Today’s Sharp Take: The AI Power Problem is Solved

Source: Nate’s Newsletter Date: 2024-09-26 URL: https://natesnewsletter.substack.com/p/todays-sharp-take-the-ai-power-problem

Summary

Written in September 2024 around Mira Murati’s departure from OpenAI and a rumored $6.5B capital raise, Nate argues that the “AI power problem” (infrastructure and energy constraints) is being solved by the sheer volume of capital flowing into the sector. Personnel changes at OpenAI are dismissed as noise; the investment pipeline is the real signal.

Implications

Capital thread. The “money solves infrastructure” thesis was partially right: data center buildouts and nuclear/power deals accelerated through 2024-2025. But capital adequacy doesn’t automatically translate to energy availability or regulatory approval, and those bottlenecks have proven stickier than the September 2024 read suggested.

AI economics thread. The framing reveals a common 2024-era assumption: that compute and power were the binding constraints, and capital unlocks them linearly. The more interesting 2025 question is whether the constraint has shifted — to talent, regulatory capacity, or model plateau — in ways that capital doesn’t directly address.

Watch: Whether the power/infrastructure investment wave of 2024-2025 produces the data center capacity it promised, or whether regulatory, grid, and siting delays push the actual capacity inflection into 2026-2027.