LunarCrush LLM | creator/twitter::929211084/posts

[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.]

[@iamtrask](/creator/twitter/iamtrask)
"To turn this post into practical advise. 1) AI researchers could pivot to form a boutique investment analysis firm which only does one thing. translate SOTA research into GPU demand forecasts 2) firms holding large NVIDIA positions should either hire (1) or an in-house AI researcher / intern to keep them up-to-date on efficiency breakthroughs 3) it's probably worth hedging some between NVIDIA (who would lose with an efficiency gain) and AI companies (who would gain with an efficiency gain)"  
[X Link](https://x.com/iamtrask/status/1980019973620900247) [@iamtrask](/creator/x/iamtrask) 2025-10-19T21:15Z 79K followers, XXX engagements


"this is a good example of why detecting AI generated content is an unsolvable task also why deepfake detection is impossible the information bottleneck is too great in all cases a human & an AI can generate the same text (i wrote that tweet. i love emdashes have for years)"  
[X Link](https://x.com/iamtrask/status/1967273060614250669) [@iamtrask](/creator/x/iamtrask) 2025-09-14T17:03Z 78.9K followers, 37.2K engagements


"I've just drafted a new blogpost "GPU demand is (1Mx) distorted by efficiency problems which are being solved" Mid-2024 Andrej Karpathy trained GPT-2 for $XX. Six months later Andreessen Horowitz reported LLM costs falling 10x annually. Two months after that DeepSeek shocked markets with radical reductions in training and inference requirements. For AI researchers this is all good news. For executives policymakers and investors forecasting GPU demand. less so. Many were caught off guard. The problem isnt that executives / policymakers / investors lacked access to information per se its that"  
[X Link](https://x.com/iamtrask/status/1979996110761660793) [@iamtrask](/creator/x/iamtrask) 2025-10-19T19:40Z 79K followers, 22K engagements


"@sotirov Fwiw - I went out of my way to craft a "library" analogy which tried to make this intuitive. Command-F for the word "librar" and I hope it helps 3"  
[X Link](https://x.com/iamtrask/status/1980020367545581625) [@iamtrask](/creator/x/iamtrask) 2025-10-19T21:16Z 79K followers, XXX engagements

[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.]

@iamtrask "To turn this post into practical advise. 1) AI researchers could pivot to form a boutique investment analysis firm which only does one thing. translate SOTA research into GPU demand forecasts 2) firms holding large NVIDIA positions should either hire (1) or an in-house AI researcher / intern to keep them up-to-date on efficiency breakthroughs 3) it's probably worth hedging some between NVIDIA (who would lose with an efficiency gain) and AI companies (who would gain with an efficiency gain)"
X Link @iamtrask 2025-10-19T21:15Z 79K followers, XXX engagements

"this is a good example of why detecting AI generated content is an unsolvable task also why deepfake detection is impossible the information bottleneck is too great in all cases a human & an AI can generate the same text (i wrote that tweet. i love emdashes have for years)"
X Link @iamtrask 2025-09-14T17:03Z 78.9K followers, 37.2K engagements

"I've just drafted a new blogpost "GPU demand is (1Mx) distorted by efficiency problems which are being solved" Mid-2024 Andrej Karpathy trained GPT-2 for $XX. Six months later Andreessen Horowitz reported LLM costs falling 10x annually. Two months after that DeepSeek shocked markets with radical reductions in training and inference requirements. For AI researchers this is all good news. For executives policymakers and investors forecasting GPU demand. less so. Many were caught off guard. The problem isnt that executives / policymakers / investors lacked access to information per se its that"
X Link @iamtrask 2025-10-19T19:40Z 79K followers, 22K engagements

"@sotirov Fwiw - I went out of my way to craft a "library" analogy which tried to make this intuitive. Command-F for the word "librar" and I hope it helps 3"
X Link @iamtrask 2025-10-19T21:16Z 79K followers, XXX engagements