The Minutes of Dylan Patel on hardware https://minutesof.com/dylan-patel/on/hardware/ Everything Dylan Patel has said on hardware: 5 verbatim quotes between November 2024 and August 2026, each with a timestamp and a link to the recording it… en Sun, 30 Aug 2026 16:22:19 +0000 Patel argues silicon supply can be optimized for either high throughput or high interactivity use cases. https://minutesof.com/q/6a3bd7b5-c74b-46a6-9dd5-8db72aa9576f/ https://minutesof.com/q/6a3bd7b5-c74b-46a6-9dd5-8db72aa9576f/ “Supply of silicon can go many ways. You can either leverage it to high throughput things or high interactivity things.” — SemiAnalysis Mon, 17 Aug 2026 15:00:06 +0000 hardware Patel says InferenceX now has over $80 million of compute across multiple vendors. https://minutesof.com/q/eb8b0b35-dc54-4b31-af11-cfea01d719e6/ https://minutesof.com/q/eb8b0b35-dc54-4b31-af11-cfea01d719e6/ “we have over $80,000,000 of compute GPUs from NVIDIA AMD, TPUs from, Google, Tranium from Amazon, and we run this benchmark constantly on the newest inference engine, newest drivers, newest, PyTorch version,” — RAISE Summit Thu, 16 Jul 2026 16:44:57 +0000 hardware Patel claims hardware improved 30x from Hopper to Blackwell for DeepSeek on optimized deployments. https://minutesof.com/q/1aeaaf0c-1080-4d21-a79c-2bc3d127baf5/ https://minutesof.com/q/1aeaaf0c-1080-4d21-a79c-2bc3d127baf5/ “from Hopper to Blackwell, which is all we've had over the last three years, roughly 30x improvement on DeepSeek, on the most optimized deployment, which you can see on InferenceX there's about a 30x improvement.” — Sequoia Capital Tue, 30 Jun 2026 12:00:25 +0000 hardware Patel explains Bumpgate issue where thermal expansion differences caused GPU solder ball failures in Apple lap https://minutesof.com/q/0a5f3bdf-1c4b-45e3-ae44-c701f4cf2fee/ https://minutesof.com/q/0a5f3bdf-1c4b-45e3-ae44-c701f4cf2fee/ “And what ended up happening is because of that different rate of expansion, the solder balls connecting the chip and the board would crack.” — Matthew Berman Mon, 30 Jun 2025 17:27:19 +0000 hardware Patel claims four to five percent of NVIDIA GPUs fail during initial burn-in testing within two weeks. https://minutesof.com/q/36c00658-21e4-44c9-9919-c879522edbb7/ https://minutesof.com/q/36c00658-21e4-44c9-9919-c879522edbb7/ “Buy a good chip from NVIDIA, and then it ends up failing like constantly, right?” — Scaling Intelligence Tue, 12 Nov 2024 04:17:00 +0000 hardware