) FY25 guide raised Segment growth: EDA +16% (71% of rev) IP +25% (13%) System Design & Analysis +35% (16%) @Cadence $CDNS"
X Link 2025-07-28T21:19Z [----] followers, [----] engagements
"Chiplets demand new tools. 3D-IC and heterogeneous integration break traditional flows. Chiplets need full-system floorplanning thermal + interconnect optimization. Cadence is betting on agentic AI to handle the complexity"
X Link 2025-07-28T21:43Z [----] followers, [---] engagements
"Agentic AI scales design Cadence says Cerebrus AI Studio can now manage 30-50M instance SoC designs up from 3-5M. Thats a 10x leap in scale just as chiplets' integration push limits of human-led flows"
X Link 2025-07-28T21:44Z [----] followers, [--] engagements
"Big gain for Nvidia (big loss for Groq) Hosted my first hackathon @lifebypixels is going to change the game at @NVIDIAAIDev and how we engage with the community Hosted my first hackathon @lifebypixels is going to change the game at @NVIDIAAIDev and how we engage with the community"
X Link 2025-07-28T23:35Z [----] followers, [---] engagements
"Everywhere theres AI theres Micron. Today Micron announced a new portfolio of SSDs purpose-built for AI workloads with three industry firsts that raise the bar for performance capacity and efficiency in AI data centers. 🧵 $MU @MicronTech"
X Link 2025-07-29T14:56Z [----] followers, [----] engagements
"Why it matters: AI models are bigger context windows longer and token throughput demands higher. Storage must scale in performance and density while cutting energy use. Microns G9 NAND plus vertical integration delivers speed efficiency and security at every stage of the AI data pipeline. $MU @MicronTech"
X Link 2025-07-29T14:56Z [----] followers, [---] engagements
"From data lakes transformation training inference Micron now has an SSD for every stage: [----] ION: massive capacity for capture & ingest 7600: low-latency transformation 9650: PCIe Gen6 throughput for training/inference Everywhere theres AI theres Micron. $MU @MicronTech"
X Link 2025-07-29T14:56Z [----] followers, [---] engagements
"Local GenAI at frontier scale is real. Heres the proof 🧵 @AMD runs @AIatMetas 109Bparameter Llama4Scout with toolcalling locally on a Ryzen AI MAX+ laptop. Enabled by: Large memory in consumer silicon MCP dev tooling Quantized models with huge context windows"
X Link 2025-07-29T22:01Z [----] followers, [----] engagements
"Yesterday's cloud models are here at the edge. Models at this scale usually need multiGPU datacenter rigs with 80GB+ VRAM each. @AIatAMD pulled it off with: 96GB Variable Graphics Memory Optimized llama.cpp via Vulkan 4bit quantization CPU + GPU + NPU acceleration in Ryzen AI MAX+"
X Link 2025-07-29T22:01Z [----] followers, [--] engagements
"AMD Variable Graphics Memory lets Ryzen AI 300series (and newer) CPUs turn a chunk of system RAM into dedicated VRAM for the iGPU"
X Link 2025-07-29T22:01Z [----] followers, [--] engagements
"DISCLAIMER: The video is sped up 5X. So the demo is useful and on a laptop but it is still time-consuming. Yet there are often use cases like Deep Research where it's fine to fire off a query and wait for it to run in the background. On the other hand painfully long wait times are an argument for more powerful AI workstations or even little desktop AI servers. Keep it local but much faster"
X Link 2025-07-29T22:01Z [----] followers, [--] engagements
"The other local breakthrough: long context length. Many local LLMs cap at 4K32K tokens. This ran at 256K tokens (Flash Attention + KV cache Q8). Thats enough to keep an entire SEC filing research paper or large codebase in memory while reasoning"
X Link 2025-07-29T22:01Z [----] followers, [--] engagements
"MCP + longcontext local inference changes whats possible. MCP lets an LLM call APIs browse and automate apps. Long context means it can remember all intermediate results and instructions in one session chaining actions together without losing state"
X Link 2025-07-29T22:01Z [----] followers, [--] engagements
"Of course it's private too. Lots of work family and personal use cases where I might as well keep information photos and videos private if I can"
X Link 2025-07-29T22:01Z [----] followers, [---] engagements
"MCP + long context + big local memory = private capable agentic AI that works across tools and steps. No cloud required. AI PCs might be a real pullforward in [----] at this rate not just hype. Need developers to make that come true though Thats it for today. $AMD @AMD"
X Link 2025-07-29T22:01Z [----] followers, [---] engagements
"@zephyr_z9 @boztank The caveat is that OAI doesn't have a sustainable business (yet). @Meta does. Zuck can afford the CapEx and OpEx; he's for sure in this game in [--] years. OpenAI depends on Microsoft Softbank Oracle benevolence. Less clear how that plays out in [--] years"
X Link 2025-07-30T13:54Z [----] followers, [---] engagements
"No Apple No problem says Qualcomm. "CEO Cristiano Amon: We're forecasting fiscal twenty twenty five to be the second consecutive year of greater than 15% YoY growth in total QCT non-Apple revenues" $QCOM @Qualcomm"
X Link 2025-07-30T20:55Z [----] followers, [----] engagements
"Qualcomm re-enters the AI datacenter market. Strategy: ARM Orion CPUs + NPUs optimized for inference efficiency AlphaWave interconnect IP and NVIDIA NVLink Fusion compatibility. In talks with a hyperscaler; targeting revenue in FY28. $QCOM @Qualcomm"
X Link 2025-07-30T21:00Z [----] followers, [----] engagements
"Qualcomm is formally entering the robotics game. Target: $1T TAM in the next decade. Snapdragons heterogeneous compute (CPU+GPU+NPU) lowlatency connectivity and sensor fusion built for automotive maps well to humanoids & industrial robots. $QCOM @Qualcomm"
X Link 2025-07-30T21:07Z [----] followers, [----] engagements
"Qualcomms AI datacenter push: Orion ARM CPUs for hyperscalers & inference head nodes AI accelerator cards + full racks focused on tokens/$ & tokens/W AlphaWave IP for leading scaleout connectivity In advanced talks with a major customer; aiming for a "halo effect" to validate platform $QCOM @Qualcomm"
X Link 2025-07-30T21:11Z [----] followers, [---] engagements
"$ARM Q1 FY26 revenue hit $1.05B. Best Q1 ever 2nd highest quarter in company history. Royalty revenue $585M (+25% YoY). Licensing $468M. AI demand is driving record performance. @Arm"
X Link 2025-07-30T21:15Z [----] followers, [---] engagements
"Microsoft CEO Satya Nadella: "It was a very strong close to what was a record fiscal year for us. All up Microsoft Cloud surpassed $168B in annual revenue up 23%. The rate of innovation and the speed of diffusion is unlike anything we have seen. To that end we are building the most comprehensive suite of AI products and tech stack at massive scale." @Microsoft $MSFT"
X Link 2025-07-30T21:35Z [----] followers, [----] engagements
"$ARM says demand for its Compute Subsystems (CSS) exceeded expectations [--] customers in market; royalties [--] Arm v9 [--] new CSS licenses this quarter (2 DC [--] PC) Xiaomi X Ring [--] & Samsung Galaxy Flip [--] ship on CSS-based Exynos [----] New Zena CSS targets AI-driven automotive Major OEM commits to multi-gen Arm GPU for flagship phones @Arm"
X Link 2025-07-30T21:40Z [----] followers, [----] engagements
"$ARM CEO says @Nvidia is giving them a massive boost: from nearly zero share in Hopper (x86 head nodes) to almost total share in Blackwell (Grace Blackwell). Combined with steady wins in generalpurpose data centers Arm says hyperscaler CPU share is now nearing 50%. @Arm"
X Link 2025-07-30T21:51Z [----] followers, [----] engagements
"The [----] narrative: 💀 AI will kill Google Search 💀 Search ads will collapse 💀 Google is doomed [----] reality: 📈 Search revenue is growing 📈 Ads are thriving 📈 GenAI is additive What happened Read it all here👇 Why GenAI is Good for Google. Also: Google Search vs. ChatGPT Google vs Meta Risks. 👇 https://t.co/Xn5DdetSdq https://t.co/wKQOFq1Pf7 Why GenAI is Good for Google. Also: Google Search vs. ChatGPT Google vs Meta Risks. 👇 https://t.co/Xn5DdetSdq https://t.co/wKQOFq1Pf7"
X Link 2025-07-31T17:04Z [----] followers, [----] engagements
"https://www.chipstrat.com/p/why-genai-is-good-for-google https://www.chipstrat.com/p/why-genai-is-good-for-google"
X Link 2025-07-31T17:04Z [----] followers, [---] engagements
"🔥🔥🔥 "The winners may not be the ones with the best models but the ones with the best financing terms. When the debt comes due having the smartest chatbot wont matter if you cant pay the bills." This is an underappreciated but important angle The AI arms race isn't just about chips it's about who can finance fastest. $CRWV flooding markets w/ debt $NVDA powering trillion-dollar capex OpenAI's Stargate backed by banks xAI doubling GPUs. The real race is in credit markets. Full breakdown: https://t.co/k8AtbG4JWY The AI arms race isn't just about chips it's about who can finance fastest. $CRWV"
X Link 2025-07-31T20:50Z [----] followers, [----] engagements
"Strong growth but AWS is late to GenAI and still supply constrained. The risk is that Azure and GCP scale faster and lock in share before AWS can close the gap. From Andy Jassy on the AWS earnings call: "In the rapidly evolving world of generative AI AWS continues to build a large fast growing triple digit year over year percentage multibillion dollar business with more demand than we have supply for at the moment." $AMZN $GOOG $MSFT"
X Link 2025-07-31T21:25Z [----] followers, [----] engagements
"Jassy's rebuttal to "youre growing slower than Azure/GCP: Growth % can mislead when youre already much bigger. Azure is 65% the size of AWS so even with lower % growth it can add considerable revenue in dollars. Also calls on other value props of AWS like performance reliability etc $AMZN $MSFT $GOOG"
X Link 2025-07-31T21:41Z [----] followers, [---] engagements
"A better Siri is nice but the real unlock is empowering 3rdparty devs including local GenAI via Foundation Models Framework. Tim Cook says Apple is putting all of our energy behind AI. Hopefully that extends to the dev ecosystem. Some AFM talent has already been poached $AAPL @Apple"
X Link 2025-07-31T21:52Z [----] followers, [---] engagements
"GenAI will transform productivity apps. Will the incumbents reinvent their own software or will startups beat them to it We are working on productivity (apps) now because that is where the most value is created. You can make a lot of money doing this and build your team and tech." Fundamental Research Labs nabs $30M to build AI agents across verticals TechCrunch https://t.co/Q2OI2uVFCC Fundamental Research Labs nabs $30M to build AI agents across verticals TechCrunch https://t.co/Q2OI2uVFCC"
X Link 2025-08-01T15:22Z [----] followers, [---] engagements
"Google Search is dead long live ChatGPT right Not so fast. Revenue: UP 📈 Search: UP 🔍 Earnings: Strong 💪 GenAI isnt killing Google; its boosting Google Heres why. 👇 $GOOG"
X Link 2025-08-05T15:11Z [----] followers, [----] engagements
"https://www.chipstrat.com/p/why-genai-is-good-for-google https://www.chipstrat.com/p/why-genai-is-good-for-google"
X Link 2025-08-05T15:11Z [----] followers, [---] engagements
"gpt-oss-120b for production general purpose high reasoning use cases that fits into a single H100 GPU (117B parameters with 5.1B active parameters) gpt-oss-20b for lower latency and local or specialized use cases (21B parameters with 3.6B active parameters) Our open models are here. Both of them. https://t.co/9tFxefOXcg Our open models are here. Both of them. https://t.co/9tFxefOXcg"
X Link 2025-08-05T17:02Z [----] followers, [----] engagements
"AMD announces its 33rd consecutive quarter of YoY CPU server share gains 🔥 Thats over [--] years of uninterrupted progress an incredible streak and a huge testament to the teams making it happen. $AMD @AMD"
X Link 2025-08-05T21:07Z [----] followers, [---] engagements
"🚨 AMD MI350 Series entered volume production ahead of schedule in June. 📈 Steep ramp in 2H [----] to power multiple large-scale deployments including Oracles 27000+ node AI cluster with MI355X accelerators. $AMD @AMD"
X Link 2025-08-05T21:12Z [----] followers, [----] engagements
"Its the inference era where atscale inference workloads are the name of the game. $AMD says MI355 delivers up to 40% more tokens per dollar for said workloads. Investors will be watching for this to show up as higher Instinct sales and market share gains. Hope to hear more in the commentary. @AMD"
X Link 2025-08-05T21:14Z [----] followers, [----] engagements
"The MI350 ramp is the big Instinct story for Q3 driven by deployments at multiple customers. So far it doesn't sound like we'll see much of that quite yet in Q2. If execution stays on track Q3 could be the start of a small share uptick $AMD @AMD"
X Link 2025-08-05T21:20Z [----] followers, [----] engagements
"AMDs AI roadmap is aimed at gigawattscale deployments the kind needed to reach tens of billions in annual Instinct revenue. Strong MI350 traction + positive MI400 feedback + relationships with right customers could make that ramp a reality. $AMD @AMD"
X Link 2025-08-05T21:58Z [----] followers, [---] engagements
"Inference is becoming more complex and tokenintensive as reasoning models and distributed inference make rackscale architectures important for inference as well. $AMD aims to be a fullscale solution provider not just a chip vendor for large deployments and says customers are validating that approach. @AMD"
X Link 2025-08-05T22:03Z [----] followers, [---] engagements
"As I said on @chipstrat back in January CPU TAM is increasing thanks to AI. $AMD is seeing the same"
X Link 2025-08-05T22:06Z [----] followers, [----] engagements
"AWS is losing share just as GenAI inference demand explodes (reasoning & agents) $AMZN cant afford to fall behind because half of Amazons profit comes from AWS Time to stop coasting and start buying GPUs. If Nvidia is slow to allocate get some Instinct 355X too. Fast. Trainium2/3 is a long-term TCO play. Gotta shore up GPUs ASAP. From Q2 call: Jassy: we could be doing more revenue. we're working hard on changing that and how much capacity we have. the single biggest constraint is power. Get creative to overcome those constraints $AMZN needs to convince investors it has a plan here. $AMZN"
X Link 2025-08-06T20:29Z [----] followers, [----] engagements
"Read more here: https://www.chipstrat.com/p/aws-is-missing-the-ai-infra-moment https://www.chipstrat.com/p/aws-is-missing-the-ai-infra-moment"
X Link 2025-08-06T20:29Z [----] followers, [---] engagements
"SkyWater increased American capacity for foundational silicon significantly. If you haven't heard of $SKYT I wrote a short intro below. https://t.co/yDolBAutll https://t.co/yDolBAutll"
X Link 2025-08-06T22:05Z [----] followers, [----] engagements
"Watching closely. $AMZN relies heavily on AWS: 18% of revenue but 53% of operating profit. Yet AWS grew just 17% last quarter while Google Cloud grew 32% and Azure 39%. Amazon cant afford to coast during the GenAI wave Amazon needs to regain AI infra momentum fast"
X Link 2025-08-07T15:35Z [----] followers, [--] engagements
"@munster_gene Definitely agree that Google's search user experience can and will be improved. The good news for Google is that LLMs make everything about search better. And Google has something that OpenAI doesn't. a profitable business model"
X Link 2025-08-08T15:18Z [----] followers, [----] engagements
"@handleym99 1/ OAI will make money in two ways A) commercial B) consumer A: The commercial biz will make money from the API (supporting "wrapper" companies) plus selling subscriptions (e.g. $200/mo). But that won't be enough to support widespread free consumer inference"
X Link 2025-08-08T17:58Z [----] followers, [--] engagements
"Joined @IvanaSpear & @NPetallides on @SchwabNetwork to talk Mag [--]. Nvidia: High ROIC leading. Azure & GCP: Winning AI cloud; AWS short on GPUs & power. Google: GenAI boosts ads. Apple: On-device AI still forming. $NVDA $MSFT $GOOG $AAPL $META $AMZN $TSLA https://www.youtube.com/watchv=Q6Chi6JXAbQ https://www.youtube.com/watchv=Q6Chi6JXAbQ"
X Link 2025-08-11T20:23Z [----] followers, [----] engagements
"Demand steady pricing power doing the heavy lifting for $MU as it raises Q4 guide: Rev: $10.7B $11.2B GM: 42% 44.5% EPS: $2.50 $2.85 Sumit Sadana: This is primarily about pricing pricing has been strong across end markets"
X Link 2025-08-12T14:13Z [----] followers, [---] engagements
"$MU isnt the commodity memory maker of old. We have either very robust share or a trajectory of share gains in every single high-margin portion of our business. Mix shift + pricing + share gains are driving results leaving them really well positioned for FY26"
X Link 2025-08-12T14:18Z [----] followers, [---] engagements
"AI boom = double whammy for $MU. HBM demand is surging and with 3:1 HBM wafers to DDR5 fab capacity is being diverted to HBM. Thats tightening DDR5 supply even as demand stays solid driving higher pricing for both HBM and DDR5"
X Link 2025-08-12T14:23Z [----] followers, [----] engagements
"$MU says HBM4E will be even less of a commodity as it becomes part of the compute architecture. Customers plan to embed GPU logic into the HBM base die offloading functions and creating custom memory tailored to their specs"
X Link 2025-08-12T14:28Z [----] followers, [----] engagements
"Continuing the "not a commodity" HBM4e explanation $MU says embedding customer IP in the base die is so costly that most will stick to only one or two suppliers"
X Link 2025-08-12T14:31Z [----] followers, [---] engagements
"@IntTradeNetwork Fair point Maybe custom AI ASIC makers too like Broadcom"
X Link 2025-08-12T14:40Z [----] followers, [--] engagements
"Yes @awscloud is missing the AI infra moment. The cloud wars are heating up and the leader of the last decade is now playing catch-up. AWS was the first major cloud provider with a commanding lead for years. But in the GenAI era the game has shifted. Heres the twist: AWS is only 18% of Amazons revenue yet it drives over half the companys profit. That makes it not just important but existential Since the large language model boom AWS has been losing net new revenue share to Microsoft Azure and Google Cloud. Why - Microsoft bought GPUs early and at scale. - AWS hesitated and now it is supply"
X Link 2025-08-12T18:48Z [----] followers, [---] engagements
"At the @Oppenheimer Tech Conference @Qualcomm's Nakul Duggal says $QCOM will commercialize it's first automotive driving stack with @BMW with more details coming in September at IAA Mobility in Munich. Big milestone for Snapdragon Ride as Qualcomm moves deeper into ADAS"
X Link 2025-08-12T20:07Z [----] followers, [----] engagements
"CoreWeave: increased backlog ($30B) yet questions about how they fund that buildout. Good point from @anurag_rana4 that the biggest CSPs have very profitable primary business to fund the land grab buildout right now. $CRWV $NBIS $MSFT $GOOG $AMZN $ORCL CoreWeave reported adjusted earnings per share loss of [--] cents for the June quarter compared to Wall Street's consensus estimate for a [--] cent loss. @anurag_rana4 breaks down the results on "Bloomberg The Close" https://t.co/FtdaKiMS7Q https://t.co/q5Ou4GB6fk CoreWeave reported adjusted earnings per share loss of [--] cents for the June quarter"
X Link 2025-08-12T21:05Z [----] followers, [----] engagements
"With AI shifting from training to inference especially reasoning its good to see $CRWV positioning as the best purpose-built AI cloud for both. Customer diversification signals confidence in CoreWeaves inference capabilities"
X Link 2025-08-12T21:18Z [----] followers, [---] engagements
"$CRWV seeing a "massive increase" in inference workloads as validated by watching the power meter lol"
X Link 2025-08-12T21:55Z [----] followers, [----] engagements
"WOW update your financial models $CRWV implies a 7-year useful life for GPUs. H100s & A100s still earn $ after training contracts end repurposed for inference. To be fair Jensen has been saying this all along :) $CRWV $NVDA @CoreWeave @nvidia"
X Link 2025-08-12T22:04Z [----] followers, 26.7K engagements
"$QCOM's ADAS play has three flavors - Chip + QCOM software stack (BMW global launch) - Chip + 3rd-party stack (common in China) - Chip + OEMs own stack BMW is the first global rollout of QCOMs co-developed stack opening the door to a verified non-China turn-key merchant stack for other OEMs. From the JP Morgan Hardware & Semis Forum. @Qualcomm"
X Link 2025-08-13T20:36Z [----] followers, [---] engagements
"$QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and centralized compute. The same stack maps directly to robotics drones and humanoids and thanks to the deal with BMW QCOM can resell it globally. @Qualcomm"
X Link 2025-08-13T20:41Z [----] followers, [---] engagements
"$QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip redundancy in mid-tier and swap in higher ADAS silicon for premium. @Qualcomm"
X Link 2025-08-13T20:45Z [----] followers, [----] engagements
"@BMW @Qualcomm https://x.com/theaustinlyons/status/1955732341755928936 $QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip redundancy in mid-tier and swap in higher ADAS silicon for premium. @Qualcomm https://t.co/MspJ9c5U8E https://x.com/theaustinlyons/status/1955732341755928936 $QCOM sees Ride Flex and a portfolio of low-mid-premium offerings for OEMs as the differentiator vs $NVDA & $MBLY. One chip can power cockpit + ADAS in low-tier cars scale to dual-chip"
X Link 2025-08-13T21:27Z [----] followers, [---] engagements
"@BMW @Qualcomm https://x.com/theaustinlyons/status/1955731417503343093 $QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and centralized compute. The same stack maps directly to robotics drones and humanoids and thanks to https://t.co/h0LEk08K07 https://x.com/theaustinlyons/status/1955731417503343093 $QCOM sees its Snapdragon Ride SoC as the perfect Physical AI platform. Sensor fusion low power high performance wireless connectivity advanced cameras object tracking and"
X Link 2025-08-13T21:27Z [----] followers, [---] engagements
"Fun 10-min run with @AlexCoffeyCS on mega tech + semis $AAPL Why partnering for models makes sense $TSLA EV business strain = funding risk for future bets (robotaxis robotics) $MSFT Infra + apps both winning from AI $INTC How it fell behind whats next $AMD How its catching up whats holding it back $MU Why memory matters and Microns position in HBM @SchwabNetwork @Apple @Tesla @Microsoft @intel @AMD @MicronTech https://www.youtube.com/watchv=Qj3XCKVmfUQ https://www.youtube.com/watchv=Qj3XCKVmfUQ"
X Link 2025-08-14T19:05Z [----] followers, [----] engagements
"Watch here https://www.youtube.com/watchv=Qj3XCKVmfUQ https://www.youtube.com/watchv=Qj3XCKVmfUQ"
X Link 2025-08-14T19:05Z [----] followers, [---] engagements
"$AMAT down 10% after hours on softer Q4 guide. China digestion + fab timing weighing near term. Record Q3 rev/EPS beat prior guide with strength in DRAM etch metal dep & OLED display margins. Long-term AI chipmaking build-out HBM/packaging share gains still on track"
X Link 2025-08-14T20:16Z [----] followers, [----] engagements
"Intel can stand on its own the White House just wants to bridge the gap and help find anchor tenants get to full fabs profitability and back to business. I had the pleasure of being on @BBCWorld with @stevelai last night to talk about reports of the White House seeking an equity stake in @intel"
X Link 2025-08-15T12:00Z [----] followers, [--] engagements
"Intel can stand on its own the White House just wants to bridge the gap and help find anchor tenants get to full fabs profitability and back to business. I had the pleasure of being on @BBCWorld with @stevelai last night to talk about reports of the White House seeking an equity stake in @intel $INTC"
X Link 2025-08-15T12:11Z [----] followers, [----] engagements
"Thinking about writing on $NVDA scale-up and scale-out networking but want your feedback first. What are yall seeing from NVIDIA that you like or dont like Pros cons questions hot takes"
X Link 2025-08-16T22:26Z [----] followers, [----] engagements
"Yes risk of monopsony. See $AMATs earnings call; beyond China they are softening the guide due to a leading edge foundry pulling back. @Mojo_flyin @rwang07 @OmerCheeema Ive said it before and Ill say it again; ASMLs risk is monopsony only one buyer of its leading machines. Someone else has to emerge opposite of TSMC. @Mojo_flyin @rwang07 @OmerCheeema Ive said it before and Ill say it again; ASMLs risk is monopsony only one buyer of its leading machines. Someone else has to emerge opposite of TSMC"
X Link 2025-08-17T12:37Z [----] followers, [----] engagements
"@DasTechnocrat @business The equity stake wouldnt be net new money so not enough to stand up Ohio plant. Also the key question is how to force fabless to use Intel Foundry ie get fabless companies to overcome switching costs. Hence our expectation of tarrifs"
X Link 2025-08-18T17:55Z [----] followers, [---] engagements
"@chilling_dc @DasTechnocrat @business @chilling_dc You're saying the administration could say to AMD or Nvidia for example "commit to making something at Intel Foundry or no China market access for you""
X Link 2025-08-18T21:47Z [----] followers, [--] engagements
"@chilling_dc @DasTechnocrat @business Problem with threatening all chips is that's a huge tax on the US tech ecosystem. It would probably be better to tariff only leading-edge chips (to incentivize Intel 18A/14A) or something quite targeted. Not sure Xilinx's FPGAs or AMD's Xbox chips etc should be tariffed"
X Link 2025-08-18T21:58Z [----] followers, [---] engagements
"SoftBanks $2B bet on $INTC is fantastic. For Intel a vote of confidence and a bit of clean capital. Speculatively some potential volume from Arm or related silicon. Meanwhile Masa Son bought in ahead b4 the biggest sovereign backstop gets in. Still need customers and more capital but a good day. @intel SoftBank Group and Intel Corporation Sign $2B Investment Agreement https://t.co/jTn8kYzQWh SoftBank Group and Intel Corporation Sign $2B Investment Agreement https://t.co/jTn8kYzQWh"
X Link 2025-08-19T02:34Z [----] followers, [----] engagements
"$INTC's wild few weeks Jul [--] Earnings: slows Ohio 14A fab build-out Aug [--] Trump calls CEO Lip-Bu Tan highly conflicted and urges him to quit Aug [--] Tan meets Trump at WH; Trump reverses tone calls his story amazing Aug [--] Leak: WH exploring equity stake in Intel to back U.S. fabs Aug [--] AM Report: plan could convert CHIPS grants into 10% govt stake Aug [--] PM SoftBank invests $2B at $23/share Whew. What's next Details on WH equity stake"
X Link 2025-08-19T13:46Z [----] followers, [----] engagements
"See the background slide for $CDNS projection of $1.2T semiconductor TAM in [----] "The acceleration of silicon" - Anirudh Devgan President and CEO of Cadence Learn more: https://t.co/te3CZWD5l6 https://t.co/xeGgFK8Iow "The acceleration of silicon" - Anirudh Devgan President and CEO of Cadence Learn more: https://t.co/te3CZWD5l6 https://t.co/xeGgFK8Iow"
X Link 2025-08-19T17:41Z [----] followers, [----] engagements
"Teslas auto business is nearly worthless but the robotaxi business will be huge in [----] 🤔😅 We expect Tesla to win 35% market share versus competitors Waymo at 15% Uber at 38% and Lyft at 13% generating almost $250 billion in revenue in [----]. After discounting the robotaxi EBITDA of $145 billion at 8.5% discount rate we estimate an implied value of Teslas robotaxi business at $298.61 per share energy business at $30.73 per share and auto business at $28.09 per share totaling an implied fair value of $357.43 per share. William Blair went to Austin to compare $TSLA robotaxi and $GOOGL Waymo:"
X Link 2025-08-20T10:35Z [----] followers, [---] engagements
"Not saying restricted just pointing out everyone wants negotiation power and supply chain diversification but that requires actually doing business with Intel Foundry. And if US government is truly bought into IF for national security it makes sense that the govt would encourage fabless designers to give IF a chance e.g. with a small product line"
X Link 2025-08-20T20:18Z [----] followers, [--] engagements
"@rwang07 $9B in a single month for networking or is that for the quarter"
X Link 2025-08-21T11:21Z [----] followers, [---] engagements
"$NVDA BYOC (bring your own CPU). Integrate NVLink C2C IP into your silicon and get [---] GB/s CPU-to-accelerator bandwidth. Think of it as replacing Grace/Vera on a Bianca board with your own silicon. @nvidia @NVIDIANetworking $NVDA"
X Link 2025-08-21T20:53Z [----] followers, [----] engagements
"15 years ago Andy Grove wrote Losing semiconductor manufacturing means losing the industrial commonsthe knowhow suppliers and ecosystems that drive future innovation. Quite prescient as the U.S. works to rebuild and restore its semi manufacturing base. $INTC How to make an American job. https://t.co/8U5CsvzJB6 https://t.co/x8yfyBb54r How to make an American job. https://t.co/8U5CsvzJB6 https://t.co/x8yfyBb54r"
X Link 2025-08-22T12:30Z [----] followers, [----] engagements
"Managing jitter across distributed datacenters 🤯 You have scale-up scale-out and now scale-across. @NVIDIA moat widens with Spectrum-XGS Ethernet that can tie GPU clusters together across distributed locations. GPU super clusters can now be tied between hyperscalers neoclouds etc. Wild. https://t.co/415uq919Wu https://t.co/ltNqZ0Vbrw You have scale-up scale-out and now scale-across. @NVIDIA moat widens with Spectrum-XGS Ethernet that can tie GPU clusters together across distributed locations. GPU super clusters can now be tied between hyperscalers neoclouds etc. Wild. https://t.co/415uq919Wu"
X Link 2025-08-22T15:25Z [----] followers, [---] engagements
"Looking forward to $INTC updates on: 1) new investors (incl. PE) 2) anchor 14A customers 3) 14A build-out restart Reminder: 20A/18A in AZ was 49% ($15B) co-funded by Brookfield under Intels Smart Capital program. BIG NEWS: The United States of America now owns 10% of Intel one of our great American technology companies. This historic agreement strengthens U.S. leadership in semiconductors which will both grow our economy and help secure Americas technological edge. Thanks to Intel https://t.co/AYMuX14Rgi BIG NEWS: The United States of America now owns 10% of Intel one of our great American"
X Link 2025-08-22T20:44Z [----] followers, [----] engagements
"Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky) US GOVT AGREES TO 433.3MLN SHARES OF INTEL COMMON STOCK AT 20.47/SHARE NOW [-----] US GOVT AGREES TO 433.3MLN SHARES OF INTEL COMMON STOCK AT 20.47/SHARE NOW 24.94"
X Link 2025-08-22T20:55Z [----] followers, 11.6K engagements
"Also US Gov't got in cheaper than Masa and Softbank ($20.47 vs $23) Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky) Hey nice $INTC gets the money now as opposed to CHIPS Act grants which are distributed over time (if you're lucky)"
X Link 2025-08-22T20:58Z [----] followers, [---] engagements
"@PatrickMoorhead @howardlutnick @LipBuTan1 Looks like DJT had a post on Truth Social https://x.com/RapidResponse47/status/1958999176324555140 https://t.co/suLz96EODS https://x.com/RapidResponse47/status/1958999176324555140 https://t.co/suLz96EODS"
X Link 2025-08-22T21:37Z [----] followers, [----] engagements
"Some analysis and thoughts on the U.S. Governments investment in Intel Corporation per the Friday announcement. $INTC @intel"
X Link 2025-08-24T14:04Z [----] followers, [---] engagements
"More analysis and thoughts on the U.S. Governments investment in $INTC"
X Link 2025-08-24T17:22Z [----] followers, [----] engagements
"MI450 is when the competition really starts 👀 $AMD hits hard on third gen of their roadmap Frank AMD Sr Fellow presented today at hotchips conference explaining how scale up BW is 10x greater than scale out BW. This is the reason why MI355's scale up world size of [--] is not enough for frontier diagg prefill+wideEP inferencing and frontier training compared to GB200's https://t.co/bqrXnzQ8Uj Frank AMD Sr Fellow presented today at hotchips conference explaining how scale up BW is 10x greater than scale out BW. This is the reason why MI355's scale up world size of [--] is not enough for frontier"
X Link 2025-08-24T22:22Z [----] followers, [----] engagements
"My thoughts on @dMatrix_AI and 3D-stacked digital in-memory compute (3DIMC): d-Matrix is taking direct aim at the memory wall with 3DIMC vertically stacking DRAM with logic dies (see image below). The company claims a 10x boost in bandwidth and 10x better energy efficiency versus HBM4. These gains could reshape inference economics in an era where datacenter power is capped and agentic AI demands low-latency interactivity and high throughput at scale. Theres precedent here. Nearly [--] years ago AMD bet on stacked DRAM and co-developed HBM. The early pain of costly interposers and low yields"
X Link 2025-08-25T14:03Z [----] followers, [----] engagements
"At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its real or I wouldnt be here.""
X Link 2025-08-25T19:15Z [----] followers, 34.3K engagements
"Don also mentioned Clearwater Forest's 17% IPC uplift from Sierra Forest $INTC At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its real or I wouldnt be here." https://t.co/PBiTyghJxG At @hotchipsorg someone asked if Intel 18A was on track. Don Soltis from $INTC literally held up a Clearwater Forest SoC and basically said Yes.this was built on 18A. Everything were talking about for Clearwater Forest is on 18A. Its"
X Link 2025-08-25T19:18Z [----] followers, [----] engagements
"Rapidus will hit 2nm in [----] a node or two behind $TSM & $INTC. Hence the pitch is time-to-silicon not node leadership. Who values that Curious to see. AI chip startups hyperscalers refining in-house silicon (i.e to catch up to TPU v7) But thats low-volume R&D value prop too. Can it scale to HVM CEO says 25K WPM by [----]. @hotchipsorg"
X Link 2025-08-26T21:42Z [----] followers, [----] engagements
"$NVDA Spectrum-X Ethernet already doing ARR of $10B Plus they still sell Infiniband :) It's all about AI networking these days. $NVDA CFO: "escalating demands of AI compute clusters necessitate high efficiency and low latency networking" It's all about AI networking these days. $NVDA CFO: "escalating demands of AI compute clusters necessitate high efficiency and low latency networking""
X Link 2025-08-27T21:16Z [----] followers, [----] engagements
"$NVDA earnings shout out to @hotchipsorg and $CRWV. "At Hotchips we introduced Spectrum-XGS Ethernet a technology designed to unify disparate data centers into gigascale AI super factories. CoreWeave is an initial adopter of the solution which is projected to double GPU-to-GPU communication speed.""
X Link 2025-08-27T21:18Z [----] followers, [----] engagements
"My takeaways from $NVDA earnings: Networking = huge bright spot China = upside optionality Thor = underappreciated future play Nearly $50B in a single quarter 🤯 Air-cooled RTX Pro servers = nice incremental add End-to-end AI system = Nvidias unbeatable edge @nvidia @NVIDIANetworkng"
X Link 2025-08-27T23:02Z [----] followers, 12.1K engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing