https://minutesof.com/dylan-patel/on/deepseek/ Everything Dylan Patel has said on deepseek: 9 verbatim quotes between February 2025 and August 2026, each with a timestamp and a link to the recording it… en Sun, 30 Aug 2026 16:22:17 +0000 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 deepseek https://minutesof.com/q/b5b15eca-a046-4a24-8309-0cfa9d92ceec/ https://minutesof.com/q/b5b15eca-a046-4a24-8309-0cfa9d92ceec/ “One example is one that DeepSeek open sourced over December of last yearJanuary, February of this year, where a single replica of inference for a single model is going to be like 140 GPUs.” — Clockwork Fri, 21 Nov 2025 17:18:51 +0000 deepseek https://minutesof.com/q/13cb1933-7756-463f-ac46-7f8cd066bee0/ https://minutesof.com/q/13cb1933-7756-463f-ac46-7f8cd066bee0/ “That's over $10,000,000 of hardware, and then that's just one replica, then you'll have a lot of replicas and you share the caching servers between them.” — No Priors: AI, Machine Learning, Tech, & Startups Thu, 14 Aug 2025 10:01:35 +0000 deepseek https://minutesof.com/q/c28a19fc-d8f5-4e4a-bc59-e7dabb126c8c/ https://minutesof.com/q/c28a19fc-d8f5-4e4a-bc59-e7dabb126c8c/ “when we looked at g p d three, the cost fell 1,200 x from g p d three's initial cost to what you can get Lama 3.23 b today.” — Alex Kantrowitz Wed, 23 Apr 2025 16:30:06 +0000 deepseek https://minutesof.com/q/42277ecd-5d1b-4270-a90f-11eb7f3a67b8/ https://minutesof.com/q/42277ecd-5d1b-4270-a90f-11eb7f3a67b8/ “It's not unexpected. Right? Like, this is actually within the trend line of what happened with GPT three is happening to GPT four level quality with DeepSeq.” — Alex Kantrowitz Wed, 23 Apr 2025 16:30:06 +0000 deepseek https://minutesof.com/q/c235c852-3da5-495d-8118-6bceb59270eb/ https://minutesof.com/q/c235c852-3da5-495d-8118-6bceb59270eb/ “The estimates that we have is, so first of all, ads in China, they say they have tens of thousands of GPUs for researchers, right?” — Special Competitive Studies Project Thu, 13 Mar 2025 16:42:15 +0000 deepseek https://minutesof.com/q/02fd4a40-e301-43f7-be83-33801b8010a9/ https://minutesof.com/q/02fd4a40-e301-43f7-be83-33801b8010a9/ “This reasoning model has a lot of overlapping training steps to DeepSeek v three, and it's confusing that you have a base model called v three that you do something to to get a chat model, and then you do some different things to get a reasoning model.” — Lex Fridman Mon, 03 Feb 2025 00:12:13 +0000 deepseek https://minutesof.com/q/99d651cc-75d0-4621-9e78-21bd90eeb8aa/ https://minutesof.com/q/99d651cc-75d0-4621-9e78-21bd90eeb8aa/ “For example, on their to get highly efficient training, they're making modifications at or below the CUDA layer for NVIDIA chips.” — Lex Fridman Mon, 03 Feb 2025 00:12:13 +0000 deepseek https://minutesof.com/q/fab5c10d-e2cd-4e52-b056-9586751d1efb/ https://minutesof.com/q/fab5c10d-e2cd-4e52-b056-9586751d1efb/ “this type of change can be big, it can be small, but they add up over time.” — Lex Fridman Mon, 03 Feb 2025 00:12:13 +0000 deepseek