The Minutes of Gavin Baker on inference https://minutesof.com/gavin-baker/on/inference/ Everything Gavin Baker has said on inference: 38 verbatim quotes between November 2019 and August 2026, each with a timestamp and a link to the recording… en Sun, 30 Aug 2026 16:21:41 +0000 Baker argues contracted compute trades at massive discount to spot, repricing will accelerate cash flows and a https://minutesof.com/q/c85f4400-5d03-4327-a817-cfa56c88e0a0/ https://minutesof.com/q/c85f4400-5d03-4327-a817-cfa56c88e0a0/ “And so, essentially, you have the contracted base of installed compute trading at a massive discount to the current spot market.” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 inference Baker reports GPU rental prices doubled from mid-$2 to nearly $4 per hour over seven months for identical clus https://minutesof.com/q/2f00fe00-eb15-4fb6-83b8-ad9eee2d466d/ https://minutesof.com/q/2f00fe00-eb15-4fb6-83b8-ad9eee2d466d/ “And they had rented a cluster of several thousand black wells, and we'll just call it somewhere in the mid $2 per GPU hour.” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 inference Baker reports inference cloud expects to pay 100% more for Blackwells when contracts expire, showing hyperscal https://minutesof.com/q/443a6457-4940-4dff-9022-a9659a14fe37/ https://minutesof.com/q/443a6457-4940-4dff-9022-a9659a14fe37/ “They went on a podcast, and they essentially said, we are planning to pay 100% more for Blackwell's when our contract expires. And that just means that essentially all the hyperscalers are under earning.” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 inference Baker cites analysis showing compute margins, quantity, and inference margins all rising simultaneously, drivi https://minutesof.com/q/273fa3f1-13d3-4f0d-9774-ac10ff1c4250/ https://minutesof.com/q/273fa3f1-13d3-4f0d-9774-ac10ff1c4250/ “The amount of compute is going up and inference margins going up. And if you multiply those three, that's how you're getting this crazy acceleration into some of the labs plus open source,” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 inference Baker estimates SpaceX monetizes compute at $50B per gigawatt versus $73B consensus, with Grok and Cursor hitt https://minutesof.com/q/898427d9-5d8c-40af-9feb-9256d6c3c56e/ https://minutesof.com/q/898427d9-5d8c-40af-9feb-9256d6c3c56e/ “And they're monetizing at something like 50,000,000,000 a gig and consensus estimates for next year are 73,000,000,000. So forget Starlink v three, forget Starlink direct to sell, Grok 4.5 and Cursor.” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 inference Baker believes Anthropic is likely already generating cash or will start this year. https://minutesof.com/q/ec5a9c7e-99d9-4536-8535-fbba88740844/ https://minutesof.com/q/ec5a9c7e-99d9-4536-8535-fbba88740844/ “I think Anthropic probably starts generating cash this year if they are not already generating cash, which I think is probably the case.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker argues America will consume all available compute, reducing edge AI bear case concerns. https://minutesof.com/q/c36d1b7b-a557-4ecb-b855-6610596f9fe3/ https://minutesof.com/q/c36d1b7b-a557-4ecb-b855-6610596f9fe3/ “And I just think the same is true of compute. It's why I'm probably less worried about like an edge AI bear case than I was.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker explains harness engineering matters significantly and harnesses are increasingly co-developed with mode https://minutesof.com/q/7991d993-a149-4684-8692-12d7dc932f1e/ https://minutesof.com/q/7991d993-a149-4684-8692-12d7dc932f1e/ “And it turns out that harness engineering is not as important as the model, but it really matters. And these harnesses in these models are increasingly being co developed.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker says understanding frontier AI now requires enterprise usage-based plans, not consumer subscriptions, du https://minutesof.com/q/7b49081d-b3f9-445d-aba4-d290d5374ebd/ https://minutesof.com/q/7b49081d-b3f9-445d-aba4-d290d5374ebd/ “To understand what Frontier AI is capable of today, even for a non coding use case, need to have Cloud Code or Codex five point Codex. And you need to be on an enterprise plan.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker says AI shifting from flat pricing to usage-based is extremely bullish as people consume more AI. https://minutesof.com/q/4b52026f-d927-4a34-90b9-14245a9cc937/ https://minutesof.com/q/4b52026f-d927-4a34-90b9-14245a9cc937/ “AI is just shifting from all you can eat to pay by the drink. Then it turns out people really like to talk to their friends long distance.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker predicts OpenAI and Anthropic will exceed $200B ARR this year due to shift to usage-based pricing. https://minutesof.com/q/60e611b0-5da1-46dd-88cc-3277d267722f/ https://minutesof.com/q/60e611b0-5da1-46dd-88cc-3277d267722f/ “So I think the shift to usage based pricing is probably why you will see OpenAI and Anthropic exceed well over $200,000,000,000 in ARR this year.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker predicts OpenAI and Anthropic will exceed $200B ARR this year due to usage-based pricing shift. https://minutesof.com/q/e54157aa-cef1-4bd4-9d22-a8b4df8c0d13/ https://minutesof.com/q/e54157aa-cef1-4bd4-9d22-a8b4df8c0d13/ “I think the shift to usage based pricing is probably why you will see OpenAI and Anthropic exceed well over $200,000,000,000 in ARR this year.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker predicts GPUs will have 10-15 year useful lives due to prefill-inference disaggregation, extending older https://minutesof.com/q/ca1f41f5-e05d-4029-a1db-504f8b4e433d/ https://minutesof.com/q/ca1f41f5-e05d-4029-a1db-504f8b4e433d/ “The disaggregation of inference means that I think these GPUs are going to have ten or fifteen year lives.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker argues GPU useful lives will extend to 10-15 years due to inference disaggregation, contradicting AI ske https://minutesof.com/q/d42a1b86-60fc-431a-8a6b-892fc8405b51/ https://minutesof.com/q/d42a1b86-60fc-431a-8a6b-892fc8405b51/ “The disaggregation of inference means that I think these GPUs are going to have ten or fifteen year lives. The AI skeptics are like, oh, these companies are all cooking their books.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker argues GPU useful lives will extend to 10-15 years due to prefill/decode disaggregation, not 1-2 years. https://minutesof.com/q/d97e5674-e5aa-447d-9755-f683cbf66ab1/ https://minutesof.com/q/d97e5674-e5aa-447d-9755-f683cbf66ab1/ “The useful life of GPU is only a year or two. The useful life of CPU is only four years because the rapid technological change.” — Invest Like the Best Wed, 20 May 2026 08:00:00 +0000 inference Baker says AI models shifting to usage-based pricing with overage reveals no ceiling on spending yet. https://minutesof.com/q/eb16479a-ee2b-4d2c-9779-1e7eb5762367/ https://minutesof.com/q/eb16479a-ee2b-4d2c-9779-1e7eb5762367/ “We're just moving from these all you can eat plans to usage based plans with overage, where those usage tokens cost a lot more, and we're finding out that there's we're nowhere near the amount of, you know, people ceiling price for how much they'll spend.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker emphasizes only 0.1% of the world uses AI models properly yet there's massive shortage despite trillions https://minutesof.com/q/20c46b63-3663-4f9f-9324-17a82821dfcb/ https://minutesof.com/q/20c46b63-3663-4f9f-9324-17a82821dfcb/ “And we're in an insane shortage despite spending cumulatively trillions of dollars. What happens when 5% of the world's population is using these models the way the cutting edge 10 basis points are?” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference What happens when 5% of the world's population is using these models the way the cutting edge 10 basis points https://minutesof.com/q/7c0eb714-2699-491f-b983-09b25778e0f7/ https://minutesof.com/q/7c0eb714-2699-491f-b983-09b25778e0f7/ “What happens when 5% of the world's population is using these models the way the cutting edge 10 basis points are? Like, it's just it's unimaginable. This is why orbital compute is a necessity.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Tranium, by far. Tranium is going to be to 2026, especially in the second half of this year when Tranium three https://minutesof.com/q/a91f962e-6711-454f-b1db-5c6d4437360d/ https://minutesof.com/q/a91f962e-6711-454f-b1db-5c6d4437360d/ “Tranium, by far. Tranium is going to be to 2026, especially in the second half of this year when Tranium three really ramps, as TPUs were to twenty twenty five.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker predicts Trainium will dominate 2026 like TPUs did in 2025, with Trainium 3 ramping in second half. https://minutesof.com/q/77488a06-75d9-432d-b0cd-0afa0e21f41f/ https://minutesof.com/q/77488a06-75d9-432d-b0cd-0afa0e21f41f/ “Tranium is going to be to 2026, especially in the second half of this year when Tranium three really ramps, as TPUs were to twenty twenty five.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker argues Trainium is most underestimated because frontier mixture-of-expert models require switched scale- https://minutesof.com/q/9f8eb50e-f98e-49c0-8051-f5a028e01a39/ https://minutesof.com/q/9f8eb50e-f98e-49c0-8051-f5a028e01a39/ “And so Trainium is for sure the most underestimated, not only because of those design choices, but because the all of these frontier models are what are called mixture of expert models.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker states only NVIDIA and Amazon Trainium have functioning switched scale-up networks for inference today. https://minutesof.com/q/05badf52-27d8-4742-8532-91f442850ac7/ https://minutesof.com/q/05badf52-27d8-4742-8532-91f442850ac7/ “And the only two functioning switched scale up networks in the world today are the ones that power NVIDIA GPUs and Amazon's Trainiums.” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker reveals Atreides could have invested over $50 million in CoreWeave at $1.1 billion valuation but was con https://minutesof.com/q/4bbf9d84-bae2-460e-a752-a60a13663534/ https://minutesof.com/q/4bbf9d84-bae2-460e-a752-a60a13663534/ “I could've Atreides could've invested over $50,000,000 in the round at 1,100,000,000, And I was conflicted out by Crusoe,” — Sohn Conference Foundation Fri, 15 May 2026 18:17:41 +0000 inference Baker identifies three axes of AI scaling: pretraining, inference time compute, and now reasoning as the third https://minutesof.com/q/5f490d92-e1c8-4877-b369-6b470be69d65/ https://minutesof.com/q/5f490d92-e1c8-4877-b369-6b470be69d65/ “And then we started scaling around inference time compute. And it's very clear that we have now added a third axis of scaling performance, and that is reasoning.” — All-In Podcast Sat, 04 Jan 2025 00:23:00 +0000 inference Baker predicts Tesla FSD will achieve 100x improvement quickly as compute scales to GPT-4.5 levels. https://minutesof.com/q/a913c721-311e-4f5d-a524-73f6cda49ae1/ https://minutesof.com/q/a913c721-311e-4f5d-a524-73f6cda49ae1/ “I think they're going to go really fast to GPT-4.5 compute, which means you're going to get, using these orders of magnitude, you're going get a 100x improvement really fast.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 inference Baker argues AI revolution stems from cloud computing power and mobile-generated data, not algorithmic advance https://minutesof.com/q/c6581532-cbbe-425f-bb38-995439fb7bb4/ https://minutesof.com/q/c6581532-cbbe-425f-bb38-995439fb7bb4/ “The only thing that has enabled the AI revolution that we're living through, which I think we're at the bottom of the first inning in, is one, we had the ability to do cloud computing, so just apply significantly more computational power to old algorithms, and then b, we had dramatically more data.” — Invest Like the Best Tue, 26 Nov 2019 10:30:00 +0000 inference Baker states data quantity is the single most predictive element of AI quality, not algorithms or infrastructu https://minutesof.com/q/32e64b43-db5d-48ee-815c-272156cf2971/ https://minutesof.com/q/32e64b43-db5d-48ee-815c-272156cf2971/ “The single most predictive element of knowledge about AI quality is the quantity of data used to train the algorithm.” — Invest Like the Best Tue, 26 Nov 2019 10:30:00 +0000 inference Baker cites research showing every 10x increase in training data doubles AI quality. https://minutesof.com/q/1c159533-7002-43df-a5db-be3d8cc9d824/ https://minutesof.com/q/1c159533-7002-43df-a5db-be3d8cc9d824/ “and it's been very well established in multiple papers from both Google and Microsoft research that for every order of magnitude increase in the data you use to train an algorithm, the quality of the AI doubles.” — Invest Like the Best Tue, 26 Nov 2019 10:30:00 +0000 inference