The Minutes of Gavin Baker on ai infrastructure https://minutesof.com/gavin-baker/on/ai-infrastructure/ Everything Gavin Baker has said on ai infrastructure: 15 verbatim quotes between August 2024 and August 2026, each with a timestamp and a link to the… en Sun, 30 Aug 2026 16:21:41 +0000 Baker identifies memory as the single most important factor for increasing token output per compute unit. https://minutesof.com/q/498c1547-6142-4dfd-9717-055d939ec2fc/ https://minutesof.com/q/498c1547-6142-4dfd-9717-055d939ec2fc/ “It's the single most important thing you could do to increase token output per unit of compute, and then that obviously, definitionally, actually lowers costs,” — Invest Like the Best Tue, 04 Aug 2026 08:00:00 +0000 ai infrastructure Baker explains MFU runs at 35-40%, measuring the percentage of theoretical compute actually used for training. https://minutesof.com/q/91f8d62c-9179-4722-83b1-a80397265ed3/ https://minutesof.com/q/91f8d62c-9179-4722-83b1-a80397265ed3/ “MFU, model flops utilization, and that generally runs around 35 to 40%. And that's literally the percentage of compute, theoretical compute flops that you're actually applying to trading.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 ai infrastructure Baker explains higher MFU allows 25% faster time to market with same GPU and power spending. https://minutesof.com/q/2ef0281e-1acf-4b2f-a09a-427746b64fa5/ https://minutesof.com/q/2ef0281e-1acf-4b2f-a09a-427746b64fa5/ “You have the same amount of GPUs and the same amount of power presumably. You could choose between faster time to market.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 ai infrastructure Baker explains that 50% MFU versus 40% MFU enables 25% faster time to market for AI models. https://minutesof.com/q/b18ea10e-8f3d-44a4-afda-d9363c3a12ee/ https://minutesof.com/q/b18ea10e-8f3d-44a4-afda-d9363c3a12ee/ “You could choose between faster time to market. If you run a 50% MFU and your competitor's running 40 for an equivalent amount of trading flops, you could be in market 25% faster.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 ai infrastructure Baker proposes a new metric decomposing MFU that he developed the morning of the interview. https://minutesof.com/q/27c177e8-03c2-4627-956e-295dd3a471e4/ https://minutesof.com/q/27c177e8-03c2-4627-956e-295dd3a471e4/ “MFU is the most important metric because it gives you all of these advantages and ways to differentiate yourself amongst five people, six people who've trained these GPT core class models.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 ai infrastructure