Now Playing
Sovereign AI at $100B: Infrastructure Boom or Fragmentation Trap?
Audio is available on Spreaker — see link below.
Global sovereign AI spending has crossed $100 billion annually, but owning compute infrastructure is not the same as controlling AI capability. This episode breaks down the fragmentation risk, OpenAI's agent breach, U.S. policy whiplash, and why data centers are now military targets.
Audio is available on Spreaker — see link below.
Global spending on sovereign AI compute just crossed one hundred billion dollars a year. That's not a projection.
Sovereign AI actually has three layers. Data residency is the easiest, and it's the most regulated.
OpenAI had two significant disclosures this cycle, and they sit right next to each other in a way that matters. The first: its coding agents are now producing three-point-one agent-workdays for every one human workday.
The U.S. policy picture moved fast this cycle, and not coherently. Commerce Secretary Lutnick and OSTP Director Kratsios locked in all twenty G20 members behind the Carolina Principles, a framework that favors existing sector regulators over creating new AI-specific bodies.
One data point reframes the whole infrastructure conversation. In March of this year, Iranian drones struck AWS facilities in the Gulf.
Two things to track from here. First, whether national compute programs produce any shared interoperability standards, or whether each country ends up with infrastructure that can't talk to anyone else's.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.