https://minutesof.com/on/deepseek/ Everyone on the record about deepseek: 21 verbatim quotes from 5 people, January 2025 to June 2026, in one chronology, each with a timestamp and a link to… en Sun, 30 Aug 2026 22:32:55 +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.” — Dylan Patel, Sequoia Capital Tue, 30 Jun 2026 12:00:25 +0000 deepseek https://minutesof.com/q/0747bd60-bb55-46d1-98e7-3331e41198ed/ https://minutesof.com/q/0747bd60-bb55-46d1-98e7-3331e41198ed/ “No. For sure. And in fact, I I hear disturbing things all the time about the Chinese stack making great inroads in The Middle East and in the Southern Hemisphere, etcetera, because, frankly, they're working really quickly to put their chips and their models, their open source models, Deepsea, Kimi k two, Quinn, etcetera, and exporting those to the world.” — Brad Gerstner, Altimeter Capital Tue, 10 Feb 2026 21:40:23 +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.” — Dylan Patel, 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.” — Dylan Patel, No Priors: AI, Machine Learning, Tech, & Startups Thu, 14 Aug 2025 10:01:35 +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.” — Dylan Patel, Alex Kantrowitz Wed, 23 Apr 2025 16:30:06 +0000 deepseek https://minutesof.com/q/a64a56f5-df99-4993-bd99-c6e00a54110d/ https://minutesof.com/q/a64a56f5-df99-4993-bd99-c6e00a54110d/ “I think what was really surprising was that it was a Chinese company for the first time. Right? Because Google and and OpenAI and Anthropic and Meta have all traded blows.” — Dylan Patel, Alex Kantrowitz Wed, 23 Apr 2025 16:30:06 +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.” — Dylan Patel, 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?” — Dylan Patel, Special Competitive Studies Project Thu, 13 Mar 2025 16:42:15 +0000 deepseek https://minutesof.com/q/2fad991a-2404-440c-9dd6-13bf493b7e2b/ https://minutesof.com/q/2fad991a-2404-440c-9dd6-13bf493b7e2b/ “forty eight hours after r one was posted, they had 500 variants on Hugging Face. And today I pinged them this morning before we started, they're up to 1,300.” — Bill Gurley, BG2 Pod Wed, 05 Feb 2025 23:13:28 +0000 deepseek https://minutesof.com/q/7915177a-c38e-49f6-8c41-f1704a1047cb/ https://minutesof.com/q/7915177a-c38e-49f6-8c41-f1704a1047cb/ “On top of that, we learned, you know, they are massively compute constrained right now. So you've seen some tweets about this, people that are, you know, getting server delay and all this stuff” — Brad Gerstner, BG2 Pod Wed, 05 Feb 2025 23:13:28 +0000 deepseek https://minutesof.com/q/f2fcae4d-e3e9-474b-9572-95f3a2e55fad/ https://minutesof.com/q/f2fcae4d-e3e9-474b-9572-95f3a2e55fad/ “if you look at the models that are apples to apples on the API right now, that that DeepSeek's pricing about one twentieth of OpenAI.” — Bill Gurley, BG2 Pod Wed, 05 Feb 2025 23:13:28 +0000 deepseek https://minutesof.com/q/dd419e62-f500-4370-8d57-c84a06675948/ https://minutesof.com/q/dd419e62-f500-4370-8d57-c84a06675948/ “They were able to do that because they figured out a way to separate the parameters and work on things with smaller parameter counts faster, which no one else had done before.” — Bill Gurley, BG2 Pod Wed, 05 Feb 2025 23:13:28 +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.” — Dylan Patel, 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.” — Dylan Patel, Lex Fridman Mon, 03 Feb 2025 00:12:13 +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.” — Dylan Patel, Lex Fridman Mon, 03 Feb 2025 00:12:13 +0000 deepseek https://minutesof.com/q/e157002e-0548-4018-aec6-14d72fa3e027/ https://minutesof.com/q/e157002e-0548-4018-aec6-14d72fa3e027/ “When you would ask it, who are you? Like what model are you? Five out of eight times, v three would tell you that it was ChatGPT four.” — David Sacks, All-In Podcast Fri, 31 Jan 2025 21:52:00 +0000 deepseek https://minutesof.com/q/c35e5045-0094-4d11-8d15-94e56f582513/ https://minutesof.com/q/c35e5045-0094-4d11-8d15-94e56f582513/ “you add up the the cost of a compute cluster with 50,000 plus hoppers and it's gonna be over $1,000,000,000.” — David Sacks, All-In Podcast Fri, 31 Jan 2025 21:52:00 +0000 deepseek https://minutesof.com/q/a4ede3f9-51af-4927-b48a-d925bdf7fed7/ https://minutesof.com/q/a4ede3f9-51af-4927-b48a-d925bdf7fed7/ “Dylan Patel, who's leading semiconductor analyst, has estimated that DeepSeac has about 50,000 hoppers. And specifically, he said they have about 10,000 h one hundreds, They have 10,000 h eight hundreds and 30,000 h twenties.” — David Sacks, All-In Podcast Fri, 31 Jan 2025 21:52:00 +0000 deepseek https://minutesof.com/q/73cecf18-6302-4ffe-ba6b-32c73ed695f1/ https://minutesof.com/q/73cecf18-6302-4ffe-ba6b-32c73ed695f1/ “On this one, I'm with Palmer Luckey and Brad Gerstner and others, and I think this has been pretty much corroborated by everyone I've talked to that that number should be debunked.” — David Sacks, All-In Podcast Fri, 31 Jan 2025 21:52:00 +0000 deepseek https://minutesof.com/q/62b103f6-6774-4c42-9053-300c3f5ea97c/ https://minutesof.com/q/62b103f6-6774-4c42-9053-300c3f5ea97c/ “if you had said to people a few weeks ago that the second company to release a reasoning model along the lines of o one would be a Chinese company, I think people would have been surprised by that.” — David Sacks, All-In Podcast Fri, 31 Jan 2025 21:52:00 +0000 deepseek https://minutesof.com/q/5fe5835f-3bc6-4b11-b3f5-6a03e58df6ec/ https://minutesof.com/q/5fe5835f-3bc6-4b11-b3f5-6a03e58df6ec/ “I think you will see the Frontier Labs stop releasing their leading edge models to prevent knowledge distillation and their IP effectively being stolen.” — Gavin Baker, All-In Podcast Sat, 04 Jan 2025 00:23:00 +0000 deepseek