Catalyst with Shayle Kann
Catalyst with Shayle Kann

Inside the AI power wars

July 9, 2026

AI Summary

5 min read

The hyperscalers—Google, Amazon, Microsoft, and Meta—are not monolithic in their approach to powering AI infrastructure. While the public narrative focuses on their chip strategies, a quieter but equally intense competition is playing out among their energy teams. Google has long been the most sophisticated, with a large, experienced energy team that pioneered 24/7 clean energy matching and carbon-aware workload shifting. But the landscape is shifting rapidly. Microsoft, once a clear second, has lost key personnel to poaching, and Amazon has stepped up dramatically with deals like the Talen Energy acquisition and large-scale PPAs for gas and nuclear. Meta, meanwhile, is doing the most unusual stuff of all, precisely because it doesn't have a cloud business. It builds infrastructure almost entirely for its own use, which gives it more flexibility to experiment with behind-the-meter generation and novel data center designs.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

  • 1 (00:06) **Introduction: The Undercovered AI Power War** - Shayle Kann sets up the episode by noting that while chip strategies are well understood, the power strategies of AI companies are not, and that a war for energy talent is underway.
  • 2 (03:12) **Hyperscaler Power Strategies: Not Monolithic** - Jeremy Eliahu explains that the major cloud companies have meaningfully different approaches to power procurement.
  • 3 (08:42) **The Talent War for Energy Teams** - The small group of experts in data center energy provisioning are now hot commodities, with significant team switching happening.
  • 4 (10:06) **Why Behind-the-Meter (BTM) Generation?** - The core driver is speed and certainty, not a desire to go off-grid permanently.
  • 5 (11:45) **The Real Driver: AI Labs, Not Hyperscalers** - The most important question is about the power needs of AI labs like OpenAI and Anthropic, who are the true demand drivers.
  • 6 (14:29) **The Generation vs. Transmission Constraint Debate** - Shayle and Jeremy debate whether the bottleneck is generation capacity or transmission and distribution (T&D) infrastructure.
  • 7 (17:29) **Frontier Labs (OpenAI, Anthropic) Energy Strategy** - These AI labs are hiring energy teams and evaluating sites themselves, but face a massive financing constraint.

+ Full timestamped outline available in the app

Show Notes

For those steeped in the world of AI development, the major differences between tech giants’ chip strategies are easy enough to see: the speed of Google’s TPUs, the affordability of Amazon’s Trainium, and Nvidia’s market dominance. But those same companies’ respective energy strategies often fly under the radar.

As data center buildouts surge across the U.S., the tech industry is hitting a massive wall: a power grid that can’t move fast enough to support them. In this episode, Shayle sits down with Jeremie Eliahou Ontiveros, who leads coverage of infrastructure and power at SemiAnalysis, to lift the curtain on the hyper-competitive world of AI energy procurement. 

They cover topics including:

  • The hyperscaler leaderboard: Why Google remains the most energy-sophisticated tech giant, as others continue to innovate
  • Why power has effectively become revenue for frontier labs
  • How Google uses its massive balance sheet to provide financial backstops for Anthropic’s data center buildouts
  • Why the interconnection queue is forcing AI labs like OpenAI and Anthropic to bring their own generation
  • The sudden market frenzy over modular speed-to-power options—ranging from aeroderivative gas turbines and massive reciprocating engines to billions of dollars of fuel cells