[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

[@ResearchQf](/creator/twitter/ResearchQf)
"If some of this talk about increasing orders from @OpenAI partners (plus higher CSP4 capex and some incremental sovereigns into 2026) as well as continued supply improvements are real . Should hear more soon if valid. Hon Hai AI servers were slightly disappointing in June but they should & need to improve significantly through 2H. Have to deconvolve large companies like Hon Hai more carefully. Apr up big MoM wasn't about AI. Neither was May down MoM"  
[X Link](https://x.com/ResearchQf/status/1952367118617379233) [@ResearchQf](/creator/x/ResearchQf) 2025-08-04T13:52Z 16.7K followers, 31.5K engagements


"1) Hon Hai takes baton from smaller ODMs on schedule. If you missed earnings call HH guided AI servers as much as XXX% Q/Q or potentially 4x Q/Q. Nomura after visiting HH on 8/15 during tech tour: HH observes its CSP clients are "desperate" to get more GB200s and would not want to wait for GB300. GB300 ramp smoother than GB200 which isn't new. This is beyond July monthlies and certain temporary GB200 to GB300 transitions at smaller ODMs where demand is not an issue. Orders were stronger everywhere. This has yet to be fully absorbed and reported by analysts but that's already rolling this"  
[X Link](https://x.com/ResearchQf/status/1957196875825586362) [@ResearchQf](/creator/x/ResearchQf) 2025-08-17T21:44Z 16.7K followers, 44.6K engagements


"1) Mainstream DRAM and NAND spot prices were up XX% last night alone. DDR5 spot just doubled. I've never seen this before. $MU and SK Hynix almost doubled since local bearish peak for DRAM & HBM in Aug. Large parts of buy & sell-side as well as Fintwit. Do math for "commodity" DRAM margins. Assume spot goes sideways and contract (which lags spot) catches up. If XXX price/50 cost = XX% GM. XXX price/50 cost = XX% GM (). It's not surprising every time I turn around any incremental price negotiations for HBM3e and HBM4 sound better. Have discussed varying fungibility timeframes between DDR and"  
[X Link](https://x.com/ResearchQf/status/1980402502391812351) [@ResearchQf](/creator/x/ResearchQf) 2025-10-20T22:35Z 16.7K followers, 92K engagements


"3) $SNDK up 2.8x & Kioxia up 2.7x since this post. I'd mentioned that if one had payed close attention sharp improvement in NAND was clear earlier. X of top X stocks YTD in S&P XXX is storage or memory. If SNDK had spun off earlier would be X of top 5"  
[X Link](https://x.com/ResearchQf/status/1980402506040823997) [@ResearchQf](/creator/x/ResearchQf) 2025-10-20T22:35Z 16.7K followers, 6366 engagements


"4) $STX & $WDC kept going up and more than doubled since first "overbought" with 80's RSI. They are finally correcting after being more extended than ever in history. Don't know if today tomorrow next W or Q but they will likely be buys or adds soon"  
[X Link](https://x.com/ResearchQf/status/1980402507487801806) [@ResearchQf](/creator/x/ResearchQf) 2025-10-20T22:35Z 16.7K followers, 5920 engagements


"1) $NVDA is more likely have bigger upside(s) than usual by Jan Q. Based on a pretty broad set of data points. It's also not impossible that Oct Q could also be quite decent. Given NVL run rate by Q4 one week in timing could mean over +/- $3B. But timing won't matter if NVDA DC rev has larger than usual midpoint for combined Oct + Jan Q vs $101B DC cons for that period. This is independent bottoms up & top down build. I.e. most sell side models look more conservative than typical. There's always a spread could be wrong but my midpoint is higher than normal vs expectations. Subject to quick"  
[X Link](https://x.com/ResearchQf/status/1963223293021901104) [@ResearchQf](/creator/x/ResearchQf) 2025-09-03T12:51Z 16.6K followers, 21.1K engagements


"1) $COHR risk/reward improved IT and I rotated a part of LITE back into COHR. It's still a smaller position. $LITE almost tripled past year and outperformed COHR 2.2x. For good reasons. LITE was and still is better positioned across a range of current and future interconnect technologies & products. As a result LITE's forward earnings cons rose 30%. COHR's forward cons was almost flat after last X Q's. But based on data today COHR numbers also have greater upside vs downside while valuation compressed. LITE also became much too big in the portfolio but that's portfolio dependent"  
[X Link](https://x.com/ResearchQf/status/1968799164902367237) [@ResearchQf](/creator/x/ResearchQf) 2025-09-18T22:07Z 16.6K followers, 28.8K engagements


"3) To gauge TAM count copper vs optical in your server. Will take long time but AI accelerating change. MRVL chart. Topology & XPU definition dependent but no wonder 800G/1.6T datapoints improved so much. Starting to feel this $COHR $LITE $MXL $MTSI etc"  
[X Link](https://x.com/ResearchQf/status/1806783080759210350) [@ResearchQf](/creator/x/ResearchQf) 2024-06-28T20:13Z 16.7K followers, 6663 engagements


"Doubt this was a main driver for the selloff today. However most reporters also aren't financial analysts and @theinformation has been mixed. $125M gross profit on $900M revenue or a disastrous XX% gross margin E.g. $ORCL just began filling in its Abilene DC. Try back of envelopes but depending on unknown depreciation & fixed cost overheads possible GM for massive DCs that are quickly built and are just starting to be utilized has a . very wide range. There are assumptions under which it wouldn't be shocking for initial GM to be negative at very low utilization for a facility of this"  
[X Link](https://x.com/ResearchQf/status/1975625672665059512) [@ResearchQf](/creator/x/ResearchQf) 2025-10-07T18:14Z 16.7K followers, 28.5K engagements


"1) Traveling the west coast. Here for high level AI overviews and application anecdotes rather than infra. One anecdote. Several decent musicians I know were shocked by Suno music we generated real time. One texted "This is shocking This will shake up the music industry industry" Wait til 2027. I love music but have little talent but we created what almost seemed like new genres by blending different eras regions and classes of music"  
[X Link](https://x.com/ResearchQf/status/1981770860165050422) [@ResearchQf](/creator/x/ResearchQf) 2025-10-24T17:12Z 16.7K followers, 27K engagements


"1) Trump Xi meeting could be very important. 2nd day also seems to have gone well. Incremental developments since Madrid and recent saber rattling appear to include "substantial" soybean purchases what could be more proactive mentions of tech export controls & fentanyl as well as final TikTok details. Importantly both Hi Lefeng and Li Chenggang also made positive comments Sunday"  
[X Link](https://x.com/ResearchQf/status/1982460212465274990) [@ResearchQf](/creator/x/ResearchQf) 2025-10-26T14:52Z 16.7K followers, 46.4K engagements