#  @sallywf Sally Ward-Foxton
Sally Ward-Foxton posts on X about ai, ceo, inference, in the the most. They currently have [------] followers and [---] posts still getting attention that total [-----] engagements in the last [--] hours.
### Engagements: [-----] [#](/creator/twitter::19663365/interactions)
- [--] Week [-----] +235%
- [--] Month [------] -30%
- [--] Months [------] +42%
- [--] Year [-------] -32%
### Mentions: [--] [#](/creator/twitter::19663365/posts_active)
- [--] Week [--] +460%
- [--] Month [--] +258%
- [--] Months [--] +145%
- [--] Year [--] +129%
### Followers: [------] [#](/creator/twitter::19663365/followers)
- [--] Week [-----] +0.12%
- [--] Month [-----] +0.52%
- [--] Months [-----] +21%
- [--] Year [-----] +31%
### CreatorRank: [---------] [#](/creator/twitter::19663365/influencer_rank)
### Social Influence
**Social category influence**
[technology brands](/list/technology-brands) 25.58% [stocks](/list/stocks) 18.14% [finance](/list/finance) 6.05% [social networks](/list/social-networks) 4.19% [travel destinations](/list/travel-destinations) 3.72% [events](/list/events) #946 [countries](/list/countries) 3.26% [fashion brands](/list/fashion-brands) 0.93% [cryptocurrencies](/list/cryptocurrencies) 0.47% [currencies](/list/currencies) 0.47%
**Social topic influence**
[ai](/topic/ai) 31.16%, [ceo](/topic/ceo) 9.77%, [inference](/topic/inference) 7.91%, [in the](/topic/in-the) 6.98%, [ip](/topic/ip) 4.65%, [llm](/topic/llm) 3.72%, [london](/topic/london) 2.79%, [$googl](/topic/$googl) 2.33%, [jim](/topic/jim) 2.33%, [data center](/topic/data-center) 2.33%
**Top accounts mentioned or mentioned by**
[@nvidia](/creator/undefined) [@iancutress](/creator/undefined) [@intel](/creator/undefined) [@tenstorrent](/creator/undefined) [@eetimes](/creator/undefined) [@groqinc](/creator/undefined) [@sambanovaai](/creator/undefined) [@cerebras](/creator/undefined) [@synopsys](/creator/undefined) [@ambarellainc](/creator/undefined) [@amd](/creator/undefined) [@brainchipinc](/creator/undefined) [@ayarlabs](/creator/undefined) [@dmatrixai](/creator/undefined) [@ndahad](/creator/undefined) [@rodrigoliang](/creator/undefined) [@esperantotech](/creator/undefined) [@jimkxa](/creator/undefined) [@positronai](/creator/undefined) [@trsohmers](/creator/undefined)
**Top assets mentioned**
[Alphabet Inc Class A (GOOGL)](/topic/$googl) [Microsoft Corp. (MSFT)](/topic/microsoft) [Ambarella, Inc. Ordinary Shares (AMBA)](/topic/$amba) [Flex Ltd. Ordinary Shares (FLEX)](/topic/flex) [Lends (LENDS)](/topic/lends) [Gains (GAINS)](/topic/gains) [Synopsys, Inc. (SNPS)](/topic/synopsys)
### Top Social Posts
Top posts by engagements in the last [--] hours
"@tesla_teamx Intel has made GPUs including a very famous data center GPU for decades"
[X Link](https://x.com/sallywf/status/2019275227105030268) 2026-02-05T05:01Z [----] followers, [--] engagements
"I was actively enjoying an entire feed full of British garden birds this week then made the mistake of clicking on a picture of a poem. Now every tweet is a poem and I'm looking for that one tweet about the semiconductor industry to engage with so I can get back to reality. Argh don't like how the algo is overly reactive. i've spent years liking animal-related content on here but if you like one tweet about potatoes or whatever it's like potatoes all the way down until another animal tweet appears and you have to treat it like a life preserver don't like how the algo is overly reactive. i've"
[X Link](https://x.com/sallywf/status/2022271740160721380) 2026-02-13T11:28Z [----] followers, [---] engagements
"RT @JP_Biz: Ok brother do you think you could make Outlook search work when you have a minute"
[X Link](https://x.com/sallywf/status/2022298238796050807) 2026-02-13T13:14Z [----] followers, [----] engagements
"AI chip startup @positron_ai has raised $230 million as it continues to sell its FPGA-based AI inference servers and work on its chiplet-based ASIC. Among strategic investors were financial trading firms who need really fast tokens @trsohmers told me: https://www.eetimes.com/positron-230-million-funding-led-by-financial-trading-firms/ https://www.eetimes.com/positron-230-million-funding-led-by-financial-trading-firms/"
[X Link](https://x.com/sallywf/status/2019092714357407866) 2026-02-04T16:56Z [----] followers, [----] engagements
"Oh my sweet summer child. Intel will start making GPUs a market dominated by Nvidia https://t.co/QotvII3gMH Intel will start making GPUs a market dominated by Nvidia https://t.co/QotvII3gMH"
[X Link](https://x.com/sallywf/status/2019264963949265042) 2026-02-05T04:20Z [----] followers, [----] engagements
"RT @Ronanchamberss: junior BBC journalist: hey so for my first piece i was thinking about writing something on how London based AI lab Deep"
[X Link](https://x.com/sallywf/status/2020786276450853264) 2026-02-09T09:06Z [----] followers, [--] engagements
"(Reporting from Web Summit Qatar:) @imec_int's Max Mirgoli told me Imec is opening an R&D base in Doha in order to take advantage of talent from countries that might find it hard to get visas for US or Europe: https://www.eetimes.com/imec-to-open-rd-center-in-qatar/ https://www.eetimes.com/imec-to-open-rd-center-in-qatar/"
[X Link](https://x.com/sallywf/status/2020772481817293124) 2026-02-09T08:11Z [----] followers, [---] engagements
"At @websummitqatar I chatted to @Atom_Computing and @qctrlHQ about sovereign quantum computing strategies. Countries should invest in quantum BEFORE we hit quantum advantage they said focusing on developing talent while avoiding techno-nationalism: https://www.eetimes.com/the-sovereign-quantum-race-starts-before-quantum-advantage/ https://www.eetimes.com/the-sovereign-quantum-race-starts-before-quantum-advantage/"
[X Link](https://x.com/sallywf/status/2021554829831602416) 2026-02-11T12:00Z [----] followers, [---] engagements
"3 things I learned at Web Summit Qatar last week: https://youtube.com/shorts/K4rO6IdvVwYfeature=share https://youtube.com/shorts/K4rO6IdvVwYfeature=share"
[X Link](https://x.com/sallywf/status/2021915196667965892) 2026-02-12T11:52Z [----] followers, [---] engagements
"My reporting from Qatar @websummitqatar - all in once place. Had a brilliant trip and hope to return next year. Qatars Deep Tech Ambitions Come Into Focus https://t.co/oMM3rs6XFh https://t.co/OF2ozf5hPL Qatars Deep Tech Ambitions Come Into Focus https://t.co/oMM3rs6XFh https://t.co/OF2ozf5hPL"
[X Link](https://x.com/sallywf/status/2022235555334082666) 2026-02-13T09:05Z [----] followers, [---] engagements
"Qatar plans to capitalise on AI by investing in strategic tech and companies. Which parts of the AI stack should a country like Qatar involve itself in and which parts should it tackle first I spoke with @AyarLabs @dMatrix_AI to get some perspective: https://www.eetimes.com/qatars-ai-ambitions-span-the-tech-stack/ https://www.eetimes.com/qatars-ai-ambitions-span-the-tech-stack/"
[X Link](https://x.com/sallywf/status/2022345932470091836) 2026-02-13T16:23Z [----] followers, [---] engagements
"Breaking News: Flex Logix has been acquired by Analog Devices. The 10-year-old eFPGA companys technical assets including its InferX AI IP and its entire technical team will transfer to existing customer ADI. @efpga @ADI_News https://www.eetimes.com/flex-logix-acquired-by-analog-devices/ https://www.eetimes.com/flex-logix-acquired-by-analog-devices/"
[X Link](https://x.com/sallywf/status/1854938950151917667) 2024-11-08T17:28Z [----] followers, [---] engagements
"Very Excited for @eetimes AI Everywhere - our virtual conference on AI chips which is free to attend. I will be hosting day [--] which is about AI in the data center and moderating a panel about inference at scale. Don't miss it The rapid growth of Large Language Models (#LLMs) has driven the development of valuable #AI capabilities. Check out more in the keynote session by @GroqInc at our virtual event AI Everywhere 👉🏻 Register now: https://t.co/6ng3dtgnO1 https://t.co/KCa9D4jDjf The rapid growth of Large Language Models (#LLMs) has driven the development of valuable #AI capabilities. Check"
[X Link](https://x.com/sallywf/status/1859246752383033702) 2024-11-20T14:45Z [----] followers, [----] engagements
"I had a chat with @BlumindInc at Electronica. The company is pressing forwards with its analog keyword spotting chip and a CNN accelerator. But the company has LLMs in its sights as the tech lends itself to bigger chiplet-based designs: https://www.eetimes.com/blumind-hints-at-big-chip-on-roadmap/ https://www.eetimes.com/blumind-hints-at-big-chip-on-roadmap/"
[X Link](https://x.com/sallywf/status/1859577057904132545) 2024-11-21T12:38Z [----] followers, [---] engagements
".@tenstorrent has raised a whopping $693 million in a round that will put any doubts about the investment community's appetite for AI chip companies aside. The company has $150 million in deals committed mainly IP with next gen chip coming in early 2025: https://www.eetimes.com/tenstorrent-raises-693-million-series-d/ https://www.eetimes.com/tenstorrent-raises-693-million-series-d/"
[X Link](https://x.com/anyuser/status/1863629989087723784) 2024-12-02T17:03Z [----] followers, [----] engagements
"My hot take: @intel will never find another Gelsinger. The 30-year Intel veteran former Intel CTO and former VMWare CEO is irreplaceable. https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/ https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/"
[X Link](https://x.com/anyuser/status/1864243323910730054) 2024-12-04T09:40Z [----] followers, 54K engagements
"@sysadafterdark A potato could tell you're wrong. "Gelsinger killed intel for short term gains": these short term gains you speak of. Are they in the room with us now"
[X Link](https://x.com/sallywf/status/1864350334614716681) 2024-12-04T16:45Z [----] followers, [---] engagements
"Intel has hired [--] board members both with industry chops presumably to counter criticism about the board's experience in the wake of Pat's exit. Former ASML CEO Eric Meurice and longtime Microchip CEO Steve Sanghi (one foundry side one product side). https://www.intc.com/news-events/press-releases/detail/1721/intel-appoints-semiconductor-leaders-eric-meurice-and-steve https://www.intc.com/news-events/press-releases/detail/1721/intel-appoints-semiconductor-leaders-eric-meurice-and-steve"
[X Link](https://x.com/sallywf/status/1864719860468514843) 2024-12-05T17:13Z [----] followers, [---] engagements
"Wondering if beginning more of my tweets with "IT'S ABOUT TO START THE BIGGEST TECH BATTLE OF OUR TIME" (and then getting the product name wrong) would get me [---] million views per tweet Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time: https://t.co/KAUpGpDcKJ Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI"
[X Link](https://x.com/anyuser/status/1869700365299728505) 2024-12-19T11:04Z [----] followers, [----] engagements
"Had a chat with @CEVA_IP's Moshe Sheier about the company's NeuPro-Nano including why tinyML needs a standalone core that can do DSP+AI (not separate accelerators or a scaled-down NeuPro-M) and why the company is already working with @EdgeImpulse. More: https://www.eetimes.com/ceva-seeks-to-exploit-synergies-in-portfolio-with-nano-npu/ https://www.eetimes.com/ceva-seeks-to-exploit-synergies-in-portfolio-with-nano-npu/"
[X Link](https://x.com/sallywf/status/1870105827501064511) 2024-12-20T13:55Z [----] followers, [---] engagements
"In a new type of collaboration @SiMa_Inc and @Synopsys will make an IP product for automotive based on SiMa's ML IP and Synopsys' digital twin tech. The automotive industry is calling for more influence on silicon both companies told me: https://www.eetimes.com/synopsys-collaborates-with-sima-ai-on-automotive-ai-ip/ https://www.eetimes.com/synopsys-collaborates-with-sima-ai-on-automotive-ai-ip/"
[X Link](https://x.com/sallywf/status/1871138749477949713) 2024-12-23T10:20Z [----] followers, [---] engagements
"The US government announced stricter rules on the export of AI chips which now need a license to be exported to almost any country. Closed model weights are also restricted. The aim is to prevent China from developing frontier models. My summary: https://www.eetimes.com/uproar-as-biden-administration-limits-ai-chip-shipments/ https://www.eetimes.com/uproar-as-biden-administration-limits-ai-chip-shipments/"
[X Link](https://x.com/sallywf/status/1879847549680697632) 2025-01-16T11:05Z [----] followers, [---] engagements
"He does say "$100 billion already committed" in this video. (Will OpenAI build chips "we'll see" he says) At WSJ Journal House Davos @OpenAI Chief Product Officer Kevin Weil discusses #Stargate outlines #ChatGPTs [----] roadmap https://t.co/PacIV7mYjV #AI #HPC At WSJ Journal House Davos @OpenAI Chief Product Officer Kevin Weil discusses #Stargate outlines #ChatGPTs [----] roadmap https://t.co/PacIV7mYjV #AI #HPC"
[X Link](https://x.com/sallywf/status/1882366249520521600) 2025-01-23T09:54Z [----] followers, [---] engagements
"Feel like this is very technically feasible and would go a long way towards getting people out of social media echo chambers or at least in showing people echo chambers exist imo all apps with a 'for you' timeline should allow you to toggle a bizzaro switch and browse the 'NOT for you' timeline just to feel what the opposite is like imo all apps with a 'for you' timeline should allow you to toggle a bizzaro switch and browse the 'NOT for you' timeline just to feel what the opposite is like"
[X Link](https://x.com/sallywf/status/1886331176933015666) 2025-02-03T08:29Z [----] followers, [---] engagements
"DARE the largest chip project ever funded by the EU has kicked off with $260 million in funding for HPC chip sovereignty. The money will go to a consortium including @AxeleraAI who will develop an AI accelerator chiplet for HPC: https://www.eetimes.com/european-project-gets-260-million-for-hpc-chip-sovereignty/ https://www.eetimes.com/european-project-gets-260-million-for-hpc-chip-sovereignty/"
[X Link](https://x.com/sallywf/status/1897951599223288109) 2025-03-07T10:04Z [----] followers, [---] engagements
".@EdgeImpulse has been snapped up by @Qualcomm. Congratulations Zach Jan Adam and team (My colleague @ndahad has the story at EE Times dot com). https://www.eetimes.com/qualcomm-hopes-to-expand-its-intelligent-iot-with-edge-impulse-acquisition/ https://www.eetimes.com/qualcomm-hopes-to-expand-its-intelligent-iot-with-edge-impulse-acquisition/"
[X Link](https://x.com/sallywf/status/1899092791470227591) 2025-03-10T13:39Z [----] followers, [----] engagements
"My hot take: A sigh of relief as Lip-Bu Tan is appointed CEO of @intel https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/ https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/"
[X Link](https://x.com/anyuser/status/1899991868198576281) 2025-03-13T01:12Z [----] followers, [----] engagements
"While this year's production felt like less of a spectacle there was certainly plenty of content in Jensen Huang's keynote at @nvidia GTC. He talked a lot about tokens single user speeds and what @dylan522p calls "tokenomics". https://www.eetimes.com/huang-talks-tokens-reveals-roadmap-at-gtc-2025/ https://www.eetimes.com/huang-talks-tokens-reveals-roadmap-at-gtc-2025/"
[X Link](https://x.com/sallywf/status/1902431480309280915) 2025-03-19T18:46Z [----] followers, [---] engagements
"ICYMI: @ndahad and I recap @nvidia GTC and @Synopsys SNUG conferences live from California https://www.youtube.com/watchv=dUViIU7dWYk https://www.youtube.com/watchv=dUViIU7dWYk"
[X Link](https://x.com/sallywf/status/1903303341368807477) 2025-03-22T04:30Z [----] followers, [---] engagements
".@OpenAI's Richard Ho keynoted @Synopsys Snug & I was on the front row Appearing to confirm he is working on AI silicon Ho said his team is addressing every layer of the stack for huge computers where GPUs will nevertheless "be a huge part of that mix": https://www.eetimes.com/openai-hardware-chief-ai-scaling-laws-will-continue/ https://www.eetimes.com/openai-hardware-chief-ai-scaling-laws-will-continue/"
[X Link](https://x.com/sallywf/status/1905568381366853863) 2025-03-28T10:31Z [----] followers, [---] engagements
".@nvidia @NVIDIAAI unveiled some impressive inference performance results at GTC but how did they do it Ian Buck went through some of the details of Dynamo the company's new inference serving software: https://www.eetimes.com/how-did-nvidia-improve-hopper-inference-performance-30x/ https://www.eetimes.com/how-did-nvidia-improve-hopper-inference-performance-30x/"
[X Link](https://x.com/sallywf/status/1906984734951145561) 2025-04-01T08:19Z [----] followers, [---] engagements
"@IanCutress "SXSW London will monitor coverage generated by all attending media to ensure coverage plans and assignments were delivered. Those with no proof of coverage may be declined media accreditation for any future conferences." -- yeah it's a no from me"
[X Link](https://x.com/sallywf/status/1920505853674656085) 2025-05-08T15:47Z [----] followers, [---] engagements
"@IanCutress @sxsw I expect they get a lot of chancers trying to get in to see the "celebrities" but making admittance dependent on coverage means they will encourage more chancers and less bona fide press which I expect is the opposite of what they want to achieve"
[X Link](https://x.com/sallywf/status/1920512093062615057) 2025-05-08T16:12Z [----] followers, [---] engagements
"@IanCutress @nvidia @MediaTek @MarvellTech @AlchipTECH @dylan522p It's like a Formula [--] Paddock interview hahahaha"
[X Link](https://x.com/sallywf/status/1924566210462810328) 2025-05-19T20:41Z [----] followers, [---] engagements
"Are you at The AI Summit in London tomorrow I will be presenting two short fireside chats with @SambaNovaAI CEO @RodrigoLiang. Pop along and hear Rodrigo's views on a variety of topics from open source vs proprietary models to AI's energy consumption: https://london.theaisummit.com/speakers/sally-ward-foxton https://london.theaisummit.com/speakers/sally-ward-foxton"
[X Link](https://x.com/sallywf/status/1932415545548079126) 2025-06-10T12:32Z [----] followers, [---] engagements
"@DylanOnChips Thanks for your service 🫡"
[X Link](https://x.com/sallywf/status/1932452731890933881) 2025-06-10T15:00Z [----] followers, [---] engagements
".@EnCharge_AI has launched a [---] TOPS [--] TOPS/W AI chip for the PC. It uses the company's capacitor-based analog compute scheme which is not affected by temperature or process variations. More in the article: https://www.eetimes.com/encharge-picks-the-pc-for-its-first-analog-ai-chip/ https://www.eetimes.com/encharge-picks-the-pc-for-its-first-analog-ai-chip/"
[X Link](https://x.com/sallywf/status/1934521492756181260) 2025-06-16T08:00Z [----] followers, [---] engagements
"Sad news as @EsperantoTech all but shuts down. The tech is still for sale or can be licensed CEO Art Swift told me but the majority of employees have already moved on. The company's European teams were decimated by competitors Swift said: https://www.eetimes.com/ai-startup-esperanto-winds-down-silicon-business/ https://www.eetimes.com/ai-startup-esperanto-winds-down-silicon-business/"
[X Link](https://x.com/sallywf/status/1942113628590571650) 2025-07-07T06:49Z [----] followers, [---] engagements
"My report from @DACconference is on the state of multi-agent AI systems in EDA and the vision for increasingly autonomous workflows. Featuring @Microsoft @Synopsys and @Siemens EDA: https://www.eetimes.com/dac-2025-towards-multi-agent-systems-in-eda/ https://www.eetimes.com/dac-2025-towards-multi-agent-systems-in-eda/"
[X Link](https://x.com/sallywf/status/1942289400735392035) 2025-07-07T18:27Z [----] followers, [---] engagements
"Big milestone for South Korean startup @FuriosaAI as LG picks Furiosa hardware for its LLM appliance. I spoke to CEO June Paik about the rigorous testing LG put Furiosa's hardware through and Furiosa's new multi-chip tensor parallel capabilities: https://www.eetimes.com/lg-to-ship-ai-appliance-with-furiosa-chips/ https://www.eetimes.com/lg-to-ship-ai-appliance-with-furiosa-chips/"
[X Link](https://x.com/sallywf/status/1947982308352335951) 2025-07-23T11:29Z [----] followers, [---] engagements
"OK I'll bite. Yes making a transformer chip does mean making a large matmul engine. But there is a lot more to efficient transformer inference than GEMM/FLOPS. You need to be able to feed and use those FLOPS efficiently. (1/x) Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well https://t.co/Ju8QX9gcc7 Although "baking transformers in silicon" may sound cool it is mostly"
[X Link](https://x.com/anyuser/status/1948374682027094171) 2025-07-24T13:28Z [----] followers, 27.4K engagements
"In other words I think there are several dimensions that could be specialised on for transformers beyond generic GEMM capability. This is instinctive to people who understand the concept of ASICs outside of data center/AI (3/x)"
[X Link](https://x.com/anyuser/status/1948374807269027843) 2025-07-24T13:28Z [----] followers, [----] engagements
"If its just about FLOPS why is current best practice disaggregated inference ie split prefill and decode stages across heterogeneous hardware (currently mostly across different types of GPUs but if transformers are just about FLOPS why use different hardware at all) (4/x)"
[X Link](https://x.com/anyuser/status/1948374919667925033) 2025-07-24T13:29Z [----] followers, [----] engagements
"If theres no point optimizing for AI/transformers beyond GPU how do other semiconductor companies with AI-specialised architectures (see: @GroqInc @CerebrasSystems) consistently beat Nvidias inference performance I guess it depends on whether you think best = fastest. (5/x)"
[X Link](https://x.com/sallywf/status/1948375086563479953) 2025-07-24T13:29Z [----] followers, [----] engagements
"@GroqInc @cerebras Speaking of Groq and Cerebras neither uses HBM so I dont think your memory wall argument holds water here either. (6/x)"
[X Link](https://x.com/anyuser/status/1948375156218384528) 2025-07-24T13:30Z [----] followers, [----] engagements
"@GroqInc @CerebrasSystems Im sure you know that GPUs often have famously poor memory bandwidth utilization for AI (in the order of 30%); with hardware optimization it is perfectly possible to utilize more of that bandwidth than GPUs can/do (see: @positron_ai) (7/x)"
[X Link](https://x.com/sallywf/status/1948375235444593065) 2025-07-24T13:30Z [----] followers, [----] engagements
"@itsclivetime Thanks Clive really appreciate your thoughtful response and I agree on a lot. Partly it was the flippancy of the original tweet I wanted to temper. To your point - I do hope programmability remains if it means we can look forward to seeing more brain-like models in production"
[X Link](https://x.com/sallywf/status/1948674246785270124) 2025-07-25T09:18Z [----] followers, [---] engagements
"@SemiAnalysis_ If there is a joke here I don't get it. @grok please educate me as to how this isn't misogynistic bullshit"
[X Link](https://x.com/sallywf/status/1950481577139417498) 2025-07-30T09:00Z [----] followers, [---] engagements
"South Korean edge AI chip startup DeepX has partnered with Chinese hyperscaler Baidu. The companies will optimize Baidus ERNIE LLM for DeepX hardware. This will enable direct access to one of the largest AI developer communities in China: https://www.eetimes.com/deepx-partners-with-baidu-to-attack-china-market/ https://www.eetimes.com/deepx-partners-with-baidu-to-attack-china-market/"
[X Link](https://x.com/sallywf/status/1956342487062126665) 2025-08-15T13:09Z [----] followers, [---] engagements
".@Cornami_Inc is touting its silicon for FHE or AI acceleration but the real promise is in doing both - AI inference on encrypted data - former CEO Wally Rhines told me. Even without the encryption the company has competitive tokens-per-second figures: https://www.eetimes.com/cornami-targets-fhe-encrypted-ai-in-real-time/ https://www.eetimes.com/cornami-targets-fhe-encrypted-ai-in-real-time/"
[X Link](https://x.com/sallywf/status/1962794979291431372) 2025-09-02T08:29Z [----] followers, [---] engagements
"Next week I'll be at the AI Infra Summit in Santa Clara - please come and see my panel discussion about how AI in EDA will affect chip startups featuring AI EDA and Startup experts @Cognichip Stelios Diamantidis @Synopsys Arun Venkatachar and Silicon Catalyst 's Laura Swan"
[X Link](https://x.com/sallywf/status/1962796521843208500) 2025-09-02T08:35Z [----] followers, [----] engagements
"An ambitious new collaboration Project Orbion aims to build a real-time digital twin of the Earth by combining various imaging data sources and offering them for commercial use. I spoke to some of the partners in San Francisco this week: https://www.eetimes.com/project-aims-to-make-real-time-digital-twin-of-the-world/ https://www.eetimes.com/project-aims-to-make-real-time-digital-twin-of-the-world/"
[X Link](https://x.com/sallywf/status/1966550717482532913) 2025-09-12T17:13Z [----] followers, [---] engagements
"I had a great time chatting to @Infineon's Gregory Guez about the major trends defining AI in the smart home including power efficiency privacy and security. Listen to our full conversation on my audio podcast: #sponsored https://www.eetimes.com/podcasts/living-on-the-edge-intelligence-power-efficiency-and-security-take-smart-home-devices-to-the-next-level/ https://www.eetimes.com/podcasts/living-on-the-edge-intelligence-power-efficiency-and-security-take-smart-home-devices-to-the-next-level/"
[X Link](https://x.com/sallywf/status/1966551619098534135) 2025-09-12T17:16Z [----] followers, [---] engagements
"AI metasurface startup @neurophos has test silicon - showing the final AI inference accelerator will reach close to [---] TOPS/W (4-bit). CEO Patrick Bowen gave me an update on which software startup the company has acquired and its new data center partner: https://www.eetimes.com/neurophos-photonic-ai-test-chip-hits-300-tops-w/ https://www.eetimes.com/neurophos-photonic-ai-test-chip-hits-300-tops-w/"
[X Link](https://x.com/sallywf/status/1970817263944212480) 2025-09-24T11:47Z [----] followers, [---] engagements
"In the first of a series of reports from Stanford University's SPRC Symposium I spoke to @NttResearch's Timothy McKenna about TFLN (thin-film lithium niobate) and the material's potential for photonic computing chips: https://www.eetimes.com/can-tfln-make-photonic-compute-competitive/ https://www.eetimes.com/can-tfln-make-photonic-compute-competitive/"
[X Link](https://x.com/sallywf/status/1973299997144916047) 2025-10-01T08:12Z [----] followers, [---] engagements
"The unique capabilities of the Xcore mean XMOS is well-positioned to apply generative AI to its toolchain so the designer can specify at the behavioural level what the system needs to do and have generative AI implement it CEO Mark Lippett told me: https://www.eetimes.com/genai-can-raise-level-of-abstraction-for-xmos-tools/ https://www.eetimes.com/genai-can-raise-level-of-abstraction-for-xmos-tools/"
[X Link](https://x.com/sallywf/status/1975841776624877611) 2025-10-08T08:32Z [----] followers, [---] engagements
"World's worst-kept secret. https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration/ https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration/"
[X Link](https://x.com/sallywf/status/1977808124355793362) 2025-10-13T18:46Z [----] followers, [---] engagements
"I'm on @EETimesOnAir discussing London Tech Week and the battle between cities to become the hub of AI tech plus alternative AI accelerator hardware from CogX. #cogx19 #AI #LondonTechWeek #aitechnology #Graphcore #Mythic #Oxfordquantumcircuits #quantumcomputing #analogcomputing #AI hubs; #HDmaps for #robocars; #supercomputer bragging rights. @sallywf @rickbmerritt @Porsche44 @VSI_Labs @EETimesOnAir https://t.co/d8iRRCo8bU #AI hubs; #HDmaps for #robocars; #supercomputer bragging rights. @sallywf @rickbmerritt @Porsche44 @VSI_Labs @EETimesOnAir https://t.co/d8iRRCo8bU"
[X Link](https://x.com/sallywf/status/1143061835681320960) 2019-06-24T07:42Z [----] followers, [--] engagements
".@MLPerf training results are out There's a new GPT-3 benchmark and @NVIDIA + @CoreWeave trained it the fastest. Their only challenger: @Intel @HabanaLabs. #ai #aibenchmark #mlperf #gpt3 #aitraining #aihardware https://www.eetimes.com/nvidia-leads-habana-challenges-on-mlperf-gpt-3-benchmark/ https://www.eetimes.com/nvidia-leads-habana-challenges-on-mlperf-gpt-3-benchmark/"
[X Link](https://x.com/sallywf/status/1674137013103087631) 2023-06-28T19:25Z [----] followers, [----] engagements
"Hands on with @EsperantoTech silicon at #dac60 - [----] RISC-V cores on this 25W inference chip. They have Meta OPT 13B up and running on one chip. #ai #llm #aihardware #aiinference"
[X Link](https://x.com/anyuser/status/1678567311563796480) 2023-07-11T00:50Z [----] followers, 28.2K engagements
".@nvidia showed off Grace Hopper for the first time in the latest round of MLPerf inference benchmarks. @HabanaLabs showed competitive performance with Gaudi2 on LLMs and Google debuted the TPUv5e. Check the article for details. #mlperf #ai #llm https://www.eetimes.com/nvidia-shows-off-grace-hopper-in-mlperf-inference/ https://www.eetimes.com/nvidia-shows-off-grace-hopper-in-mlperf-inference/"
[X Link](https://x.com/sallywf/status/1701602583360807316) 2023-09-12T14:24Z [----] followers, [----] engagements
".@StabilityAI will buy [----] Gaudi2s plus Intel Xeon CPUs for an AI supercomputer that will be in the top [--] in the world (largest in Europe) said Gelsinger in his keynote this morning. Momentum on the AI continuum@IntelAI https://t.co/44eSDscRpb Momentum on the AI continuum@IntelAI https://t.co/44eSDscRpb"
[X Link](https://x.com/sallywf/status/1704226750555984138) 2023-09-19T20:11Z [----] followers, [----] engagements
"A few weeks ago I visited @Ambarella_Inc HQ - they showed me Llama2 up and running on their autonomous driving chip. Will they make an LLM accelerator Its probably something we need to consider Ambarella CEO Fermi Wang told me. Read more: https://www.eetimes.com/ambarella-demos-llm-inference-on-autonomous-driving-chip/ https://www.eetimes.com/ambarella-demos-llm-inference-on-autonomous-driving-chip/"
[X Link](https://x.com/sallywf/status/1724097819865231470) 2023-11-13T16:12Z [----] followers, [---] engagements
"Chinese automotive OEMs are able to move much quicker than their European counterparts and they've defined new classes of ADAS specifically for China said @Ambarella_Inc CEO. We also discussed the relationships between OEMs tier ones and chipmakers: https://www.eetimes.com/ambarella-ceo-chinese-oems-are-copying-the-tesla-model/ https://www.eetimes.com/ambarella-ceo-chinese-oems-are-copying-the-tesla-model/"
[X Link](https://x.com/sallywf/status/1726889323960615050) 2023-11-21T09:04Z [----] followers, [---] engagements
"I saw @Ambarella_Inc's latest multi-modal LLM demos including LLaVA-34B on the N1 Ambarella's most powerful SoC which comes with an #LLM-friendly software stack. CTO Les Kohn told me multi-modal LLMs are set to revolutionise autonomous driving. Article: https://www.eetimes.com/llms-are-ready-for-robotics-and-self-driving-says-ambarella/ https://www.eetimes.com/llms-are-ready-for-robotics-and-self-driving-says-ambarella/"
[X Link](https://x.com/sallywf/status/1813129511497576920) 2024-07-16T08:32Z [----] followers, [---] engagements
""The woman seems to be enjoying herself" - LLaVA-34B on Ambarella N1 gets it right first time Thanks @Ambarella_Inc for the demos 😊"
[X Link](https://x.com/sallywf/status/1813130577416962522) 2024-07-16T08:36Z [----] followers, [---] engagements
"Another edge AI chip startup exits @kinara_ai is being acquired by @NXP for $307m. I spoke to NXP to find out why and get some details on what their plans are for Kinara and its hardware and software roadmaps. Congrats to Ravi Wajahat Rehan and team https://www.eetimes.com/nxp-acquires-ai-chip-startup-kinara/ https://www.eetimes.com/nxp-acquires-ai-chip-startup-kinara/"
[X Link](https://x.com/sallywf/status/1889605567888379921) 2025-02-12T09:20Z [----] followers, [---] engagements
"I enjoyed moderating two engaging sessions on fast inference and agentic AI at The AI Summit (London Tech Week) with @SambaNovaAI CEO @RodrigoLiang Welcome back to London Rodrigo and team"
[X Link](https://x.com/sallywf/status/1933457655340650985) 2025-06-13T09:33Z [----] followers, [---] engagements
"A short while ago I had an interesting conversation with @Ambarella_Inc CTO Les Kohn about reasoning techniques at the edge - other techniques will likely become more useful for resource constrained-environments especially for real-time compute: https://www.eetimes.com/will-robots-use-reasoning-at-the-edge/ https://www.eetimes.com/will-robots-use-reasoning-at-the-edge/"
[X Link](https://x.com/sallywf/status/1956345062805078397) 2025-08-15T13:19Z [----] followers, [---] engagements
".@GoogleResearch developed a RISC-V superscalar CPU designed for edge AI which it is giving to the industry for free in a bid to avoid fragmentation in the edge AI ecosystem. @SynaCorp is the first to incorporate this NPU into production silicon: https://www.eetimes.com/google-open-sources-npu-ip-synaptics-implements-it/ https://www.eetimes.com/google-open-sources-npu-ip-synaptics-implements-it/"
[X Link](https://x.com/sallywf/status/1978460140396278226) 2025-10-15T13:57Z [----] followers, [----] engagements
".@NextSilicon has revealed its HPC chip architecture which reconfigures itself during runtime based on the workload. It can run any HPC code out of the box CEO Elad Raz told me. The company also showed off a test chip for a big RISC-V CPU it's building: https://www.eetimes.com/nextsilicon-details-runtime-reconfigurable-architecture https://www.eetimes.com/nextsilicon-details-runtime-reconfigurable-architecture"
[X Link](https://x.com/sallywf/status/1981252192997970330) 2025-10-23T06:51Z [----] followers, [----] engagements
"@FelixCLC_ @NextSilicon "flew under the radar" mate it was reported in EE Times"
[X Link](https://x.com/sallywf/status/1985381806468636929) 2025-11-03T16:21Z [----] followers, [---] engagements
"Chatted with Moreh CEO about enabling cost-effective alternatives to Nvidia. Moreh built a training software stack for AMD GPU clusters - something AMD itself is still working on - and is about to launch an inference stack for AMD and Tenstorrent servers: https://www.eetimes.com/cost-efficient-ai-at-scale-is-a-software-problem/ https://www.eetimes.com/cost-efficient-ai-at-scale-is-a-software-problem/"
[X Link](https://x.com/sallywf/status/1986780018467295721) 2025-11-07T12:57Z [----] followers, [----] engagements
"Eta Compute has rebranded as ModelCat and launched an AI agent that builds small edge models based on customers' datasets. The agent is hardware-aware and can test hypotheses on actual chips from a number of suppliers including @NXP: https://www.eetimes.com/eta-compute-rebrands-as-modelcat-unveils-agentic-model-builder/ https://www.eetimes.com/eta-compute-rebrands-as-modelcat-unveils-agentic-model-builder/"
[X Link](https://x.com/sallywf/status/1987810730121248861) 2025-11-10T09:13Z [----] followers, [---] engagements
"Japanese crypto-mining ASIC company Triple-1 has applied its deep sub-threshold technology (design target 0.26V) to TSMC 3nm. Key use case is edge data centers that convert surplus renewable energy into Bitcoin. Could this be applied to AI More: https://www.eetimes.com/crypto-mining-asic-goes-deep-sub-threshold-on-3-nm/ https://www.eetimes.com/crypto-mining-asic-goes-deep-sub-threshold-on-3-nm/"
[X Link](https://x.com/sallywf/status/1991080072996847854) 2025-11-19T09:44Z [----] followers, [---] engagements
"@GanguliArko 😂 I mean I really really want a house robot. I'm just against these videos that show performance in a misleading way. Much of this video is sped up 10x"
[X Link](https://x.com/sallywf/status/1991502242432847978) 2025-11-20T13:41Z [----] followers, [--] engagements
".@SynaCorp is open-sourcing the compiler toolchain for its AI hardware including the open-source NPU that comes from Google. Dave Garrett talked me through the reasons behind this decision. Hear our full conversation on my audio podcast: #sponsored #ad https://www.eetimes.com/podcasts/inside-synaptics-decision-to-open-source-its-ai-compiler/ https://www.eetimes.com/podcasts/inside-synaptics-decision-to-open-source-its-ai-compiler/"
[X Link](https://x.com/sallywf/status/1993695238578196701) 2025-11-26T14:56Z [----] followers, [---] engagements
"Q.ANT has unveiled a second-gen TFLN-based photonic compute chip. The German startup is banking on accelerating non-linear functions and KAN networks efficiently. I spoke to CEO Michael Frtsch for details on new funding customers and demo: https://www.eetimes.com/q-ant-raises-series-a-debuts-second-gen-tfln-photonic-chip/ https://www.eetimes.com/q-ant-raises-series-a-debuts-second-gen-tfln-photonic-chip/"
[X Link](https://x.com/sallywf/status/1993698050380902902) 2025-11-26T15:07Z [----] followers, [---] engagements
"Lemurian Labs formerly an AI chip company has raised $28M to develop its AI software stack spanning the compiler layer all the way up to workload orchestration while enabling code portability. The CEO said he wants to do away with kernels altogether: https://www.eetimes.com/lemurian-labs-raises-28-million-for-ai-portability-software/ https://www.eetimes.com/lemurian-labs-raises-28-million-for-ai-portability-software/"
[X Link](https://x.com/sallywf/status/1996661283647901821) 2025-12-04T19:22Z [----] followers, [---] engagements
"The biggest news story in my beat last year dropped at 10pm on Christmas Eve (thank you Santa.). Nvidia has all-but acquired Groq for reportedly $20B. I was surprised to say the least. I spent the last two weeks reading bad take after bad take (1/x)"
[X Link](https://x.com/sallywf/status/2009207702249972127) 2026-01-08T10:17Z [----] followers, [----] engagements
"I have been writing about Groq for years. Much of what is on social media is ill-informed mediocre AI slop or just plain wrong. The truth is no-one outside the companies has any real idea why Nvidia all-but acquired Groq and what we know so far doesnt make much sense yet🧵(1/x)"
[X Link](https://x.com/sallywf/status/2009308696191385667) 2026-01-08T16:58Z [----] followers, 93.5K engagements
"Nvidia integrating Groq chips/systems directly into its main data center tech roadmap seems far-fetched partly because it would mean admitting GPUs cant do everything and partly because software would be a nightmare because (5/x)"
[X Link](https://x.com/sallywf/status/2009308933379219596) 2026-01-08T16:59Z [----] followers, [----] engagements
"The difference is not as simple as Groq only uses SRAM. Groq is so so different from a GPU. Its compiler is a huge part of its secret sauce. It would not fit into the CUDA ecosystem. eg there is no equivalent of a kernel in Groq arch (6/x)"
[X Link](https://x.com/sallywf/status/2009308982741958847) 2026-01-08T16:59Z [----] followers, [----] engagements
"Nvidias biggest fear is hyperscalers own chips I would speculate that Nvidia didnt want to risk one of its big hyperscale customers acquiring Groq (10/x)"
[X Link](https://x.com/sallywf/status/2009309154687549882) 2026-01-08T17:00Z [----] followers, [----] engagements
"Google TPU is great but eg Meta and Microsoft internal chip efforts are still relatively immature. Groqs hardware software and infrastructure might have made a difference there (11/x)"
[X Link](https://x.com/sallywf/status/2009309194147479967) 2026-01-08T17:00Z [----] followers, [----] engagements
"This is obviously speculation and ultimately there are still many more questions than answers regarding this deal. Here is a longer-form version of my analysis on Groq and Nvidia (/end): https://www.eetimes.com/what-is-groq-nvidia-deal-really-about/ https://www.eetimes.com/what-is-groq-nvidia-deal-really-about/"
[X Link](https://x.com/sallywf/status/2009309254004400179) 2026-01-08T17:00Z [----] followers, [----] engagements
".@Ambarella_Inc has changed its go-to-market strategy as it embraces the edge AI ecosystem particularly ISVs (the company previously supported big customers almost exclusively with no resellers and no channel). Also details on the new CV7 silicon: https://www.eetimes.com/ambarella-shifts-market-strategy-to-embrace-edge-ecosystem/ https://www.eetimes.com/ambarella-shifts-market-strategy-to-embrace-edge-ecosystem/"
[X Link](https://x.com/sallywf/status/2009356546392903761) 2026-01-08T20:08Z [----] followers, [----] engagements
"@ourtown2 Thanks. As I study AI chip companies for my full time job I never would have known this"
[X Link](https://x.com/sallywf/status/2009568493214269878) 2026-01-09T10:10Z [----] followers, [--] engagements
"@cdsmithus Yes but did it work then I am not aware of any customer who bought Groq chips back then and ran production workloads (maybe all secret). Pivot to TaaS for business reasons yes ("selling chips is too hard") but no doubt making whole stack "for internal consumption only" helped"
[X Link](https://x.com/sallywf/status/2009606175890059552) 2026-01-09T12:40Z [----] followers, [--] engagements
"@cdsmithus That's great. I hope you're right that Nvidia saw value in Groq's tech. Personally would love to see what some sort of Nvida-Groq combo would look like and I hope we can get more details at GTC"
[X Link](https://x.com/sallywf/status/2009640401993711852) 2026-01-09T14:56Z [----] followers, [--] engagements
"Troubled analog compute chip startup Mythic has pulled off a comeback with a fresh $125-million funding round. The company has fixed issues with its 1st gen chip and plans a 2nd gen data center class product and more according to its new CEO. Article: https://www.eetimes.com/mythic-rises-from-the-ashes-with-125-million-funding-round/ https://www.eetimes.com/mythic-rises-from-the-ashes-with-125-million-funding-round/"
[X Link](https://x.com/sallywf/status/2012151080348786951) 2026-01-16T13:12Z [----] followers, [----] engagements
"Delighted to get the opportunity to ask @SiFive's John Simpson all about vector processor design for AI about the new RISC-V vector profiles and about how SiFive works with these profiles even before they are set in stone. Listen in to our deep-dive: #ad https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/ https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/"
[X Link](https://x.com/sallywf/status/2013290597545553945) 2026-01-19T16:40Z [----] followers, [--] engagements
"@FelixCLC_ Let's just say you won't need the goose meme😉"
[X Link](https://x.com/sallywf/status/2013297132611023155) 2026-01-19T17:06Z [----] followers, [---] engagements
"Quadric has raised another $30 million and has at least five licensees for its programmable NPU IP. I spoke to CMO Steve Roddy who explained a little about the NPU's architecture and how @quadric_io sees its competition in the space: https://www.eetimes.com/quadric-raises-series-c-for-programmable-npu-ip/ https://www.eetimes.com/quadric-raises-series-c-for-programmable-npu-ip/"
[X Link](https://x.com/sallywf/status/2013558316547375594) 2026-01-20T10:24Z [----] followers, [----] engagements
"I had a chat with @dMatrix_AI CEO Sid Sheth a little while back about test time scaling what it is and what it means for hardware - podcast: (I also got a first-hand look at D-Matrix's 3D-stacked logic and DRAM test chip - see article link in next tweet) https://www.eetimes.com/podcasts/d-matrix-ceo-sid-sheth-on-ai-test-time-scaling/ https://www.eetimes.com/podcasts/d-matrix-ceo-sid-sheth-on-ai-test-time-scaling/"
[X Link](https://x.com/sallywf/status/2014773548468883608) 2026-01-23T18:53Z [----] followers, [----] engagements
"@dMatrix_AI Here's the article with updates on D-Matrix's roadmap and some pics https://www.eetimes.com/as-demand-for-fast-ai-tokens-grows-d-matrix-develops-fast-nic/ https://www.eetimes.com/as-demand-for-fast-ai-tokens-grows-d-matrix-develops-fast-nic/"
[X Link](https://x.com/sallywf/status/2014773707152011558) 2026-01-23T18:54Z [----] followers, [---] engagements
"@proxietuna I think Groq value at last funding was 7B - cf AMD acq of Xilinx at 49B ADI acq Maxim for 20B in [----]. and SambaNova at 1B. Groq must have a revenue figure (unlike some others) they can use to calculate value. Yes Mellanox cheap but that was pre-AI - all has since gone nuts"
[X Link](https://x.com/sallywf/status/2009271062459036103) 2026-01-08T14:28Z [----] followers, [---] engagements
"Delighted to get the opportunity to ask @SiFive's John Simpson all about vector processor design for AI about the new RISC-V matrix profiles and about how SiFive works with these profiles even before they are set in stone. Listen in to our deep-dive: #ad https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/ https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/"
[X Link](https://x.com/sallywf/status/2013293550583062889) 2026-01-19T16:52Z [----] followers, [----] engagements
".@AyarLabs and Alchip unveiled a reference design for big AI chips with optical I/O based on Ayar optical engines and TSMC COUPE tech. Ayar CTO Vladimir Stojanovic talked me through the significance of this reference platform: https://www.eetimes.com/ayar-labs-and-alchip-unveil-optical-i-o-reference-design/ https://www.eetimes.com/ayar-labs-and-alchip-unveil-optical-i-o-reference-design/"
[X Link](https://x.com/sallywf/status/1971142307861942325) 2025-09-25T09:18Z [----] followers, [---] engagements
"RT @Pranavenstein: Intel has been making GPUs for years. When you dont know what youre writing about this is the kind of headline you ge"
[X Link](https://x.com/sallywf/status/2018955030435680352) 2026-02-04T07:49Z [----] followers, [--] engagements
"My hot take: @intel will never find another Gelsinger. The 30-year Intel veteran former Intel CTO and former VMWare CEO is irreplaceable. https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/ https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/"
[X Link](https://x.com/anyuser/status/1864243323910730054) 2024-12-04T09:40Z [----] followers, 54K engagements
"Happy Tuesday everyone"
[X Link](https://x.com/anyuser/status/1820756452618756337) 2024-08-06T09:39Z [----] followers, 41.6K engagements
"Edge AI chip companies be like:"
[X Link](https://x.com/anyuser/status/1730251076480172267) 2023-11-30T15:42Z [----] followers, 28.5K engagements
"Geeking out at @nvidia with the new GH200 🤓"
[X Link](https://x.com/anyuser/status/1703916659575284114) 2023-09-18T23:39Z [----] followers, 20.7K engagements
"Hands on with @EsperantoTech silicon at #dac60 - [----] RISC-V cores on this 25W inference chip. They have Meta OPT 13B up and running on one chip. #ai #llm #aihardware #aiinference"
[X Link](https://x.com/anyuser/status/1678567311563796480) 2023-07-11T00:50Z [----] followers, 28.2K engagements
"@IanCutress .on [----] sqmm chips 😂"
[X Link](https://x.com/anyuser/status/1864992143812469235) 2024-12-06T11:15Z [----] followers, [----] engagements
"Congrats @AMD and @UntetherAI on debut @MLPerf results AMD MI300X matches H100's inference performance while Untether can beat H200's power efficiency by 3x. Featuring also: @nvidia Blackwell @intel Granite Rapids and @Google Trillium. My analysis: https://www.eetimes.com/amd-and-untether-take-on-nvidia-in-mlperf-benchmarks/ https://www.eetimes.com/amd-and-untether-take-on-nvidia-in-mlperf-benchmarks/"
[X Link](https://x.com/anyuser/status/1829068040798904373) 2024-08-29T08:06Z [----] followers, 13.5K engagements
"The Japanese government will send engineers to be trained in RISC-V and AI at @tenstorrent over the next [--] years. Tenstorrent gets free engineering talent revenue from the Japanese government and hopefully good relationships in the engineers' companies: https://www.eetimes.com/japan-sends-engineers-to-be-trained-at-tenstorrent/ https://www.eetimes.com/japan-sends-engineers-to-be-trained-at-tenstorrent/"
[X Link](https://x.com/anyuser/status/1853791550699901074) 2024-11-05T13:28Z [----] followers, 11.5K engagements
"OK I'll bite. Yes making a transformer chip does mean making a large matmul engine. But there is a lot more to efficient transformer inference than GEMM/FLOPS. You need to be able to feed and use those FLOPS efficiently. (1/x) Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well https://t.co/Ju8QX9gcc7 Although "baking transformers in silicon" may sound cool it is mostly"
[X Link](https://x.com/anyuser/status/1948374682027094171) 2025-07-24T13:28Z [----] followers, 27.4K engagements
"Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well for GEMMs. Even if you "bake transformers in silicon" aka just create an gaint systolic array most of your die area will still be taken up with SRAM cells and you will still face the memory wall since your HBM memory bandwidth will be the same as GPGPU/TPUs/Trainium"
[X Link](https://x.com/anyuser/status/1948206985306095796) 2025-07-24T02:21Z 60.1K followers, 51.1K engagements
".@BrainChip_inc is now offering its #neuromorphic #ai chips in an edge box suitable for security camera systems. Why are edge boxes becoming an increasingly popular path to market for edge AI chip makers And what is BrainChip's strategy for edge boxes https://www.eetimes.com/brainchip-adds-edge-box-to-chip-and-ip-offerings/ https://www.eetimes.com/brainchip-adds-edge-box-to-chip-and-ip-offerings/"
[X Link](https://x.com/anyuser/status/1765116749223977399) 2024-03-05T20:46Z [----] followers, [----] engagements
"Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: https://t.co/S7pQFnOVJV "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: https://t.co/S7pQFnOVJV "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X"
[X Link](https://x.com/anyuser/status/1787883356253773943) 2024-05-07T16:33Z [----] followers, [----] engagements
"Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X Elite and X Plus have [--] TOPS. Now its clear why this launch feels rushed for M4. https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/ https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/"
[X Link](https://x.com/anyuser/status/1787880145568829475) 2024-05-07T16:20Z 18.7K followers, 25.4K engagements
"Called by @tenstorrent to interview Jim Keller and get a look at new Wormhole cards Thanks for having me"
[X Link](https://x.com/anyuser/status/1659312624457371649) 2023-05-18T21:38Z [----] followers, [----] engagements
"Carmakers are testing neuromorphic technologies based on spiking neural networks. What will cars do with their newly found brain power Featuring hardware from @SynSenseNeuro @BrainChip_inc. #snn #spiking #spikingneuralnetwork #automotive #neuromorphic https://www.eetimes.com/cars-that-think-like-you/ https://www.eetimes.com/cars-that-think-like-you/"
[X Link](https://x.com/anyuser/status/1551485047252914181) 2022-07-25T08:30Z [----] followers, [---] engagements
"Neuromorphic IP vendor @BrainChip_inc now has dev kits available. The company converts CNNs to spiking networks so they can run on BrainChip's Akida processor which reduces power consumption and enables on-chip learning at the edge. #ai #neuromorphic https://www.eetimes.com/brainchip-launches-event-domain-ai-inference-dev-kits/ https://www.eetimes.com/brainchip-launches-event-domain-ai-inference-dev-kits/"
[X Link](https://x.com/anyuser/status/1451480894372450322) 2021-10-22T09:29Z [----] followers, [---] engagements
".@tenstorrent has partnered with fledgling Japanese foundry Rapidus. Why I sat down with Jim Keller for insight into the automotive AI chip market a progress report on hardware and software and its IP business which is steadily gaining momentum. https://www.eetimes.com/tenstorrent-gains-momentum-in-asia-with-rapidus-partnership/ https://www.eetimes.com/tenstorrent-gains-momentum-in-asia-with-rapidus-partnership/"
[X Link](https://x.com/anyuser/status/1729067950814978214) 2023-11-27T09:21Z [----] followers, 13K engagements
"New favourite GIF just dropped @IanCutress With subs and easy karaoke https://t.co/KyJVIR7zsR @IanCutress With subs and easy karaoke https://t.co/KyJVIR7zsR"
[X Link](https://x.com/anyuser/status/1693905807010566493) 2023-08-22T08:39Z [----] followers, [----] engagements
"@IanCutress With subs and easy karaoke"
[X Link](https://x.com/anyuser/status/1693803468346081403) 2023-08-22T01:53Z [---] followers, [----] engagements
"My hot take: A sigh of relief as Lip-Bu Tan is appointed CEO of @intel https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/ https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/"
[X Link](https://x.com/anyuser/status/1899991868198576281) 2025-03-13T01:12Z [----] followers, [----] engagements
"HAHAHAHAHAHAHAHAHAHAHAHAHAHA. deep breath hahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahah watching @potus more carefully recently has really changed my perspective on him (i wish i had done more of my own thinking and definitely fell in the npc trap). i'mnot going to agree with him on everything but i think he will be incredible for the country in many ways watching @potus more carefully recently has really"
[X Link](https://x.com/anyuser/status/1882367393734983799) 2025-01-23T09:58Z [----] followers, [----] engagements
"watching @potus more carefully recently has really changed my perspective on him (i wish i had done more of my own thinking and definitely fell in the npc trap). i'mnot going to agree with him on everything but i think he will be incredible for the country in many ways"
[X Link](https://x.com/anyuser/status/1882234406662000833) 2025-01-23T01:10Z 4.3M followers, 12.6M engagements
"Cheers @nvidia on a successful and enjoyable GTC #gtc24"
[X Link](https://x.com/anyuser/status/1770582346070110281) 2024-03-20T22:45Z [----] followers, [----] engagements
".@RajaXg's new startup Mihira will build data centers with heterogeneous compute for rendering and AI. Mihira has licensed software from Intel's recently abandoned Project Endgame. And Raja told me he hasn't ruled out chip design going forward. https://www.eetimes.com/koduris-startup-mihira-to-license-intel-endgame-software/ https://www.eetimes.com/koduris-startup-mihira-to-license-intel-endgame-software/"
[X Link](https://x.com/anyuser/status/1714630697871188091) 2023-10-18T13:13Z [----] followers, [----] engagements
"Stealthy Korean startup @FuriosaAI will present its tensor contraction processor at Hot Chips later today. I got the scoop Please enjoy this delightful interview with CEO June Paik: https://www.eetimes.com/furiosa-targets-llm-inference-with-second-gen-chip/ https://www.eetimes.com/furiosa-targets-llm-inference-with-second-gen-chip/"
[X Link](https://x.com/anyuser/status/1828140585192165815) 2024-08-26T18:40Z [----] followers, [----] engagements
"I reached [----] twitter followers 🎉🎉🎉 Thanks everyone for following along and please stay tuned for more interesting stuff on AI chips as we head into [----]. I'd like to thank my mother my colleagues everyone I've ever interviewed . (🧵1/387)"
[X Link](https://x.com/anyuser/status/1729806359640617181) 2023-11-29T10:15Z [----] followers, [----] engagements
".@nvidia has trained its NeMo #LLM on its internal documents and codebase. ChipNeMo can write Tcl scripts to control EDA tools summarize bug documentation & answer junior designers' questions. All geared towards designer productivity Bill Dally told me: https://www.eetimes.com/nvidia-trains-llm-on-chip-design/ https://www.eetimes.com/nvidia-trains-llm-on-chip-design/"
[X Link](https://x.com/anyuser/status/1719026669753618666) 2023-10-30T16:21Z [----] followers, [----] engagements
".@Arm has updated its CSS designs for hyperscalers' custom chips. Arms vision is for a significant percentage if not the vast majority of AI inference to run on CPUs. Details on two new Neoverse cores (and preliminary Llama2-7B results) in the article: https://www.eetimes.com/arm-updates-css-designs-for-hyperscalers-custom-chips/ https://www.eetimes.com/arm-updates-css-designs-for-hyperscalers-custom-chips/"
[X Link](https://x.com/anyuser/status/1760342006922870788) 2024-02-21T16:33Z [----] followers, [----] engagements
".@tenstorrent has raised a whopping $693 million in a round that will put any doubts about the investment community's appetite for AI chip companies aside. The company has $150 million in deals committed mainly IP with next gen chip coming in early 2025: https://www.eetimes.com/tenstorrent-raises-693-million-series-d/ https://www.eetimes.com/tenstorrent-raises-693-million-series-d/"
[X Link](https://x.com/anyuser/status/1863629989087723784) 2024-12-02T17:03Z [----] followers, [----] engagements
"This customer win is a huge deal for @cerebras and will be worth around $900 million in the next [--] months. Importantly it validates the market for non-GPU AI accelerators at data center scale. https://www.eetimes.com/cerebras-sells-100-million-ai-supercomputer-plans-8-more/ https://www.eetimes.com/cerebras-sells-100-million-ai-supercomputer-plans-8-more/"
[X Link](https://x.com/anyuser/status/1682025863548899331) 2023-07-20T13:53Z [----] followers, [----] engagements
".@GroqInc CEO @JonathanRoss321 is adamant his company no longer sells hardware-they are a cloud services provider and have a deal with Aramco. "The realization was you cannot sell chips as a startup its just too hard Ross told me. More in the article: https://www.eetimes.com/groq-ceo-we-no-longer-sell-hardware/ https://www.eetimes.com/groq-ceo-we-no-longer-sell-hardware/"
[X Link](https://x.com/anyuser/status/1777333385649586388) 2024-04-08T13:51Z [----] followers, 12.3K engagements
"I spent a couple of hours at @tenstorrent chatting with @jimkxa to get updates on the company's many current projects including a new AI workstation they have developed based on first-generation Wormhole chips (a next-gen Blackhole version is coming): https://www.eetimes.com/tenstorrent-to-offer-ai-workstation-for-developers/ https://www.eetimes.com/tenstorrent-to-offer-ai-workstation-for-developers/"
[X Link](https://x.com/anyuser/status/1805635553913684145) 2024-06-25T16:13Z [----] followers, [----] engagements
"Delighted to be invited to @Arm's HQ in Cambridge today. Story coming soon to @eetimes"
[X Link](https://x.com/anyuser/status/1628777798297960450) 2023-02-23T15:24Z [----] followers, [----] engagements
"Called by @tenstorrent this morning to chat with Jim Keller about the company's plans RISC-V in automotive and scientific computing the economics of chiplets and much more. Stay tuned for the @eetimes article. Thanks @jimkxa"
[X Link](https://x.com/anyuser/status/1701367541351014832) 2023-09-11T22:50Z [----] followers, [----] engagements
"Already sick of hearing people describe Arm as a "chipmaker" (in this case "shipping chips") and there are still several weeks to go on this IPO. Arm does not ship chips Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile - 96% gross margins (26% operating margins) - Aiming for 65B market cap For the full history of the company just watch this video: https://t.co/8R15GFf8Nl Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile -"
[X Link](https://x.com/anyuser/status/1694342919736447209) 2023-08-23T13:36Z [----] followers, [----] engagements
"Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile - 96% gross margins (26% operating margins) - Aiming for 65B market cap For the full history of the company just watch this video:"
[X Link](https://x.com/anyuser/status/1694019604040007811) 2023-08-22T16:12Z 68.8K followers, 430.8K engagements
"Always feel a little outclassed in a press conference's sea of laptops. I'm here with my pen that writes in multiple colours #GTC24"
[X Link](https://x.com/anyuser/status/1769757702282948898) 2024-03-18T16:08Z [----] followers, [----] engagements
"Presenting The AI Hardware Show Live at #dac60 today with @IanCutress - thanks to everyone who stopped by to watch us"
[X Link](https://x.com/anyuser/status/1679279272513662977) 2023-07-12T23:59Z [----] followers, [----] engagements
"COMING SOON"
[X Link](https://x.com/anyuser/status/1952822492696813840) 2025-08-05T20:02Z [----] followers, [----] engagements
".@tenstorrent has launched (and open-sourced) its bare-metal software stack. This is a crucial capability for deployment of #AI models into production at scale as well as for HPC applications. I also got a demo of Falcon-40B up and running on Galaxy https://www.eetimes.com/tenstorrent-engineers-talk-open-sourced-bare-metal-stack/ https://www.eetimes.com/tenstorrent-engineers-talk-open-sourced-bare-metal-stack/"
[X Link](https://x.com/anyuser/status/1753423821980274954) 2024-02-02T14:23Z [----] followers, [----] engagements
"I sat down with @nvidia's Ian Buck to go through some details on Blackwell including what's new in the 2nd-gen transformer engine and how Nvidia gets 30X the performance from GPUs with 2.5X the flops of Hopper. More details in the article: #ai #blackwell https://www.eetimes.com/nvidias-blackwell-gpu-offers-fp4-transformer-engine-sharp/ https://www.eetimes.com/nvidias-blackwell-gpu-offers-fp4-transformer-engine-sharp/"
[X Link](https://x.com/anyuser/status/1773004698313687273) 2024-03-27T15:10Z [----] followers, [----] engagements
"Wondering if beginning more of my tweets with "IT'S ABOUT TO START THE BIGGEST TECH BATTLE OF OUR TIME" (and then getting the product name wrong) would get me [---] million views per tweet Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time: https://t.co/KAUpGpDcKJ Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI"
[X Link](https://x.com/anyuser/status/1869700365299728505) 2024-12-19T11:04Z [----] followers, [----] engagements
"Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time:"
[X Link](https://x.com/anyuser/status/1869427504303481072) 2024-12-18T17:00Z 44.2K followers, 4M engagements
".@SiFive and @BrainChip_inc have partnered to show SiFive's RISC-V host processor IP working alongside BrainChip's neuromorphic AI accelerator for edge AI SoCs. #ai #aiaccelerator #neuromorphic #ip #riscv https://www.eetimes.com/sifive-and-brainchip-partner-to-demo-ip-compatibility/ https://www.eetimes.com/sifive-and-brainchip-partner-to-demo-ip-compatibility/"
[X Link](https://x.com/anyuser/status/1517067284854063105) 2022-04-21T09:06Z [----] followers, [--] engagements
".@Modular has unveiled its #AI stack w/@nvidia GPU support. The startup is tackling AI infrastructure's hard problems by building a fancy compiler and a new programming language. Is it hard Yes. But is it worth it Also yes @clattner_llvm told me: https://www.eetimes.com/modular-launches-ai-stack-adds-gpu-support/ https://www.eetimes.com/modular-launches-ai-stack-adds-gpu-support/"
[X Link](https://x.com/anyuser/status/1736768775863632366) 2023-12-18T15:21Z [----] followers, [----] engagements
"Raja Koduri @RajaXg has launched his new GPU company @realoxmiqlabs will take on the whole of the GPU technology stack from data center orchestration to hardware IP. Coming first A Python module that enables torch.cuda code to run on other hardware: https://www.eetimes.com/koduri-unveils-gpu-hardware-ip-and-software-startup/ https://www.eetimes.com/koduri-unveils-gpu-hardware-ip-and-software-startup/"
[X Link](https://x.com/anyuser/status/1952655660501368936) 2025-08-05T08:59Z [----] followers, [----] engagements
"Sad news from @UntetherAI as the RISC-V AI chip company closes its doors. Untether's engineering teams will join @AMD but the company's silicon and software are no more: https://www.eetimes.com/untether-ai-shuts-down-engineering-team-joins-amd/ https://www.eetimes.com/untether-ai-shuts-down-engineering-team-joins-amd/"
[X Link](https://x.com/anyuser/status/1932045970923020295) 2025-06-09T12:03Z [----] followers, [----] engagements
"Had an interesting conversation with @nayampally_n recently about @BrainChip_inc's second-gen architecture which has hardware support for transformers and the concept of TENNs. #AI Available at or wherever you get your podcasts: https://www.eetimes.com/podcasts/brainchip-second-gen-architecture-transformers-tenns-and-stdp/ http://eetimes.com https://www.eetimes.com/podcasts/brainchip-second-gen-architecture-transformers-tenns-and-stdp/ http://eetimes.com"
[X Link](https://x.com/anyuser/status/1735300453573796036) 2023-12-14T14:07Z [----] followers, [----] engagements
"I have a few thoughts on this article.🧵 Nvidias rivals struggle to gain ground in generative AI chip war https://t.co/o0afqQ4kqO Nvidias rivals struggle to gain ground in generative AI chip war https://t.co/o0afqQ4kqO"
[X Link](https://x.com/anyuser/status/1669390178891202587) 2023-06-15T17:03Z [----] followers, 22.6K engagements
"Nvidias rivals struggle to gain ground in generative AI chip war https://on.ft.com/43YdiBw https://on.ft.com/43YdiBw"
[X Link](https://x.com/anyuser/status/1669284177152290816) 2023-06-15T10:02Z 289.6K followers, 18.3K engagements
"Not thrilled so far #CogX19 #CogX stood in a queue of hundreds in pouring rain to pick up unnecessary lanyard. Why can't barcode alone be good for entry Missing Sadiq Khan speech for this"
[X Link](https://x.com/anyuser/status/1137992577200906241) 2019-06-10T07:59Z [----] followers, [--] engagements
"@spiralout112 @intel Anyone worthy of the job would stick to Pat's vision - but because of this they would never get the job. They'll have a bean counter in the role in [--] months"
[X Link](https://x.com/anyuser/status/1864250796730183824) 2024-12-04T10:10Z [----] followers, [----] engagements
"It's certainly been a big week for @risc_v. Read my hot take on how @intel's investment and Arm's IPO will affect the open-source ISA. #riscv #arm #ipo #isa https://www.eetimes.com/a-big-week-for-risc-v/ https://www.eetimes.com/a-big-week-for-risc-v/"
[X Link](https://x.com/anyuser/status/1492084880288714763) 2022-02-11T10:35Z [----] followers, [--] engagements
".@AMD has multiple AI software stacks for its diverse hardware. Vamsi Boppana told me there is work ongoing to unify some elements of these stacks and that ROCm is the company's #1 priority since it is crucial to the success of MI300 & MI300X. #ai #rocm https://www.eetimes.com/rocm-is-amds-no-1-priority-exec-says/ https://www.eetimes.com/rocm-is-amds-no-1-priority-exec-says/"
[X Link](https://x.com/anyuser/status/1706309212177858974) 2023-09-25T14:06Z [----] followers, [----] engagements
".@imec_int processor-in-memory test chip can achieve [----] TOPS/W for AI inference - yes you read that right. This tech will be available on @GLOBALFOUNDRIES 22FDX process soon. #fdsoi #tinyml #ai #MachineLearning https://www.eetimes.com/imec-develops-efficient-processor-in-memory-technique-for-glofo/ https://www.eetimes.com/imec-develops-efficient-processor-in-memory-technique-for-glofo/"
[X Link](https://x.com/anyuser/status/1281498995760730116) 2020-07-10T08:02Z [----] followers, [--] engagements
"Good morning to everyone EXCEPT the person who commented on YouTube that they thought I was @IanCutress in a wig"
[X Link](https://x.com/anyuser/status/1646896869300158470) 2023-04-14T15:23Z [----] followers, [----] engagements
"Happy Monday everyone"
[X Link](https://x.com/anyuser/status/1779799165934903500) 2024-04-15T09:09Z [----] followers, [----] engagements
"Eliyan's chiplet PHY can achieve advanced packaging performance from standard packaging. Omitting the silicon interposer opens up a new world of design possibilities for chiplets especially for HBM. I spoke to @RajaXg about some of these possibilities: https://www.eetimes.com/eliyan-breaks-chiplet-memory-wall-with-standard-packaging/ https://www.eetimes.com/eliyan-breaks-chiplet-memory-wall-with-standard-packaging/"
[X Link](https://x.com/anyuser/status/1754532207467667840) 2024-02-05T15:47Z [----] followers, [----] engagements
"Mercedes used @BrainChip_inc's neuromorphic chip in its latest concept car for the "Hey Mercedes" hot word. Mercedes said #neuromorphic chips use 5-10X less power than conventional equivalent which contributes to the car's 1000-km range. https://www.eetimes.com/mercedes-applies-neuromorphic-computing-in-ev-concept-car/ https://www.eetimes.com/mercedes-applies-neuromorphic-computing-in-ev-concept-car/"
[X Link](https://x.com/anyuser/status/1483730204665303044) 2022-01-19T09:17Z [----] followers, [--] engagements
"Google published a paper last week on TPUv4. You may have seen in press/on twitter Google "beating Nvidia" or making unfair comparisons. Not only did they not beat Nvidia G's comparison was appropriate and paper actually had interesting stuff besides: https://www.eetimes.com/stories-claiming-google-tpuv4-beat-nvidia-greatly-overstated/ https://www.eetimes.com/stories-claiming-google-tpuv4-beat-nvidia-greatly-overstated/"
[X Link](https://x.com/anyuser/status/1645761371122835456) 2023-04-11T12:11Z [----] followers, [----] engagements
"Is rumoured Graphcore acquisition about to close @ndahad has the scoop and I have provided some analysis in our first joint story. It was fun to speculate on what Graphcore's rumoured acquirer might do with the British startup and its team: https://www.eetimes.com/is-graphcore-deal-finally-about-to-close/ https://www.eetimes.com/is-graphcore-deal-finally-about-to-close/"
[X Link](https://x.com/anyuser/status/1788847390377607175) 2024-05-10T08:23Z [----] followers, [----] engagements
".@SambaNovaAI released new silicon and yours truly had my hands on it on Friday New chip is designed for LLM inference and adds HBM for the first time (previously they got away with DDR only). Details in my story: #ai #llm #llminference #chatgpt #aichip https://www.eetimes.com/sambanova-adds-hbm-for-llm-inference-chip/ https://www.eetimes.com/sambanova-adds-hbm-for-llm-inference-chip/"
[X Link](https://x.com/anyuser/status/1704135427429736955) 2023-09-19T14:08Z [----] followers, [----] engagements
"I called in at @tenstorrent to see the team's Llama3.1-70B demo up and running on one of their 8-chip systems at [--] tokens/second/user a figure the company considers a work-in-progress as it continues to optimise its software. More in the article: https://www.eetimes.com/tenstorrent-shows-off-single-user-llm-speed-for-workstation/ https://www.eetimes.com/tenstorrent-shows-off-single-user-llm-speed-for-workstation/"
[X Link](https://x.com/anyuser/status/1841101082136600839) 2024-10-01T13:01Z [----] followers, [----] engagements
"@IanCutress They should be reading @eetimes"
[X Link](https://x.com/anyuser/status/1870089951251189964) 2024-12-20T12:52Z [----] followers, [----] engagements
"There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: https://t.co/HOKaYA8dm5 There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: https://t.co/HOKaYA8dm5"
[X Link](https://x.com/anyuser/status/1745488497270468986) 2024-01-11T16:51Z [----] followers, [----] engagements
"There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: Get the full story on our LPU Inference Engine's performance in our latest blog and thanks to the work of our incredible team and the team over at Anyscale http://groq.link/anyscaleblog http://groq.link/anyscaleblog"
[X Link](https://x.com/anyuser/status/1745478422464500044) 2024-01-11T16:10Z 89K followers, [----] engagements
".@latticesemi new general-purpose FPGA comes in 17k-40k logic cells a market largely ignored by the big players. #fpga #programmablelogic #latticesemi #generalpurposefpga https://www.eetimes.com/lattice-reinvents-general-purpose-fpga-offering/ https://www.eetimes.com/lattice-reinvents-general-purpose-fpga-offering/"
[X Link](https://x.com/anyuser/status/1276081439419834368) 2020-06-25T09:14Z [----] followers, [--] engagements
"SOME PEOPLE THINK "IN-MEMORY COMPUTE" AND "COMPUTE-IN-MEMORY" IS THE SAME THING BUT OFTEN WHEN PEOPLE SAY "IN-MEMORY COMPUTE" THEY JUST MEAN COMPUTE AND MEMORY MIXED TOGETHER AT A FINE LEVEL OF GRANULARITY AS OPPOSED TO ANALOG COMPUTING WHICH IS TOTALLY DIFF-"
[X Link](https://x.com/anyuser/status/1560559062579720192) 2022-08-19T09:27Z [----] followers, [--] engagements
"HUGE if true and its a big IF😂 on the contrary this should be a MASSIVE surprise to anyone who knows anything about Arm 🧵 This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center. This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center"
[X Link](https://x.com/anyuser/status/1790006954506510458) 2024-05-13T13:11Z [----] followers, [----] engagements
"This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center"
[X Link](https://x.com/anyuser/status/1789830414850281863) 2024-05-13T01:29Z 18.7K followers, 11K engagements
"Filming in a secret location today with @IanCutress"
[X Link](https://x.com/anyuser/status/1818239381796254057) 2024-07-30T10:57Z [----] followers, [----] engagements
"New @intel CEO Lip-Bu Tan appeared both humble and determined in a keynote at Intel Vision this week. It seems he will take pages from his Cadence playbook for Intel including refocusing on innovation and driving a more customer-centric culture: https://www.eetimes.com/intel-ceo-you-deserve-better-we-need-to-improve-and-we-will/ https://www.eetimes.com/intel-ceo-you-deserve-better-we-need-to-improve-and-we-will/"
[X Link](https://x.com/anyuser/status/1907815663210500342) 2025-04-03T15:21Z [----] followers, [----] engagements
".@cerebras has filed the S-1 for its IPO. I read the 188-page document looking for interesting tidbits and here's what I found: https://www.eetimes.com/cerebras-ipo-paperwork-sheds-light-on-relationship-with-g42/ https://www.eetimes.com/cerebras-ipo-paperwork-sheds-light-on-relationship-with-g42/"
[X Link](https://x.com/anyuser/status/1841227318871667005) 2024-10-01T21:22Z [----] followers, [----] engagements
"It was an honour to be asked to present Opening Remarks at @eetimes Taiwan's AI+Techcon conference. Thank you for inviting me"
[X Link](https://x.com/anyuser/status/1783418474980642906) 2024-04-25T08:51Z [----] followers, [----] engagements
"Here comes Hot Chips Are you looking forward to it as much as I am Here are some notes on what I'm looking out for on the program this year: #hotchips34 @hotchipsorg https://www.eetimes.com/here-comes-hot-chips/ https://www.eetimes.com/here-comes-hot-chips/"
[X Link](https://x.com/anyuser/status/1559471919409381376) 2022-08-16T09:27Z [----] followers, [--] engagements
"It's official: @graphcoreai has been acquired by @SoftBank. Masayoshi Son laid out his grand vision for artificial superintelligence last month and it now seems clear Graphcore will be a part of that plan. Details in the article: https://www.eetimes.com/ai-chip-startup-graphcore-acquired-by-softbank/ https://www.eetimes.com/ai-chip-startup-graphcore-acquired-by-softbank/"
[X Link](https://x.com/anyuser/status/1811685564447338727) 2024-07-12T08:54Z [----] followers, [----] engagements
".@Hailo_ai's new edge chip for #GenAI has slightly less compute than the previous gen but with an optimised memory subsystem for #LLMs. It can manage [--] TOPS at INT4 or Llama2-7B at [--] tokens per second in a power envelope of 5W. #edgeai #aichips https://www.eetimes.com/hailo-debuts-edge-genai-chip-raises-120-million/ https://www.eetimes.com/hailo-debuts-edge-genai-chip-raises-120-million/"
[X Link](https://x.com/anyuser/status/1775158972833579250) 2024-04-02T13:50Z [----] followers, [----] engagements
"New @intel products CEO @MJHolthaus dropped hints about what her strategy will be. For AI she acknowledged Intel's difficulties with Gaudi and didn't mince words about Falcon Shores ("Is it going to be wonderful No. But it is a good first step.") More: https://www.eetimes.com/intel-products-ceo-gives-hints-on-new-strategy/ https://www.eetimes.com/intel-products-ceo-gives-hints-on-new-strategy/"
[X Link](https://x.com/anyuser/status/1868952992076177889) 2024-12-17T09:34Z [----] followers, [----] engagements
"On my recent trip to California I had the pleasure of sitting down in person with @AyarLabs CEO Charlie Wuischpard to find out how Ayar is building its ecosystem its technology roadmap and what the eventual exit plan is. Read my interview here: https://www.eetimes.com/at-ayar-labs-its-all-coming-together/ https://www.eetimes.com/at-ayar-labs-its-all-coming-together/"
[X Link](https://x.com/anyuser/status/1579848499998621696) 2022-10-11T14:56Z [----] followers, [--] engagements
".@kinara_ai has revamped its architecture for a second-gen edge AI accelerator - big enough and powerful enough to run Llama2-7B within its 6-W power envelope. More in the article on architectural changes etc: #ai #aiaccelerator #aihardware #edgeai #llm https://www.eetimes.com/kinara-eyes-genai-with-6-w-edge-chip/ https://www.eetimes.com/kinara-eyes-genai-with-6-w-edge-chip/"
[X Link](https://x.com/anyuser/status/1742182402019975488) 2024-01-02T13:53Z [----] followers, [----] engagements
".@graphcoreai has supercharged its IPU using TSMC's wafer-on-wafer technology. The same processor die on the same 7nm process can run 40% faster with resulting performance boost which Graphcore plans to offer at the same price as its previous-gen tech. https://www.eetimes.com/graphcore-supercharges-ipu-with-wafer-on-wafer/ https://www.eetimes.com/graphcore-supercharges-ipu-with-wafer-on-wafer/"
[X Link](https://x.com/anyuser/status/1499327740964659202) 2022-03-03T10:16Z [----] followers, [--] engagements
"Had the pleasure of chatting with @SambaNovaAI CEO @RodrigoLiang about the company's [---] trillion-parameter #ai model Samba-1. SambaNova's answer to #GPT-4 is a composition of experts not mixture of experts which is better for the enterprise he said: https://www.eetimes.com/sambanova-trains-trillion-parameter-model-to-take-on-gpt-4/ https://www.eetimes.com/sambanova-trains-trillion-parameter-model-to-take-on-gpt-4/"
[X Link](https://x.com/anyuser/status/1765366085174403329) 2024-03-06T13:17Z [----] followers, [----] engagements
"And that's a wrap The final podcast for the AI Hardware Show [----] is out Stay tuned for more in [----] 😉 Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss https://t.co/MwXbjFp0GU with @sallywf Big thanks to @VentanaMicro #sponsored https://t.co/VRv8MZC4YP Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss https://t.co/MwXbjFp0GU with"
[X Link](https://x.com/anyuser/status/1755568298169622541) 2024-02-08T12:24Z [----] followers, [----] engagements
"Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss with @sallywf Big thanks to @VentanaMicro #sponsored https://youtu.be/keAZy__gsAo https://youtu.be/keAZy__gsAo"
[X Link](https://x.com/anyuser/status/1755562263740633161) 2024-02-08T12:00Z 51.3K followers, [----] engagements
"The Snapdragon X Elite - @Qualcomm's new laptop chip with CPU cores designed at Nuvia - has [--] TOPS (INT8) total AI performance across CPU GPU and Hexagon NPU. Kevin @Krewell reports from Snapdragon Summit in Hawaii for EE Times. At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://t.co/r4wORxBUsL https://t.co/qjY1ZdJmXz At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://t.co/r4wORxBUsL https://t.co/qjY1ZdJmXz"
[X Link](https://x.com/anyuser/status/1718978783145849129) 2023-10-30T13:10Z [----] followers, [----] engagements
"At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://www.eetimes.com/qualcomm-takes-on-amd-apple-intel-with-snapdragon-x-elite/ https://www.eetimes.com/qualcomm-takes-on-amd-apple-intel-with-snapdragon-x-elite/"
[X Link](https://x.com/anyuser/status/1718976141178946023) 2023-10-30T13:00Z 36.4K followers, [----] engagements
".@tenstorrent has licensed CPU IO and memory chiplet IP to Japanese research institute LSTC who are building their own chiplet-based edge AI chip. Tenstorrent chief architect Wei-Han Lien told me he is most concerned about compatibility. More in article: https://www.eetimes.com/tenstorrent-licenses-chiplet-designs-to-japanese-institute/ https://www.eetimes.com/tenstorrent-licenses-chiplet-designs-to-japanese-institute/"
[X Link](https://x.com/anyuser/status/1763570644350894453) 2024-03-01T14:23Z [----] followers, [----] engagements
"AI chipmakers @cerebras @SambaNovaAI join @GroqInc in offering cloud inference APIs for specialized hardware. All three claim big single-user speed advantages versus GPU-based clouds but what's the motivation for this apparent change in strategy https://www.eetimes.com/token-wars-heats-up-as-cerebras-and-sambanova-enter-the-fray/ https://www.eetimes.com/token-wars-heats-up-as-cerebras-and-sambanova-enter-the-fray/"
[X Link](https://x.com/anyuser/status/1833590096068813212) 2024-09-10T19:35Z [----] followers, [----] engagements
"What about optimal sizing of the arrays for each model/layer Dataflow/datapath optimization at the architecture level Balance between memory and compute for prefill and decode stages Hardware support for other transformer operators (2/x)"
[X Link](https://x.com/anyuser/status/1948374742089470231) 2025-07-24T13:28Z [----] followers, [----] engagements
"@IanCutress We used to have another term for a programmable ASIC - ASSP - but you don't see this term any more. (In the industry we used to say "ASIC" for a fixed-function chip that's built for one customer only)"
[X Link](https://x.com/anyuser/status/1895493341506359745) 2025-02-28T15:16Z [----] followers, [----] engagements
"The legendary journalist that is @junkoyoshida is retiring today after more than three decades with @eetimes - au revoir Junko and thank you for everything Junko Paints Life with VibrantColors https://t.co/0ZZnd08Lq8 Junko Paints Life with VibrantColors https://t.co/0ZZnd08Lq8"
[X Link](https://x.com/anyuser/status/1393125487732568064) 2021-05-14T08:46Z [----] followers, [--] engagements
"Junko Paints Life with VibrantColors https://www.eetimes.eu/junko-paints-life-with-vibrant-colors/ https://www.eetimes.eu/junko-paints-life-with-vibrant-colors/"
[X Link](https://x.com/anyuser/status/1393122961373204486) 2021-05-14T08:36Z [---] followers, [--] engagements
"I called by @tenstorrent for a chat with CEO Jim Keller @jimkxa about his commitment to open-source the training systems he's building and why he wants to do the opposite of whatever Jensen does: https://www.eetimes.com/jim-keller-whatever-nvidia-does-well-do-the-opposite https://www.eetimes.com/jim-keller-whatever-nvidia-does-well-do-the-opposite"
[X Link](https://x.com/anyuser/status/1930704169997881737) 2025-06-05T19:11Z [----] followers, [----] engagements
"@IanCutress Love the thumbnail with "New Intel CEO" plastered over you. Makes it look like Gelsinger is handing the reins to Tech Tech Potato 😅"
[X Link](https://x.com/anyuser/status/1863705060678475930) 2024-12-02T22:01Z [----] followers, [----] engagements
"Delighted to speak with @petewarden about his new company Useful Sensors. The startup aims makes AI-enabled sensors "that just work". Their first product a 20x20mm person detection board is out now #ai #ml #tinyml #aisensor #sensormodule #aiot https://www.eetimes.com/pete-wardens-startup-puts-ai-in-the-sensor/ https://www.eetimes.com/pete-wardens-startup-puts-ai-in-the-sensor/"
[X Link](https://x.com/anyuser/status/1582998292295651328) 2022-10-20T07:32Z [----] followers, [--] engagements
"In other words I think there are several dimensions that could be specialised on for transformers beyond generic GEMM capability. This is instinctive to people who understand the concept of ASICs outside of data center/AI (3/x)"
[X Link](https://x.com/anyuser/status/1948374807269027843) 2025-07-24T13:28Z [----] followers, [----] engagements
"@EnerTuition @intel Strongly disagree. Gelsinger inherited years of mis-management which is still playing out. Turning a ship that size around will take more years"
[X Link](https://x.com/anyuser/status/1864402765952196690) 2024-12-04T20:13Z [----] followers, [---] engagements
"This is huge Congratulations @clattner_llvm and @Modular team Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇 https://t.co/mnHqtM8s0Y Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇 https://t.co/mnHqtM8s0Y"
[X Link](https://x.com/anyuser/status/1932503885811253379) 2025-06-10T18:23Z [----] followers, [----] engagements
"Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇"
[X Link](https://x.com/anyuser/status/1932488921138823212) 2025-06-10T17:23Z 21K followers, 118.8K engagements
"I will be moderating a panel at next week's AI Hardware Summit in London The panel will take on hardware challenges for AI in the data center and at the edge and our hardware experts will be available to answer your questions. #ai #aihardware #aihardwaresummit #event #aichips"
[X Link](https://x.com/anyuser/status/1800444292240125979) 2024-06-11T08:25Z [----] followers, [----] engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing
@sallywf Sally Ward-Foxton
Sally Ward-Foxton posts on X about ai, ceo, inference, in the the most. They currently have [------] followers and [---] posts still getting attention that total [-----] engagements in the last [--] hours.
Engagements: [-----] #
- [--] Week [-----] +235%
- [--] Month [------] -30%
- [--] Months [------] +42%
- [--] Year [-------] -32%
Mentions: [--] #
- [--] Week [--] +460%
- [--] Month [--] +258%
- [--] Months [--] +145%
- [--] Year [--] +129%
Followers: [------] #
- [--] Week [-----] +0.12%
- [--] Month [-----] +0.52%
- [--] Months [-----] +21%
- [--] Year [-----] +31%
CreatorRank: [---------] #
Social Influence
Social category influence technology brands 25.58% stocks 18.14% finance 6.05% social networks 4.19% travel destinations 3.72% events #946 countries 3.26% fashion brands 0.93% cryptocurrencies 0.47% currencies 0.47%
Social topic influence ai 31.16%, ceo 9.77%, inference 7.91%, in the 6.98%, ip 4.65%, llm 3.72%, london 2.79%, $googl 2.33%, jim 2.33%, data center 2.33%
Top accounts mentioned or mentioned by @nvidia @iancutress @intel @tenstorrent @eetimes @groqinc @sambanovaai @cerebras @synopsys @ambarellainc @amd @brainchipinc @ayarlabs @dmatrixai @ndahad @rodrigoliang @esperantotech @jimkxa @positronai @trsohmers
Top assets mentioned Alphabet Inc Class A (GOOGL) Microsoft Corp. (MSFT) Ambarella, Inc. Ordinary Shares (AMBA) Flex Ltd. Ordinary Shares (FLEX) Lends (LENDS) Gains (GAINS) Synopsys, Inc. (SNPS)
Top Social Posts
Top posts by engagements in the last [--] hours
"@tesla_teamx Intel has made GPUs including a very famous data center GPU for decades"
X Link 2026-02-05T05:01Z [----] followers, [--] engagements
"I was actively enjoying an entire feed full of British garden birds this week then made the mistake of clicking on a picture of a poem. Now every tweet is a poem and I'm looking for that one tweet about the semiconductor industry to engage with so I can get back to reality. Argh don't like how the algo is overly reactive. i've spent years liking animal-related content on here but if you like one tweet about potatoes or whatever it's like potatoes all the way down until another animal tweet appears and you have to treat it like a life preserver don't like how the algo is overly reactive. i've"
X Link 2026-02-13T11:28Z [----] followers, [---] engagements
"RT @JP_Biz: Ok brother do you think you could make Outlook search work when you have a minute"
X Link 2026-02-13T13:14Z [----] followers, [----] engagements
"AI chip startup @positron_ai has raised $230 million as it continues to sell its FPGA-based AI inference servers and work on its chiplet-based ASIC. Among strategic investors were financial trading firms who need really fast tokens @trsohmers told me: https://www.eetimes.com/positron-230-million-funding-led-by-financial-trading-firms/ https://www.eetimes.com/positron-230-million-funding-led-by-financial-trading-firms/"
X Link 2026-02-04T16:56Z [----] followers, [----] engagements
"Oh my sweet summer child. Intel will start making GPUs a market dominated by Nvidia https://t.co/QotvII3gMH Intel will start making GPUs a market dominated by Nvidia https://t.co/QotvII3gMH"
X Link 2026-02-05T04:20Z [----] followers, [----] engagements
"RT @Ronanchamberss: junior BBC journalist: hey so for my first piece i was thinking about writing something on how London based AI lab Deep"
X Link 2026-02-09T09:06Z [----] followers, [--] engagements
"(Reporting from Web Summit Qatar:) @imec_int's Max Mirgoli told me Imec is opening an R&D base in Doha in order to take advantage of talent from countries that might find it hard to get visas for US or Europe: https://www.eetimes.com/imec-to-open-rd-center-in-qatar/ https://www.eetimes.com/imec-to-open-rd-center-in-qatar/"
X Link 2026-02-09T08:11Z [----] followers, [---] engagements
"At @websummitqatar I chatted to @Atom_Computing and @qctrlHQ about sovereign quantum computing strategies. Countries should invest in quantum BEFORE we hit quantum advantage they said focusing on developing talent while avoiding techno-nationalism: https://www.eetimes.com/the-sovereign-quantum-race-starts-before-quantum-advantage/ https://www.eetimes.com/the-sovereign-quantum-race-starts-before-quantum-advantage/"
X Link 2026-02-11T12:00Z [----] followers, [---] engagements
"3 things I learned at Web Summit Qatar last week: https://youtube.com/shorts/K4rO6IdvVwYfeature=share https://youtube.com/shorts/K4rO6IdvVwYfeature=share"
X Link 2026-02-12T11:52Z [----] followers, [---] engagements
"My reporting from Qatar @websummitqatar - all in once place. Had a brilliant trip and hope to return next year. Qatars Deep Tech Ambitions Come Into Focus https://t.co/oMM3rs6XFh https://t.co/OF2ozf5hPL Qatars Deep Tech Ambitions Come Into Focus https://t.co/oMM3rs6XFh https://t.co/OF2ozf5hPL"
X Link 2026-02-13T09:05Z [----] followers, [---] engagements
"Qatar plans to capitalise on AI by investing in strategic tech and companies. Which parts of the AI stack should a country like Qatar involve itself in and which parts should it tackle first I spoke with @AyarLabs @dMatrix_AI to get some perspective: https://www.eetimes.com/qatars-ai-ambitions-span-the-tech-stack/ https://www.eetimes.com/qatars-ai-ambitions-span-the-tech-stack/"
X Link 2026-02-13T16:23Z [----] followers, [---] engagements
"Breaking News: Flex Logix has been acquired by Analog Devices. The 10-year-old eFPGA companys technical assets including its InferX AI IP and its entire technical team will transfer to existing customer ADI. @efpga @ADI_News https://www.eetimes.com/flex-logix-acquired-by-analog-devices/ https://www.eetimes.com/flex-logix-acquired-by-analog-devices/"
X Link 2024-11-08T17:28Z [----] followers, [---] engagements
"Very Excited for @eetimes AI Everywhere - our virtual conference on AI chips which is free to attend. I will be hosting day [--] which is about AI in the data center and moderating a panel about inference at scale. Don't miss it The rapid growth of Large Language Models (#LLMs) has driven the development of valuable #AI capabilities. Check out more in the keynote session by @GroqInc at our virtual event AI Everywhere 👉🏻 Register now: https://t.co/6ng3dtgnO1 https://t.co/KCa9D4jDjf The rapid growth of Large Language Models (#LLMs) has driven the development of valuable #AI capabilities. Check"
X Link 2024-11-20T14:45Z [----] followers, [----] engagements
"I had a chat with @BlumindInc at Electronica. The company is pressing forwards with its analog keyword spotting chip and a CNN accelerator. But the company has LLMs in its sights as the tech lends itself to bigger chiplet-based designs: https://www.eetimes.com/blumind-hints-at-big-chip-on-roadmap/ https://www.eetimes.com/blumind-hints-at-big-chip-on-roadmap/"
X Link 2024-11-21T12:38Z [----] followers, [---] engagements
".@tenstorrent has raised a whopping $693 million in a round that will put any doubts about the investment community's appetite for AI chip companies aside. The company has $150 million in deals committed mainly IP with next gen chip coming in early 2025: https://www.eetimes.com/tenstorrent-raises-693-million-series-d/ https://www.eetimes.com/tenstorrent-raises-693-million-series-d/"
X Link 2024-12-02T17:03Z [----] followers, [----] engagements
"My hot take: @intel will never find another Gelsinger. The 30-year Intel veteran former Intel CTO and former VMWare CEO is irreplaceable. https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/ https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/"
X Link 2024-12-04T09:40Z [----] followers, 54K engagements
"@sysadafterdark A potato could tell you're wrong. "Gelsinger killed intel for short term gains": these short term gains you speak of. Are they in the room with us now"
X Link 2024-12-04T16:45Z [----] followers, [---] engagements
"Intel has hired [--] board members both with industry chops presumably to counter criticism about the board's experience in the wake of Pat's exit. Former ASML CEO Eric Meurice and longtime Microchip CEO Steve Sanghi (one foundry side one product side). https://www.intc.com/news-events/press-releases/detail/1721/intel-appoints-semiconductor-leaders-eric-meurice-and-steve https://www.intc.com/news-events/press-releases/detail/1721/intel-appoints-semiconductor-leaders-eric-meurice-and-steve"
X Link 2024-12-05T17:13Z [----] followers, [---] engagements
"Wondering if beginning more of my tweets with "IT'S ABOUT TO START THE BIGGEST TECH BATTLE OF OUR TIME" (and then getting the product name wrong) would get me [---] million views per tweet Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time: https://t.co/KAUpGpDcKJ Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI"
X Link 2024-12-19T11:04Z [----] followers, [----] engagements
"Had a chat with @CEVA_IP's Moshe Sheier about the company's NeuPro-Nano including why tinyML needs a standalone core that can do DSP+AI (not separate accelerators or a scaled-down NeuPro-M) and why the company is already working with @EdgeImpulse. More: https://www.eetimes.com/ceva-seeks-to-exploit-synergies-in-portfolio-with-nano-npu/ https://www.eetimes.com/ceva-seeks-to-exploit-synergies-in-portfolio-with-nano-npu/"
X Link 2024-12-20T13:55Z [----] followers, [---] engagements
"In a new type of collaboration @SiMa_Inc and @Synopsys will make an IP product for automotive based on SiMa's ML IP and Synopsys' digital twin tech. The automotive industry is calling for more influence on silicon both companies told me: https://www.eetimes.com/synopsys-collaborates-with-sima-ai-on-automotive-ai-ip/ https://www.eetimes.com/synopsys-collaborates-with-sima-ai-on-automotive-ai-ip/"
X Link 2024-12-23T10:20Z [----] followers, [---] engagements
"The US government announced stricter rules on the export of AI chips which now need a license to be exported to almost any country. Closed model weights are also restricted. The aim is to prevent China from developing frontier models. My summary: https://www.eetimes.com/uproar-as-biden-administration-limits-ai-chip-shipments/ https://www.eetimes.com/uproar-as-biden-administration-limits-ai-chip-shipments/"
X Link 2025-01-16T11:05Z [----] followers, [---] engagements
"He does say "$100 billion already committed" in this video. (Will OpenAI build chips "we'll see" he says) At WSJ Journal House Davos @OpenAI Chief Product Officer Kevin Weil discusses #Stargate outlines #ChatGPTs [----] roadmap https://t.co/PacIV7mYjV #AI #HPC At WSJ Journal House Davos @OpenAI Chief Product Officer Kevin Weil discusses #Stargate outlines #ChatGPTs [----] roadmap https://t.co/PacIV7mYjV #AI #HPC"
X Link 2025-01-23T09:54Z [----] followers, [---] engagements
"Feel like this is very technically feasible and would go a long way towards getting people out of social media echo chambers or at least in showing people echo chambers exist imo all apps with a 'for you' timeline should allow you to toggle a bizzaro switch and browse the 'NOT for you' timeline just to feel what the opposite is like imo all apps with a 'for you' timeline should allow you to toggle a bizzaro switch and browse the 'NOT for you' timeline just to feel what the opposite is like"
X Link 2025-02-03T08:29Z [----] followers, [---] engagements
"DARE the largest chip project ever funded by the EU has kicked off with $260 million in funding for HPC chip sovereignty. The money will go to a consortium including @AxeleraAI who will develop an AI accelerator chiplet for HPC: https://www.eetimes.com/european-project-gets-260-million-for-hpc-chip-sovereignty/ https://www.eetimes.com/european-project-gets-260-million-for-hpc-chip-sovereignty/"
X Link 2025-03-07T10:04Z [----] followers, [---] engagements
".@EdgeImpulse has been snapped up by @Qualcomm. Congratulations Zach Jan Adam and team (My colleague @ndahad has the story at EE Times dot com). https://www.eetimes.com/qualcomm-hopes-to-expand-its-intelligent-iot-with-edge-impulse-acquisition/ https://www.eetimes.com/qualcomm-hopes-to-expand-its-intelligent-iot-with-edge-impulse-acquisition/"
X Link 2025-03-10T13:39Z [----] followers, [----] engagements
"My hot take: A sigh of relief as Lip-Bu Tan is appointed CEO of @intel https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/ https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/"
X Link 2025-03-13T01:12Z [----] followers, [----] engagements
"While this year's production felt like less of a spectacle there was certainly plenty of content in Jensen Huang's keynote at @nvidia GTC. He talked a lot about tokens single user speeds and what @dylan522p calls "tokenomics". https://www.eetimes.com/huang-talks-tokens-reveals-roadmap-at-gtc-2025/ https://www.eetimes.com/huang-talks-tokens-reveals-roadmap-at-gtc-2025/"
X Link 2025-03-19T18:46Z [----] followers, [---] engagements
"ICYMI: @ndahad and I recap @nvidia GTC and @Synopsys SNUG conferences live from California https://www.youtube.com/watchv=dUViIU7dWYk https://www.youtube.com/watchv=dUViIU7dWYk"
X Link 2025-03-22T04:30Z [----] followers, [---] engagements
".@OpenAI's Richard Ho keynoted @Synopsys Snug & I was on the front row Appearing to confirm he is working on AI silicon Ho said his team is addressing every layer of the stack for huge computers where GPUs will nevertheless "be a huge part of that mix": https://www.eetimes.com/openai-hardware-chief-ai-scaling-laws-will-continue/ https://www.eetimes.com/openai-hardware-chief-ai-scaling-laws-will-continue/"
X Link 2025-03-28T10:31Z [----] followers, [---] engagements
".@nvidia @NVIDIAAI unveiled some impressive inference performance results at GTC but how did they do it Ian Buck went through some of the details of Dynamo the company's new inference serving software: https://www.eetimes.com/how-did-nvidia-improve-hopper-inference-performance-30x/ https://www.eetimes.com/how-did-nvidia-improve-hopper-inference-performance-30x/"
X Link 2025-04-01T08:19Z [----] followers, [---] engagements
"@IanCutress "SXSW London will monitor coverage generated by all attending media to ensure coverage plans and assignments were delivered. Those with no proof of coverage may be declined media accreditation for any future conferences." -- yeah it's a no from me"
X Link 2025-05-08T15:47Z [----] followers, [---] engagements
"@IanCutress @sxsw I expect they get a lot of chancers trying to get in to see the "celebrities" but making admittance dependent on coverage means they will encourage more chancers and less bona fide press which I expect is the opposite of what they want to achieve"
X Link 2025-05-08T16:12Z [----] followers, [---] engagements
"@IanCutress @nvidia @MediaTek @MarvellTech @AlchipTECH @dylan522p It's like a Formula [--] Paddock interview hahahaha"
X Link 2025-05-19T20:41Z [----] followers, [---] engagements
"Are you at The AI Summit in London tomorrow I will be presenting two short fireside chats with @SambaNovaAI CEO @RodrigoLiang. Pop along and hear Rodrigo's views on a variety of topics from open source vs proprietary models to AI's energy consumption: https://london.theaisummit.com/speakers/sally-ward-foxton https://london.theaisummit.com/speakers/sally-ward-foxton"
X Link 2025-06-10T12:32Z [----] followers, [---] engagements
"@DylanOnChips Thanks for your service 🫡"
X Link 2025-06-10T15:00Z [----] followers, [---] engagements
".@EnCharge_AI has launched a [---] TOPS [--] TOPS/W AI chip for the PC. It uses the company's capacitor-based analog compute scheme which is not affected by temperature or process variations. More in the article: https://www.eetimes.com/encharge-picks-the-pc-for-its-first-analog-ai-chip/ https://www.eetimes.com/encharge-picks-the-pc-for-its-first-analog-ai-chip/"
X Link 2025-06-16T08:00Z [----] followers, [---] engagements
"Sad news as @EsperantoTech all but shuts down. The tech is still for sale or can be licensed CEO Art Swift told me but the majority of employees have already moved on. The company's European teams were decimated by competitors Swift said: https://www.eetimes.com/ai-startup-esperanto-winds-down-silicon-business/ https://www.eetimes.com/ai-startup-esperanto-winds-down-silicon-business/"
X Link 2025-07-07T06:49Z [----] followers, [---] engagements
"My report from @DACconference is on the state of multi-agent AI systems in EDA and the vision for increasingly autonomous workflows. Featuring @Microsoft @Synopsys and @Siemens EDA: https://www.eetimes.com/dac-2025-towards-multi-agent-systems-in-eda/ https://www.eetimes.com/dac-2025-towards-multi-agent-systems-in-eda/"
X Link 2025-07-07T18:27Z [----] followers, [---] engagements
"Big milestone for South Korean startup @FuriosaAI as LG picks Furiosa hardware for its LLM appliance. I spoke to CEO June Paik about the rigorous testing LG put Furiosa's hardware through and Furiosa's new multi-chip tensor parallel capabilities: https://www.eetimes.com/lg-to-ship-ai-appliance-with-furiosa-chips/ https://www.eetimes.com/lg-to-ship-ai-appliance-with-furiosa-chips/"
X Link 2025-07-23T11:29Z [----] followers, [---] engagements
"OK I'll bite. Yes making a transformer chip does mean making a large matmul engine. But there is a lot more to efficient transformer inference than GEMM/FLOPS. You need to be able to feed and use those FLOPS efficiently. (1/x) Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well https://t.co/Ju8QX9gcc7 Although "baking transformers in silicon" may sound cool it is mostly"
X Link 2025-07-24T13:28Z [----] followers, 27.4K engagements
"In other words I think there are several dimensions that could be specialised on for transformers beyond generic GEMM capability. This is instinctive to people who understand the concept of ASICs outside of data center/AI (3/x)"
X Link 2025-07-24T13:28Z [----] followers, [----] engagements
"If its just about FLOPS why is current best practice disaggregated inference ie split prefill and decode stages across heterogeneous hardware (currently mostly across different types of GPUs but if transformers are just about FLOPS why use different hardware at all) (4/x)"
X Link 2025-07-24T13:29Z [----] followers, [----] engagements
"If theres no point optimizing for AI/transformers beyond GPU how do other semiconductor companies with AI-specialised architectures (see: @GroqInc @CerebrasSystems) consistently beat Nvidias inference performance I guess it depends on whether you think best = fastest. (5/x)"
X Link 2025-07-24T13:29Z [----] followers, [----] engagements
"@GroqInc @cerebras Speaking of Groq and Cerebras neither uses HBM so I dont think your memory wall argument holds water here either. (6/x)"
X Link 2025-07-24T13:30Z [----] followers, [----] engagements
"@GroqInc @CerebrasSystems Im sure you know that GPUs often have famously poor memory bandwidth utilization for AI (in the order of 30%); with hardware optimization it is perfectly possible to utilize more of that bandwidth than GPUs can/do (see: @positron_ai) (7/x)"
X Link 2025-07-24T13:30Z [----] followers, [----] engagements
"@itsclivetime Thanks Clive really appreciate your thoughtful response and I agree on a lot. Partly it was the flippancy of the original tweet I wanted to temper. To your point - I do hope programmability remains if it means we can look forward to seeing more brain-like models in production"
X Link 2025-07-25T09:18Z [----] followers, [---] engagements
"@SemiAnalysis_ If there is a joke here I don't get it. @grok please educate me as to how this isn't misogynistic bullshit"
X Link 2025-07-30T09:00Z [----] followers, [---] engagements
"South Korean edge AI chip startup DeepX has partnered with Chinese hyperscaler Baidu. The companies will optimize Baidus ERNIE LLM for DeepX hardware. This will enable direct access to one of the largest AI developer communities in China: https://www.eetimes.com/deepx-partners-with-baidu-to-attack-china-market/ https://www.eetimes.com/deepx-partners-with-baidu-to-attack-china-market/"
X Link 2025-08-15T13:09Z [----] followers, [---] engagements
".@Cornami_Inc is touting its silicon for FHE or AI acceleration but the real promise is in doing both - AI inference on encrypted data - former CEO Wally Rhines told me. Even without the encryption the company has competitive tokens-per-second figures: https://www.eetimes.com/cornami-targets-fhe-encrypted-ai-in-real-time/ https://www.eetimes.com/cornami-targets-fhe-encrypted-ai-in-real-time/"
X Link 2025-09-02T08:29Z [----] followers, [---] engagements
"Next week I'll be at the AI Infra Summit in Santa Clara - please come and see my panel discussion about how AI in EDA will affect chip startups featuring AI EDA and Startup experts @Cognichip Stelios Diamantidis @Synopsys Arun Venkatachar and Silicon Catalyst 's Laura Swan"
X Link 2025-09-02T08:35Z [----] followers, [----] engagements
"An ambitious new collaboration Project Orbion aims to build a real-time digital twin of the Earth by combining various imaging data sources and offering them for commercial use. I spoke to some of the partners in San Francisco this week: https://www.eetimes.com/project-aims-to-make-real-time-digital-twin-of-the-world/ https://www.eetimes.com/project-aims-to-make-real-time-digital-twin-of-the-world/"
X Link 2025-09-12T17:13Z [----] followers, [---] engagements
"I had a great time chatting to @Infineon's Gregory Guez about the major trends defining AI in the smart home including power efficiency privacy and security. Listen to our full conversation on my audio podcast: #sponsored https://www.eetimes.com/podcasts/living-on-the-edge-intelligence-power-efficiency-and-security-take-smart-home-devices-to-the-next-level/ https://www.eetimes.com/podcasts/living-on-the-edge-intelligence-power-efficiency-and-security-take-smart-home-devices-to-the-next-level/"
X Link 2025-09-12T17:16Z [----] followers, [---] engagements
"AI metasurface startup @neurophos has test silicon - showing the final AI inference accelerator will reach close to [---] TOPS/W (4-bit). CEO Patrick Bowen gave me an update on which software startup the company has acquired and its new data center partner: https://www.eetimes.com/neurophos-photonic-ai-test-chip-hits-300-tops-w/ https://www.eetimes.com/neurophos-photonic-ai-test-chip-hits-300-tops-w/"
X Link 2025-09-24T11:47Z [----] followers, [---] engagements
"In the first of a series of reports from Stanford University's SPRC Symposium I spoke to @NttResearch's Timothy McKenna about TFLN (thin-film lithium niobate) and the material's potential for photonic computing chips: https://www.eetimes.com/can-tfln-make-photonic-compute-competitive/ https://www.eetimes.com/can-tfln-make-photonic-compute-competitive/"
X Link 2025-10-01T08:12Z [----] followers, [---] engagements
"The unique capabilities of the Xcore mean XMOS is well-positioned to apply generative AI to its toolchain so the designer can specify at the behavioural level what the system needs to do and have generative AI implement it CEO Mark Lippett told me: https://www.eetimes.com/genai-can-raise-level-of-abstraction-for-xmos-tools/ https://www.eetimes.com/genai-can-raise-level-of-abstraction-for-xmos-tools/"
X Link 2025-10-08T08:32Z [----] followers, [---] engagements
"World's worst-kept secret. https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration/ https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration/"
X Link 2025-10-13T18:46Z [----] followers, [---] engagements
"I'm on @EETimesOnAir discussing London Tech Week and the battle between cities to become the hub of AI tech plus alternative AI accelerator hardware from CogX. #cogx19 #AI #LondonTechWeek #aitechnology #Graphcore #Mythic #Oxfordquantumcircuits #quantumcomputing #analogcomputing #AI hubs; #HDmaps for #robocars; #supercomputer bragging rights. @sallywf @rickbmerritt @Porsche44 @VSI_Labs @EETimesOnAir https://t.co/d8iRRCo8bU #AI hubs; #HDmaps for #robocars; #supercomputer bragging rights. @sallywf @rickbmerritt @Porsche44 @VSI_Labs @EETimesOnAir https://t.co/d8iRRCo8bU"
X Link 2019-06-24T07:42Z [----] followers, [--] engagements
".@MLPerf training results are out There's a new GPT-3 benchmark and @NVIDIA + @CoreWeave trained it the fastest. Their only challenger: @Intel @HabanaLabs. #ai #aibenchmark #mlperf #gpt3 #aitraining #aihardware https://www.eetimes.com/nvidia-leads-habana-challenges-on-mlperf-gpt-3-benchmark/ https://www.eetimes.com/nvidia-leads-habana-challenges-on-mlperf-gpt-3-benchmark/"
X Link 2023-06-28T19:25Z [----] followers, [----] engagements
"Hands on with @EsperantoTech silicon at #dac60 - [----] RISC-V cores on this 25W inference chip. They have Meta OPT 13B up and running on one chip. #ai #llm #aihardware #aiinference"
X Link 2023-07-11T00:50Z [----] followers, 28.2K engagements
".@nvidia showed off Grace Hopper for the first time in the latest round of MLPerf inference benchmarks. @HabanaLabs showed competitive performance with Gaudi2 on LLMs and Google debuted the TPUv5e. Check the article for details. #mlperf #ai #llm https://www.eetimes.com/nvidia-shows-off-grace-hopper-in-mlperf-inference/ https://www.eetimes.com/nvidia-shows-off-grace-hopper-in-mlperf-inference/"
X Link 2023-09-12T14:24Z [----] followers, [----] engagements
".@StabilityAI will buy [----] Gaudi2s plus Intel Xeon CPUs for an AI supercomputer that will be in the top [--] in the world (largest in Europe) said Gelsinger in his keynote this morning. Momentum on the AI continuum@IntelAI https://t.co/44eSDscRpb Momentum on the AI continuum@IntelAI https://t.co/44eSDscRpb"
X Link 2023-09-19T20:11Z [----] followers, [----] engagements
"A few weeks ago I visited @Ambarella_Inc HQ - they showed me Llama2 up and running on their autonomous driving chip. Will they make an LLM accelerator Its probably something we need to consider Ambarella CEO Fermi Wang told me. Read more: https://www.eetimes.com/ambarella-demos-llm-inference-on-autonomous-driving-chip/ https://www.eetimes.com/ambarella-demos-llm-inference-on-autonomous-driving-chip/"
X Link 2023-11-13T16:12Z [----] followers, [---] engagements
"Chinese automotive OEMs are able to move much quicker than their European counterparts and they've defined new classes of ADAS specifically for China said @Ambarella_Inc CEO. We also discussed the relationships between OEMs tier ones and chipmakers: https://www.eetimes.com/ambarella-ceo-chinese-oems-are-copying-the-tesla-model/ https://www.eetimes.com/ambarella-ceo-chinese-oems-are-copying-the-tesla-model/"
X Link 2023-11-21T09:04Z [----] followers, [---] engagements
"I saw @Ambarella_Inc's latest multi-modal LLM demos including LLaVA-34B on the N1 Ambarella's most powerful SoC which comes with an #LLM-friendly software stack. CTO Les Kohn told me multi-modal LLMs are set to revolutionise autonomous driving. Article: https://www.eetimes.com/llms-are-ready-for-robotics-and-self-driving-says-ambarella/ https://www.eetimes.com/llms-are-ready-for-robotics-and-self-driving-says-ambarella/"
X Link 2024-07-16T08:32Z [----] followers, [---] engagements
""The woman seems to be enjoying herself" - LLaVA-34B on Ambarella N1 gets it right first time Thanks @Ambarella_Inc for the demos 😊"
X Link 2024-07-16T08:36Z [----] followers, [---] engagements
"Another edge AI chip startup exits @kinara_ai is being acquired by @NXP for $307m. I spoke to NXP to find out why and get some details on what their plans are for Kinara and its hardware and software roadmaps. Congrats to Ravi Wajahat Rehan and team https://www.eetimes.com/nxp-acquires-ai-chip-startup-kinara/ https://www.eetimes.com/nxp-acquires-ai-chip-startup-kinara/"
X Link 2025-02-12T09:20Z [----] followers, [---] engagements
"I enjoyed moderating two engaging sessions on fast inference and agentic AI at The AI Summit (London Tech Week) with @SambaNovaAI CEO @RodrigoLiang Welcome back to London Rodrigo and team"
X Link 2025-06-13T09:33Z [----] followers, [---] engagements
"A short while ago I had an interesting conversation with @Ambarella_Inc CTO Les Kohn about reasoning techniques at the edge - other techniques will likely become more useful for resource constrained-environments especially for real-time compute: https://www.eetimes.com/will-robots-use-reasoning-at-the-edge/ https://www.eetimes.com/will-robots-use-reasoning-at-the-edge/"
X Link 2025-08-15T13:19Z [----] followers, [---] engagements
".@GoogleResearch developed a RISC-V superscalar CPU designed for edge AI which it is giving to the industry for free in a bid to avoid fragmentation in the edge AI ecosystem. @SynaCorp is the first to incorporate this NPU into production silicon: https://www.eetimes.com/google-open-sources-npu-ip-synaptics-implements-it/ https://www.eetimes.com/google-open-sources-npu-ip-synaptics-implements-it/"
X Link 2025-10-15T13:57Z [----] followers, [----] engagements
".@NextSilicon has revealed its HPC chip architecture which reconfigures itself during runtime based on the workload. It can run any HPC code out of the box CEO Elad Raz told me. The company also showed off a test chip for a big RISC-V CPU it's building: https://www.eetimes.com/nextsilicon-details-runtime-reconfigurable-architecture https://www.eetimes.com/nextsilicon-details-runtime-reconfigurable-architecture"
X Link 2025-10-23T06:51Z [----] followers, [----] engagements
"@FelixCLC_ @NextSilicon "flew under the radar" mate it was reported in EE Times"
X Link 2025-11-03T16:21Z [----] followers, [---] engagements
"Chatted with Moreh CEO about enabling cost-effective alternatives to Nvidia. Moreh built a training software stack for AMD GPU clusters - something AMD itself is still working on - and is about to launch an inference stack for AMD and Tenstorrent servers: https://www.eetimes.com/cost-efficient-ai-at-scale-is-a-software-problem/ https://www.eetimes.com/cost-efficient-ai-at-scale-is-a-software-problem/"
X Link 2025-11-07T12:57Z [----] followers, [----] engagements
"Eta Compute has rebranded as ModelCat and launched an AI agent that builds small edge models based on customers' datasets. The agent is hardware-aware and can test hypotheses on actual chips from a number of suppliers including @NXP: https://www.eetimes.com/eta-compute-rebrands-as-modelcat-unveils-agentic-model-builder/ https://www.eetimes.com/eta-compute-rebrands-as-modelcat-unveils-agentic-model-builder/"
X Link 2025-11-10T09:13Z [----] followers, [---] engagements
"Japanese crypto-mining ASIC company Triple-1 has applied its deep sub-threshold technology (design target 0.26V) to TSMC 3nm. Key use case is edge data centers that convert surplus renewable energy into Bitcoin. Could this be applied to AI More: https://www.eetimes.com/crypto-mining-asic-goes-deep-sub-threshold-on-3-nm/ https://www.eetimes.com/crypto-mining-asic-goes-deep-sub-threshold-on-3-nm/"
X Link 2025-11-19T09:44Z [----] followers, [---] engagements
"@GanguliArko 😂 I mean I really really want a house robot. I'm just against these videos that show performance in a misleading way. Much of this video is sped up 10x"
X Link 2025-11-20T13:41Z [----] followers, [--] engagements
".@SynaCorp is open-sourcing the compiler toolchain for its AI hardware including the open-source NPU that comes from Google. Dave Garrett talked me through the reasons behind this decision. Hear our full conversation on my audio podcast: #sponsored #ad https://www.eetimes.com/podcasts/inside-synaptics-decision-to-open-source-its-ai-compiler/ https://www.eetimes.com/podcasts/inside-synaptics-decision-to-open-source-its-ai-compiler/"
X Link 2025-11-26T14:56Z [----] followers, [---] engagements
"Q.ANT has unveiled a second-gen TFLN-based photonic compute chip. The German startup is banking on accelerating non-linear functions and KAN networks efficiently. I spoke to CEO Michael Frtsch for details on new funding customers and demo: https://www.eetimes.com/q-ant-raises-series-a-debuts-second-gen-tfln-photonic-chip/ https://www.eetimes.com/q-ant-raises-series-a-debuts-second-gen-tfln-photonic-chip/"
X Link 2025-11-26T15:07Z [----] followers, [---] engagements
"Lemurian Labs formerly an AI chip company has raised $28M to develop its AI software stack spanning the compiler layer all the way up to workload orchestration while enabling code portability. The CEO said he wants to do away with kernels altogether: https://www.eetimes.com/lemurian-labs-raises-28-million-for-ai-portability-software/ https://www.eetimes.com/lemurian-labs-raises-28-million-for-ai-portability-software/"
X Link 2025-12-04T19:22Z [----] followers, [---] engagements
"The biggest news story in my beat last year dropped at 10pm on Christmas Eve (thank you Santa.). Nvidia has all-but acquired Groq for reportedly $20B. I was surprised to say the least. I spent the last two weeks reading bad take after bad take (1/x)"
X Link 2026-01-08T10:17Z [----] followers, [----] engagements
"I have been writing about Groq for years. Much of what is on social media is ill-informed mediocre AI slop or just plain wrong. The truth is no-one outside the companies has any real idea why Nvidia all-but acquired Groq and what we know so far doesnt make much sense yet🧵(1/x)"
X Link 2026-01-08T16:58Z [----] followers, 93.5K engagements
"Nvidia integrating Groq chips/systems directly into its main data center tech roadmap seems far-fetched partly because it would mean admitting GPUs cant do everything and partly because software would be a nightmare because (5/x)"
X Link 2026-01-08T16:59Z [----] followers, [----] engagements
"The difference is not as simple as Groq only uses SRAM. Groq is so so different from a GPU. Its compiler is a huge part of its secret sauce. It would not fit into the CUDA ecosystem. eg there is no equivalent of a kernel in Groq arch (6/x)"
X Link 2026-01-08T16:59Z [----] followers, [----] engagements
"Nvidias biggest fear is hyperscalers own chips I would speculate that Nvidia didnt want to risk one of its big hyperscale customers acquiring Groq (10/x)"
X Link 2026-01-08T17:00Z [----] followers, [----] engagements
"Google TPU is great but eg Meta and Microsoft internal chip efforts are still relatively immature. Groqs hardware software and infrastructure might have made a difference there (11/x)"
X Link 2026-01-08T17:00Z [----] followers, [----] engagements
"This is obviously speculation and ultimately there are still many more questions than answers regarding this deal. Here is a longer-form version of my analysis on Groq and Nvidia (/end): https://www.eetimes.com/what-is-groq-nvidia-deal-really-about/ https://www.eetimes.com/what-is-groq-nvidia-deal-really-about/"
X Link 2026-01-08T17:00Z [----] followers, [----] engagements
".@Ambarella_Inc has changed its go-to-market strategy as it embraces the edge AI ecosystem particularly ISVs (the company previously supported big customers almost exclusively with no resellers and no channel). Also details on the new CV7 silicon: https://www.eetimes.com/ambarella-shifts-market-strategy-to-embrace-edge-ecosystem/ https://www.eetimes.com/ambarella-shifts-market-strategy-to-embrace-edge-ecosystem/"
X Link 2026-01-08T20:08Z [----] followers, [----] engagements
"@ourtown2 Thanks. As I study AI chip companies for my full time job I never would have known this"
X Link 2026-01-09T10:10Z [----] followers, [--] engagements
"@cdsmithus Yes but did it work then I am not aware of any customer who bought Groq chips back then and ran production workloads (maybe all secret). Pivot to TaaS for business reasons yes ("selling chips is too hard") but no doubt making whole stack "for internal consumption only" helped"
X Link 2026-01-09T12:40Z [----] followers, [--] engagements
"@cdsmithus That's great. I hope you're right that Nvidia saw value in Groq's tech. Personally would love to see what some sort of Nvida-Groq combo would look like and I hope we can get more details at GTC"
X Link 2026-01-09T14:56Z [----] followers, [--] engagements
"Troubled analog compute chip startup Mythic has pulled off a comeback with a fresh $125-million funding round. The company has fixed issues with its 1st gen chip and plans a 2nd gen data center class product and more according to its new CEO. Article: https://www.eetimes.com/mythic-rises-from-the-ashes-with-125-million-funding-round/ https://www.eetimes.com/mythic-rises-from-the-ashes-with-125-million-funding-round/"
X Link 2026-01-16T13:12Z [----] followers, [----] engagements
"Delighted to get the opportunity to ask @SiFive's John Simpson all about vector processor design for AI about the new RISC-V vector profiles and about how SiFive works with these profiles even before they are set in stone. Listen in to our deep-dive: #ad https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/ https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/"
X Link 2026-01-19T16:40Z [----] followers, [--] engagements
"@FelixCLC_ Let's just say you won't need the goose meme😉"
X Link 2026-01-19T17:06Z [----] followers, [---] engagements
"Quadric has raised another $30 million and has at least five licensees for its programmable NPU IP. I spoke to CMO Steve Roddy who explained a little about the NPU's architecture and how @quadric_io sees its competition in the space: https://www.eetimes.com/quadric-raises-series-c-for-programmable-npu-ip/ https://www.eetimes.com/quadric-raises-series-c-for-programmable-npu-ip/"
X Link 2026-01-20T10:24Z [----] followers, [----] engagements
"I had a chat with @dMatrix_AI CEO Sid Sheth a little while back about test time scaling what it is and what it means for hardware - podcast: (I also got a first-hand look at D-Matrix's 3D-stacked logic and DRAM test chip - see article link in next tweet) https://www.eetimes.com/podcasts/d-matrix-ceo-sid-sheth-on-ai-test-time-scaling/ https://www.eetimes.com/podcasts/d-matrix-ceo-sid-sheth-on-ai-test-time-scaling/"
X Link 2026-01-23T18:53Z [----] followers, [----] engagements
"@dMatrix_AI Here's the article with updates on D-Matrix's roadmap and some pics https://www.eetimes.com/as-demand-for-fast-ai-tokens-grows-d-matrix-develops-fast-nic/ https://www.eetimes.com/as-demand-for-fast-ai-tokens-grows-d-matrix-develops-fast-nic/"
X Link 2026-01-23T18:54Z [----] followers, [---] engagements
"@proxietuna I think Groq value at last funding was 7B - cf AMD acq of Xilinx at 49B ADI acq Maxim for 20B in [----]. and SambaNova at 1B. Groq must have a revenue figure (unlike some others) they can use to calculate value. Yes Mellanox cheap but that was pre-AI - all has since gone nuts"
X Link 2026-01-08T14:28Z [----] followers, [---] engagements
"Delighted to get the opportunity to ask @SiFive's John Simpson all about vector processor design for AI about the new RISC-V matrix profiles and about how SiFive works with these profiles even before they are set in stone. Listen in to our deep-dive: #ad https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/ https://www.eetimes.com/podcasts/scaling-ai-from-edge-to-data-center-with-sifive-risc-v-vectors/"
X Link 2026-01-19T16:52Z [----] followers, [----] engagements
".@AyarLabs and Alchip unveiled a reference design for big AI chips with optical I/O based on Ayar optical engines and TSMC COUPE tech. Ayar CTO Vladimir Stojanovic talked me through the significance of this reference platform: https://www.eetimes.com/ayar-labs-and-alchip-unveil-optical-i-o-reference-design/ https://www.eetimes.com/ayar-labs-and-alchip-unveil-optical-i-o-reference-design/"
X Link 2025-09-25T09:18Z [----] followers, [---] engagements
"RT @Pranavenstein: Intel has been making GPUs for years. When you dont know what youre writing about this is the kind of headline you ge"
X Link 2026-02-04T07:49Z [----] followers, [--] engagements
"My hot take: @intel will never find another Gelsinger. The 30-year Intel veteran former Intel CTO and former VMWare CEO is irreplaceable. https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/ https://www.eetimes.com/intel-will-never-find-another-gelsinger-but-do-they-want-one/"
X Link 2024-12-04T09:40Z [----] followers, 54K engagements
"Happy Tuesday everyone"
X Link 2024-08-06T09:39Z [----] followers, 41.6K engagements
"Edge AI chip companies be like:"
X Link 2023-11-30T15:42Z [----] followers, 28.5K engagements
"Geeking out at @nvidia with the new GH200 🤓"
X Link 2023-09-18T23:39Z [----] followers, 20.7K engagements
"Hands on with @EsperantoTech silicon at #dac60 - [----] RISC-V cores on this 25W inference chip. They have Meta OPT 13B up and running on one chip. #ai #llm #aihardware #aiinference"
X Link 2023-07-11T00:50Z [----] followers, 28.2K engagements
"@IanCutress .on [----] sqmm chips 😂"
X Link 2024-12-06T11:15Z [----] followers, [----] engagements
"Congrats @AMD and @UntetherAI on debut @MLPerf results AMD MI300X matches H100's inference performance while Untether can beat H200's power efficiency by 3x. Featuring also: @nvidia Blackwell @intel Granite Rapids and @Google Trillium. My analysis: https://www.eetimes.com/amd-and-untether-take-on-nvidia-in-mlperf-benchmarks/ https://www.eetimes.com/amd-and-untether-take-on-nvidia-in-mlperf-benchmarks/"
X Link 2024-08-29T08:06Z [----] followers, 13.5K engagements
"The Japanese government will send engineers to be trained in RISC-V and AI at @tenstorrent over the next [--] years. Tenstorrent gets free engineering talent revenue from the Japanese government and hopefully good relationships in the engineers' companies: https://www.eetimes.com/japan-sends-engineers-to-be-trained-at-tenstorrent/ https://www.eetimes.com/japan-sends-engineers-to-be-trained-at-tenstorrent/"
X Link 2024-11-05T13:28Z [----] followers, 11.5K engagements
"OK I'll bite. Yes making a transformer chip does mean making a large matmul engine. But there is a lot more to efficient transformer inference than GEMM/FLOPS. You need to be able to feed and use those FLOPS efficiently. (1/x) Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well https://t.co/Ju8QX9gcc7 Although "baking transformers in silicon" may sound cool it is mostly"
X Link 2025-07-24T13:28Z [----] followers, 27.4K engagements
"Although "baking transformers in silicon" may sound cool it is mostly just a marketing slogan. 90% of Transformers FLOPS are just GEMMs and 256x256/128x128 systolic arrays in TPU/Trainium are already optimized for GEMMs. Even modern GPGPU (with Tensor cores) are optimized well for GEMMs. Even if you "bake transformers in silicon" aka just create an gaint systolic array most of your die area will still be taken up with SRAM cells and you will still face the memory wall since your HBM memory bandwidth will be the same as GPGPU/TPUs/Trainium"
X Link 2025-07-24T02:21Z 60.1K followers, 51.1K engagements
".@BrainChip_inc is now offering its #neuromorphic #ai chips in an edge box suitable for security camera systems. Why are edge boxes becoming an increasingly popular path to market for edge AI chip makers And what is BrainChip's strategy for edge boxes https://www.eetimes.com/brainchip-adds-edge-box-to-chip-and-ip-offerings/ https://www.eetimes.com/brainchip-adds-edge-box-to-chip-and-ip-offerings/"
X Link 2024-03-05T20:46Z [----] followers, [----] engagements
"Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: https://t.co/S7pQFnOVJV "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: https://t.co/S7pQFnOVJV "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X"
X Link 2024-05-07T16:33Z [----] followers, [----] engagements
"Im Paris with the family this weekbut had to tweet this part of the M4 announcement out: "The Neural Engine in M4 is Apples most capable yet and is more powerful than any neural processing unit in any AI PC today." Its [--] TOPS. The @Snapdragon X Elite and X Plus have [--] TOPS. Now its clear why this launch feels rushed for M4. https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/ https://www.apple.com/newsroom/2024/05/apple-introduces-m4-chip/"
X Link 2024-05-07T16:20Z 18.7K followers, 25.4K engagements
"Called by @tenstorrent to interview Jim Keller and get a look at new Wormhole cards Thanks for having me"
X Link 2023-05-18T21:38Z [----] followers, [----] engagements
"Carmakers are testing neuromorphic technologies based on spiking neural networks. What will cars do with their newly found brain power Featuring hardware from @SynSenseNeuro @BrainChip_inc. #snn #spiking #spikingneuralnetwork #automotive #neuromorphic https://www.eetimes.com/cars-that-think-like-you/ https://www.eetimes.com/cars-that-think-like-you/"
X Link 2022-07-25T08:30Z [----] followers, [---] engagements
"Neuromorphic IP vendor @BrainChip_inc now has dev kits available. The company converts CNNs to spiking networks so they can run on BrainChip's Akida processor which reduces power consumption and enables on-chip learning at the edge. #ai #neuromorphic https://www.eetimes.com/brainchip-launches-event-domain-ai-inference-dev-kits/ https://www.eetimes.com/brainchip-launches-event-domain-ai-inference-dev-kits/"
X Link 2021-10-22T09:29Z [----] followers, [---] engagements
".@tenstorrent has partnered with fledgling Japanese foundry Rapidus. Why I sat down with Jim Keller for insight into the automotive AI chip market a progress report on hardware and software and its IP business which is steadily gaining momentum. https://www.eetimes.com/tenstorrent-gains-momentum-in-asia-with-rapidus-partnership/ https://www.eetimes.com/tenstorrent-gains-momentum-in-asia-with-rapidus-partnership/"
X Link 2023-11-27T09:21Z [----] followers, 13K engagements
"New favourite GIF just dropped @IanCutress With subs and easy karaoke https://t.co/KyJVIR7zsR @IanCutress With subs and easy karaoke https://t.co/KyJVIR7zsR"
X Link 2023-08-22T08:39Z [----] followers, [----] engagements
"@IanCutress With subs and easy karaoke"
X Link 2023-08-22T01:53Z [---] followers, [----] engagements
"My hot take: A sigh of relief as Lip-Bu Tan is appointed CEO of @intel https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/ https://www.eetimes.com/opinion-a-sigh-of-relief-as-lip-bu-tan-is-appointed-ceo-of-intel/"
X Link 2025-03-13T01:12Z [----] followers, [----] engagements
"HAHAHAHAHAHAHAHAHAHAHAHAHAHA. deep breath hahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahahah watching @potus more carefully recently has really changed my perspective on him (i wish i had done more of my own thinking and definitely fell in the npc trap). i'mnot going to agree with him on everything but i think he will be incredible for the country in many ways watching @potus more carefully recently has really"
X Link 2025-01-23T09:58Z [----] followers, [----] engagements
"watching @potus more carefully recently has really changed my perspective on him (i wish i had done more of my own thinking and definitely fell in the npc trap). i'mnot going to agree with him on everything but i think he will be incredible for the country in many ways"
X Link 2025-01-23T01:10Z 4.3M followers, 12.6M engagements
"Cheers @nvidia on a successful and enjoyable GTC #gtc24"
X Link 2024-03-20T22:45Z [----] followers, [----] engagements
".@RajaXg's new startup Mihira will build data centers with heterogeneous compute for rendering and AI. Mihira has licensed software from Intel's recently abandoned Project Endgame. And Raja told me he hasn't ruled out chip design going forward. https://www.eetimes.com/koduris-startup-mihira-to-license-intel-endgame-software/ https://www.eetimes.com/koduris-startup-mihira-to-license-intel-endgame-software/"
X Link 2023-10-18T13:13Z [----] followers, [----] engagements
"Stealthy Korean startup @FuriosaAI will present its tensor contraction processor at Hot Chips later today. I got the scoop Please enjoy this delightful interview with CEO June Paik: https://www.eetimes.com/furiosa-targets-llm-inference-with-second-gen-chip/ https://www.eetimes.com/furiosa-targets-llm-inference-with-second-gen-chip/"
X Link 2024-08-26T18:40Z [----] followers, [----] engagements
"I reached [----] twitter followers 🎉🎉🎉 Thanks everyone for following along and please stay tuned for more interesting stuff on AI chips as we head into [----]. I'd like to thank my mother my colleagues everyone I've ever interviewed . (🧵1/387)"
X Link 2023-11-29T10:15Z [----] followers, [----] engagements
".@nvidia has trained its NeMo #LLM on its internal documents and codebase. ChipNeMo can write Tcl scripts to control EDA tools summarize bug documentation & answer junior designers' questions. All geared towards designer productivity Bill Dally told me: https://www.eetimes.com/nvidia-trains-llm-on-chip-design/ https://www.eetimes.com/nvidia-trains-llm-on-chip-design/"
X Link 2023-10-30T16:21Z [----] followers, [----] engagements
".@Arm has updated its CSS designs for hyperscalers' custom chips. Arms vision is for a significant percentage if not the vast majority of AI inference to run on CPUs. Details on two new Neoverse cores (and preliminary Llama2-7B results) in the article: https://www.eetimes.com/arm-updates-css-designs-for-hyperscalers-custom-chips/ https://www.eetimes.com/arm-updates-css-designs-for-hyperscalers-custom-chips/"
X Link 2024-02-21T16:33Z [----] followers, [----] engagements
".@tenstorrent has raised a whopping $693 million in a round that will put any doubts about the investment community's appetite for AI chip companies aside. The company has $150 million in deals committed mainly IP with next gen chip coming in early 2025: https://www.eetimes.com/tenstorrent-raises-693-million-series-d/ https://www.eetimes.com/tenstorrent-raises-693-million-series-d/"
X Link 2024-12-02T17:03Z [----] followers, [----] engagements
"This customer win is a huge deal for @cerebras and will be worth around $900 million in the next [--] months. Importantly it validates the market for non-GPU AI accelerators at data center scale. https://www.eetimes.com/cerebras-sells-100-million-ai-supercomputer-plans-8-more/ https://www.eetimes.com/cerebras-sells-100-million-ai-supercomputer-plans-8-more/"
X Link 2023-07-20T13:53Z [----] followers, [----] engagements
".@GroqInc CEO @JonathanRoss321 is adamant his company no longer sells hardware-they are a cloud services provider and have a deal with Aramco. "The realization was you cannot sell chips as a startup its just too hard Ross told me. More in the article: https://www.eetimes.com/groq-ceo-we-no-longer-sell-hardware/ https://www.eetimes.com/groq-ceo-we-no-longer-sell-hardware/"
X Link 2024-04-08T13:51Z [----] followers, 12.3K engagements
"I spent a couple of hours at @tenstorrent chatting with @jimkxa to get updates on the company's many current projects including a new AI workstation they have developed based on first-generation Wormhole chips (a next-gen Blackhole version is coming): https://www.eetimes.com/tenstorrent-to-offer-ai-workstation-for-developers/ https://www.eetimes.com/tenstorrent-to-offer-ai-workstation-for-developers/"
X Link 2024-06-25T16:13Z [----] followers, [----] engagements
"Delighted to be invited to @Arm's HQ in Cambridge today. Story coming soon to @eetimes"
X Link 2023-02-23T15:24Z [----] followers, [----] engagements
"Called by @tenstorrent this morning to chat with Jim Keller about the company's plans RISC-V in automotive and scientific computing the economics of chiplets and much more. Stay tuned for the @eetimes article. Thanks @jimkxa"
X Link 2023-09-11T22:50Z [----] followers, [----] engagements
"Already sick of hearing people describe Arm as a "chipmaker" (in this case "shipping chips") and there are still several weeks to go on this IPO. Arm does not ship chips Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile - 96% gross margins (26% operating margins) - Aiming for 65B market cap For the full history of the company just watch this video: https://t.co/8R15GFf8Nl Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile -"
X Link 2023-08-23T13:36Z [----] followers, [----] engagements
"Here's what you need to know about the Arm IPO: - They've shipped over a quarter TRILLION chips - They have 99% market share in mobile - 96% gross margins (26% operating margins) - Aiming for 65B market cap For the full history of the company just watch this video:"
X Link 2023-08-22T16:12Z 68.8K followers, 430.8K engagements
"Always feel a little outclassed in a press conference's sea of laptops. I'm here with my pen that writes in multiple colours #GTC24"
X Link 2024-03-18T16:08Z [----] followers, [----] engagements
"Presenting The AI Hardware Show Live at #dac60 today with @IanCutress - thanks to everyone who stopped by to watch us"
X Link 2023-07-12T23:59Z [----] followers, [----] engagements
"COMING SOON"
X Link 2025-08-05T20:02Z [----] followers, [----] engagements
".@tenstorrent has launched (and open-sourced) its bare-metal software stack. This is a crucial capability for deployment of #AI models into production at scale as well as for HPC applications. I also got a demo of Falcon-40B up and running on Galaxy https://www.eetimes.com/tenstorrent-engineers-talk-open-sourced-bare-metal-stack/ https://www.eetimes.com/tenstorrent-engineers-talk-open-sourced-bare-metal-stack/"
X Link 2024-02-02T14:23Z [----] followers, [----] engagements
"I sat down with @nvidia's Ian Buck to go through some details on Blackwell including what's new in the 2nd-gen transformer engine and how Nvidia gets 30X the performance from GPUs with 2.5X the flops of Hopper. More details in the article: #ai #blackwell https://www.eetimes.com/nvidias-blackwell-gpu-offers-fp4-transformer-engine-sharp/ https://www.eetimes.com/nvidias-blackwell-gpu-offers-fp4-transformer-engine-sharp/"
X Link 2024-03-27T15:10Z [----] followers, [----] engagements
"Wondering if beginning more of my tweets with "IT'S ABOUT TO START THE BIGGEST TECH BATTLE OF OUR TIME" (and then getting the product name wrong) would get me [---] million views per tweet Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time: https://t.co/KAUpGpDcKJ Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI"
X Link 2024-12-19T11:04Z [----] followers, [----] engagements
"Nvidia just released a $249 computer that will change AI forever. It's called the Jetson Nano - a tiny device that can run AI models locally. This means AI can run without connecting to the cloud. And it's about to spark the biggest tech battle of our time:"
X Link 2024-12-18T17:00Z 44.2K followers, 4M engagements
".@SiFive and @BrainChip_inc have partnered to show SiFive's RISC-V host processor IP working alongside BrainChip's neuromorphic AI accelerator for edge AI SoCs. #ai #aiaccelerator #neuromorphic #ip #riscv https://www.eetimes.com/sifive-and-brainchip-partner-to-demo-ip-compatibility/ https://www.eetimes.com/sifive-and-brainchip-partner-to-demo-ip-compatibility/"
X Link 2022-04-21T09:06Z [----] followers, [--] engagements
".@Modular has unveiled its #AI stack w/@nvidia GPU support. The startup is tackling AI infrastructure's hard problems by building a fancy compiler and a new programming language. Is it hard Yes. But is it worth it Also yes @clattner_llvm told me: https://www.eetimes.com/modular-launches-ai-stack-adds-gpu-support/ https://www.eetimes.com/modular-launches-ai-stack-adds-gpu-support/"
X Link 2023-12-18T15:21Z [----] followers, [----] engagements
"Raja Koduri @RajaXg has launched his new GPU company @realoxmiqlabs will take on the whole of the GPU technology stack from data center orchestration to hardware IP. Coming first A Python module that enables torch.cuda code to run on other hardware: https://www.eetimes.com/koduri-unveils-gpu-hardware-ip-and-software-startup/ https://www.eetimes.com/koduri-unveils-gpu-hardware-ip-and-software-startup/"
X Link 2025-08-05T08:59Z [----] followers, [----] engagements
"Sad news from @UntetherAI as the RISC-V AI chip company closes its doors. Untether's engineering teams will join @AMD but the company's silicon and software are no more: https://www.eetimes.com/untether-ai-shuts-down-engineering-team-joins-amd/ https://www.eetimes.com/untether-ai-shuts-down-engineering-team-joins-amd/"
X Link 2025-06-09T12:03Z [----] followers, [----] engagements
"Had an interesting conversation with @nayampally_n recently about @BrainChip_inc's second-gen architecture which has hardware support for transformers and the concept of TENNs. #AI Available at or wherever you get your podcasts: https://www.eetimes.com/podcasts/brainchip-second-gen-architecture-transformers-tenns-and-stdp/ http://eetimes.com https://www.eetimes.com/podcasts/brainchip-second-gen-architecture-transformers-tenns-and-stdp/ http://eetimes.com"
X Link 2023-12-14T14:07Z [----] followers, [----] engagements
"I have a few thoughts on this article.🧵 Nvidias rivals struggle to gain ground in generative AI chip war https://t.co/o0afqQ4kqO Nvidias rivals struggle to gain ground in generative AI chip war https://t.co/o0afqQ4kqO"
X Link 2023-06-15T17:03Z [----] followers, 22.6K engagements
"Nvidias rivals struggle to gain ground in generative AI chip war https://on.ft.com/43YdiBw https://on.ft.com/43YdiBw"
X Link 2023-06-15T10:02Z 289.6K followers, 18.3K engagements
"Not thrilled so far #CogX19 #CogX stood in a queue of hundreds in pouring rain to pick up unnecessary lanyard. Why can't barcode alone be good for entry Missing Sadiq Khan speech for this"
X Link 2019-06-10T07:59Z [----] followers, [--] engagements
"@spiralout112 @intel Anyone worthy of the job would stick to Pat's vision - but because of this they would never get the job. They'll have a bean counter in the role in [--] months"
X Link 2024-12-04T10:10Z [----] followers, [----] engagements
"It's certainly been a big week for @risc_v. Read my hot take on how @intel's investment and Arm's IPO will affect the open-source ISA. #riscv #arm #ipo #isa https://www.eetimes.com/a-big-week-for-risc-v/ https://www.eetimes.com/a-big-week-for-risc-v/"
X Link 2022-02-11T10:35Z [----] followers, [--] engagements
".@AMD has multiple AI software stacks for its diverse hardware. Vamsi Boppana told me there is work ongoing to unify some elements of these stacks and that ROCm is the company's #1 priority since it is crucial to the success of MI300 & MI300X. #ai #rocm https://www.eetimes.com/rocm-is-amds-no-1-priority-exec-says/ https://www.eetimes.com/rocm-is-amds-no-1-priority-exec-says/"
X Link 2023-09-25T14:06Z [----] followers, [----] engagements
".@imec_int processor-in-memory test chip can achieve [----] TOPS/W for AI inference - yes you read that right. This tech will be available on @GLOBALFOUNDRIES 22FDX process soon. #fdsoi #tinyml #ai #MachineLearning https://www.eetimes.com/imec-develops-efficient-processor-in-memory-technique-for-glofo/ https://www.eetimes.com/imec-develops-efficient-processor-in-memory-technique-for-glofo/"
X Link 2020-07-10T08:02Z [----] followers, [--] engagements
"Good morning to everyone EXCEPT the person who commented on YouTube that they thought I was @IanCutress in a wig"
X Link 2023-04-14T15:23Z [----] followers, [----] engagements
"Happy Monday everyone"
X Link 2024-04-15T09:09Z [----] followers, [----] engagements
"Eliyan's chiplet PHY can achieve advanced packaging performance from standard packaging. Omitting the silicon interposer opens up a new world of design possibilities for chiplets especially for HBM. I spoke to @RajaXg about some of these possibilities: https://www.eetimes.com/eliyan-breaks-chiplet-memory-wall-with-standard-packaging/ https://www.eetimes.com/eliyan-breaks-chiplet-memory-wall-with-standard-packaging/"
X Link 2024-02-05T15:47Z [----] followers, [----] engagements
"Mercedes used @BrainChip_inc's neuromorphic chip in its latest concept car for the "Hey Mercedes" hot word. Mercedes said #neuromorphic chips use 5-10X less power than conventional equivalent which contributes to the car's 1000-km range. https://www.eetimes.com/mercedes-applies-neuromorphic-computing-in-ev-concept-car/ https://www.eetimes.com/mercedes-applies-neuromorphic-computing-in-ev-concept-car/"
X Link 2022-01-19T09:17Z [----] followers, [--] engagements
"Google published a paper last week on TPUv4. You may have seen in press/on twitter Google "beating Nvidia" or making unfair comparisons. Not only did they not beat Nvidia G's comparison was appropriate and paper actually had interesting stuff besides: https://www.eetimes.com/stories-claiming-google-tpuv4-beat-nvidia-greatly-overstated/ https://www.eetimes.com/stories-claiming-google-tpuv4-beat-nvidia-greatly-overstated/"
X Link 2023-04-11T12:11Z [----] followers, [----] engagements
"Is rumoured Graphcore acquisition about to close @ndahad has the scoop and I have provided some analysis in our first joint story. It was fun to speculate on what Graphcore's rumoured acquirer might do with the British startup and its team: https://www.eetimes.com/is-graphcore-deal-finally-about-to-close/ https://www.eetimes.com/is-graphcore-deal-finally-about-to-close/"
X Link 2024-05-10T08:23Z [----] followers, [----] engagements
".@SambaNovaAI released new silicon and yours truly had my hands on it on Friday New chip is designed for LLM inference and adds HBM for the first time (previously they got away with DDR only). Details in my story: #ai #llm #llminference #chatgpt #aichip https://www.eetimes.com/sambanova-adds-hbm-for-llm-inference-chip/ https://www.eetimes.com/sambanova-adds-hbm-for-llm-inference-chip/"
X Link 2023-09-19T14:08Z [----] followers, [----] engagements
"I called in at @tenstorrent to see the team's Llama3.1-70B demo up and running on one of their 8-chip systems at [--] tokens/second/user a figure the company considers a work-in-progress as it continues to optimise its software. More in the article: https://www.eetimes.com/tenstorrent-shows-off-single-user-llm-speed-for-workstation/ https://www.eetimes.com/tenstorrent-shows-off-single-user-llm-speed-for-workstation/"
X Link 2024-10-01T13:01Z [----] followers, [----] engagements
"@IanCutress They should be reading @eetimes"
X Link 2024-12-20T12:52Z [----] followers, [----] engagements
"There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: https://t.co/HOKaYA8dm5 There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: https://t.co/HOKaYA8dm5"
X Link 2024-01-11T16:51Z [----] followers, [----] engagements
"There's a first time for everything right Our first public benchmark results are here. Not only are we a part of the LLMPerf Leaderboard by @anyscalecompute we're LEADING with up to 18x faster #LLM #inference performance compared to top cloud-based providers: Get the full story on our LPU Inference Engine's performance in our latest blog and thanks to the work of our incredible team and the team over at Anyscale http://groq.link/anyscaleblog http://groq.link/anyscaleblog"
X Link 2024-01-11T16:10Z 89K followers, [----] engagements
".@latticesemi new general-purpose FPGA comes in 17k-40k logic cells a market largely ignored by the big players. #fpga #programmablelogic #latticesemi #generalpurposefpga https://www.eetimes.com/lattice-reinvents-general-purpose-fpga-offering/ https://www.eetimes.com/lattice-reinvents-general-purpose-fpga-offering/"
X Link 2020-06-25T09:14Z [----] followers, [--] engagements
"SOME PEOPLE THINK "IN-MEMORY COMPUTE" AND "COMPUTE-IN-MEMORY" IS THE SAME THING BUT OFTEN WHEN PEOPLE SAY "IN-MEMORY COMPUTE" THEY JUST MEAN COMPUTE AND MEMORY MIXED TOGETHER AT A FINE LEVEL OF GRANULARITY AS OPPOSED TO ANALOG COMPUTING WHICH IS TOTALLY DIFF-"
X Link 2022-08-19T09:27Z [----] followers, [--] engagements
"HUGE if true and its a big IF😂 on the contrary this should be a MASSIVE surprise to anyone who knows anything about Arm 🧵 This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center. This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center"
X Link 2024-05-13T13:11Z [----] followers, [----] engagements
"This should not be a surprise. Arm is a major vendor of chip IP that major tech companies depend on and this was a gap it is closing. Will be interesting to see how it scales from client to data center"
X Link 2024-05-13T01:29Z 18.7K followers, 11K engagements
"Filming in a secret location today with @IanCutress"
X Link 2024-07-30T10:57Z [----] followers, [----] engagements
"New @intel CEO Lip-Bu Tan appeared both humble and determined in a keynote at Intel Vision this week. It seems he will take pages from his Cadence playbook for Intel including refocusing on innovation and driving a more customer-centric culture: https://www.eetimes.com/intel-ceo-you-deserve-better-we-need-to-improve-and-we-will/ https://www.eetimes.com/intel-ceo-you-deserve-better-we-need-to-improve-and-we-will/"
X Link 2025-04-03T15:21Z [----] followers, [----] engagements
".@cerebras has filed the S-1 for its IPO. I read the 188-page document looking for interesting tidbits and here's what I found: https://www.eetimes.com/cerebras-ipo-paperwork-sheds-light-on-relationship-with-g42/ https://www.eetimes.com/cerebras-ipo-paperwork-sheds-light-on-relationship-with-g42/"
X Link 2024-10-01T21:22Z [----] followers, [----] engagements
"It was an honour to be asked to present Opening Remarks at @eetimes Taiwan's AI+Techcon conference. Thank you for inviting me"
X Link 2024-04-25T08:51Z [----] followers, [----] engagements
"Here comes Hot Chips Are you looking forward to it as much as I am Here are some notes on what I'm looking out for on the program this year: #hotchips34 @hotchipsorg https://www.eetimes.com/here-comes-hot-chips/ https://www.eetimes.com/here-comes-hot-chips/"
X Link 2022-08-16T09:27Z [----] followers, [--] engagements
"It's official: @graphcoreai has been acquired by @SoftBank. Masayoshi Son laid out his grand vision for artificial superintelligence last month and it now seems clear Graphcore will be a part of that plan. Details in the article: https://www.eetimes.com/ai-chip-startup-graphcore-acquired-by-softbank/ https://www.eetimes.com/ai-chip-startup-graphcore-acquired-by-softbank/"
X Link 2024-07-12T08:54Z [----] followers, [----] engagements
".@Hailo_ai's new edge chip for #GenAI has slightly less compute than the previous gen but with an optimised memory subsystem for #LLMs. It can manage [--] TOPS at INT4 or Llama2-7B at [--] tokens per second in a power envelope of 5W. #edgeai #aichips https://www.eetimes.com/hailo-debuts-edge-genai-chip-raises-120-million/ https://www.eetimes.com/hailo-debuts-edge-genai-chip-raises-120-million/"
X Link 2024-04-02T13:50Z [----] followers, [----] engagements
"New @intel products CEO @MJHolthaus dropped hints about what her strategy will be. For AI she acknowledged Intel's difficulties with Gaudi and didn't mince words about Falcon Shores ("Is it going to be wonderful No. But it is a good first step.") More: https://www.eetimes.com/intel-products-ceo-gives-hints-on-new-strategy/ https://www.eetimes.com/intel-products-ceo-gives-hints-on-new-strategy/"
X Link 2024-12-17T09:34Z [----] followers, [----] engagements
"On my recent trip to California I had the pleasure of sitting down in person with @AyarLabs CEO Charlie Wuischpard to find out how Ayar is building its ecosystem its technology roadmap and what the eventual exit plan is. Read my interview here: https://www.eetimes.com/at-ayar-labs-its-all-coming-together/ https://www.eetimes.com/at-ayar-labs-its-all-coming-together/"
X Link 2022-10-11T14:56Z [----] followers, [--] engagements
".@kinara_ai has revamped its architecture for a second-gen edge AI accelerator - big enough and powerful enough to run Llama2-7B within its 6-W power envelope. More in the article on architectural changes etc: #ai #aiaccelerator #aihardware #edgeai #llm https://www.eetimes.com/kinara-eyes-genai-with-6-w-edge-chip/ https://www.eetimes.com/kinara-eyes-genai-with-6-w-edge-chip/"
X Link 2024-01-02T13:53Z [----] followers, [----] engagements
".@graphcoreai has supercharged its IPU using TSMC's wafer-on-wafer technology. The same processor die on the same 7nm process can run 40% faster with resulting performance boost which Graphcore plans to offer at the same price as its previous-gen tech. https://www.eetimes.com/graphcore-supercharges-ipu-with-wafer-on-wafer/ https://www.eetimes.com/graphcore-supercharges-ipu-with-wafer-on-wafer/"
X Link 2022-03-03T10:16Z [----] followers, [--] engagements
"Had the pleasure of chatting with @SambaNovaAI CEO @RodrigoLiang about the company's [---] trillion-parameter #ai model Samba-1. SambaNova's answer to #GPT-4 is a composition of experts not mixture of experts which is better for the enterprise he said: https://www.eetimes.com/sambanova-trains-trillion-parameter-model-to-take-on-gpt-4/ https://www.eetimes.com/sambanova-trains-trillion-parameter-model-to-take-on-gpt-4/"
X Link 2024-03-06T13:17Z [----] followers, [----] engagements
"And that's a wrap The final podcast for the AI Hardware Show [----] is out Stay tuned for more in [----] 😉 Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss https://t.co/MwXbjFp0GU with @sallywf Big thanks to @VentanaMicro #sponsored https://t.co/VRv8MZC4YP Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss https://t.co/MwXbjFp0GU with"
X Link 2024-02-08T12:24Z [----] followers, [----] engagements
"Our final podcast of the first series It's our longest one ever. Well when you speak about @Intel @AMD @NVIDIA and all the big CPU+GPU accelerators they have there's a lot to discuss with @sallywf Big thanks to @VentanaMicro #sponsored https://youtu.be/keAZy__gsAo https://youtu.be/keAZy__gsAo"
X Link 2024-02-08T12:00Z 51.3K followers, [----] engagements
"The Snapdragon X Elite - @Qualcomm's new laptop chip with CPU cores designed at Nuvia - has [--] TOPS (INT8) total AI performance across CPU GPU and Hexagon NPU. Kevin @Krewell reports from Snapdragon Summit in Hawaii for EE Times. At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://t.co/r4wORxBUsL https://t.co/qjY1ZdJmXz At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://t.co/r4wORxBUsL https://t.co/qjY1ZdJmXz"
X Link 2023-10-30T13:10Z [----] followers, [----] engagements
"At its annual gathering in Hawaii Qualcomm showed off its PC-ready Snapdragon X Elite with powerful #AI acceleration. https://www.eetimes.com/qualcomm-takes-on-amd-apple-intel-with-snapdragon-x-elite/ https://www.eetimes.com/qualcomm-takes-on-amd-apple-intel-with-snapdragon-x-elite/"
X Link 2023-10-30T13:00Z 36.4K followers, [----] engagements
".@tenstorrent has licensed CPU IO and memory chiplet IP to Japanese research institute LSTC who are building their own chiplet-based edge AI chip. Tenstorrent chief architect Wei-Han Lien told me he is most concerned about compatibility. More in article: https://www.eetimes.com/tenstorrent-licenses-chiplet-designs-to-japanese-institute/ https://www.eetimes.com/tenstorrent-licenses-chiplet-designs-to-japanese-institute/"
X Link 2024-03-01T14:23Z [----] followers, [----] engagements
"AI chipmakers @cerebras @SambaNovaAI join @GroqInc in offering cloud inference APIs for specialized hardware. All three claim big single-user speed advantages versus GPU-based clouds but what's the motivation for this apparent change in strategy https://www.eetimes.com/token-wars-heats-up-as-cerebras-and-sambanova-enter-the-fray/ https://www.eetimes.com/token-wars-heats-up-as-cerebras-and-sambanova-enter-the-fray/"
X Link 2024-09-10T19:35Z [----] followers, [----] engagements
"What about optimal sizing of the arrays for each model/layer Dataflow/datapath optimization at the architecture level Balance between memory and compute for prefill and decode stages Hardware support for other transformer operators (2/x)"
X Link 2025-07-24T13:28Z [----] followers, [----] engagements
"@IanCutress We used to have another term for a programmable ASIC - ASSP - but you don't see this term any more. (In the industry we used to say "ASIC" for a fixed-function chip that's built for one customer only)"
X Link 2025-02-28T15:16Z [----] followers, [----] engagements
"The legendary journalist that is @junkoyoshida is retiring today after more than three decades with @eetimes - au revoir Junko and thank you for everything Junko Paints Life with VibrantColors https://t.co/0ZZnd08Lq8 Junko Paints Life with VibrantColors https://t.co/0ZZnd08Lq8"
X Link 2021-05-14T08:46Z [----] followers, [--] engagements
"Junko Paints Life with VibrantColors https://www.eetimes.eu/junko-paints-life-with-vibrant-colors/ https://www.eetimes.eu/junko-paints-life-with-vibrant-colors/"
X Link 2021-05-14T08:36Z [---] followers, [--] engagements
"I called by @tenstorrent for a chat with CEO Jim Keller @jimkxa about his commitment to open-source the training systems he's building and why he wants to do the opposite of whatever Jensen does: https://www.eetimes.com/jim-keller-whatever-nvidia-does-well-do-the-opposite https://www.eetimes.com/jim-keller-whatever-nvidia-does-well-do-the-opposite"
X Link 2025-06-05T19:11Z [----] followers, [----] engagements
"@IanCutress Love the thumbnail with "New Intel CEO" plastered over you. Makes it look like Gelsinger is handing the reins to Tech Tech Potato 😅"
X Link 2024-12-02T22:01Z [----] followers, [----] engagements
"Delighted to speak with @petewarden about his new company Useful Sensors. The startup aims makes AI-enabled sensors "that just work". Their first product a 20x20mm person detection board is out now #ai #ml #tinyml #aisensor #sensormodule #aiot https://www.eetimes.com/pete-wardens-startup-puts-ai-in-the-sensor/ https://www.eetimes.com/pete-wardens-startup-puts-ai-in-the-sensor/"
X Link 2022-10-20T07:32Z [----] followers, [--] engagements
"In other words I think there are several dimensions that could be specialised on for transformers beyond generic GEMM capability. This is instinctive to people who understand the concept of ASICs outside of data center/AI (3/x)"
X Link 2025-07-24T13:28Z [----] followers, [----] engagements
"@EnerTuition @intel Strongly disagree. Gelsinger inherited years of mis-management which is still playing out. Turning a ship that size around will take more years"
X Link 2024-12-04T20:13Z [----] followers, [---] engagements
"This is huge Congratulations @clattner_llvm and @Modular team Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇 https://t.co/mnHqtM8s0Y Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇 https://t.co/mnHqtM8s0Y"
X Link 2025-06-10T18:23Z [----] followers, [----] engagements
"Were one step closer to truly scalable hardware-independent AI. Todays premiere introduced major updates that simplify deployment speed up development and bring powerful performance to any GPU. Here's what dropped 👇"
X Link 2025-06-10T17:23Z 21K followers, 118.8K engagements
"I will be moderating a panel at next week's AI Hardware Summit in London The panel will take on hardware challenges for AI in the data center and at the edge and our hardware experts will be available to answer your questions. #ai #aihardware #aihardwaresummit #event #aichips"
X Link 2024-06-11T08:25Z [----] followers, [----] engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing