The Power Allocation

Meta created Meta Compute to build computing capacity faster than anyone else. Hyperion in Louisiana will scale to 5 gigawatts — equivalent to 5 nuclear reactors.

Show Notes

Meta has created Meta Compute to centralize and accelerate its AI infrastructure buildout. With Prometheus in Ohio and Hyperion scaling to 5 gigawatts in Louisiana, Meta is building at scales that rival utility infrastructure projects. This episode examines Meta's compute strategy and its implications.

Key topics covered:

  • Meta Compute division reporting directly to Zuckerberg
  • Prometheus supercluster activating in Ohio 2026
  • Hyperion campus scaling to 5 GW in Louisiana
  • $135 billion AI investment potential in 2026
  • Llama 4 and distributed inference requirements

Related keywords: Meta Compute, Meta data centers, Hyperion Louisiana, Prometheus Ohio, Meta AI infrastructure, Zuckerberg compute

Brought to you by Spring Street Management Group.

]]>

What is The Power Allocation?

The AI boom isn't constrained by chips, algorithms, or talent. It's constrained by electricity.

The Power Allocation is a daily briefing on AI infrastructure — where capital is actually being deployed. Each episode cuts through the hype to examine the physical realities shaping the AI buildout: power constraints, grid interconnection, land acquisition, data center financing, cooling infrastructure, and utility relationships.

This isn't a software podcast. This is an infrastructure podcast.

Who it's for: Institutional investors, infrastructure allocators, data center developers, utilities, family offices, and anyone positioning capital for the physical layer of artificial intelligence.

What you'll learn:

Why power availability — not GPU supply — is the binding constraint on AI compute
How hyperscalers are locking in multi-decade power purchase agreements
Where data centers are relocating and why grid geography is reshaping the industry
The financing structures turning compute facilities into bond-like assets
What execution timelines, permitting delays, and interconnection queues mean for capital deployment
Format: 3-6 minute episodes. Dense. Clear. No hype.

New episodes daily. Subscribe wherever you listen.