On Sovereign AI — with a Senior Policy Advisor

This is the episode I have been building toward for months. My guest is a senior policy advisor who was directly involved in drafting portions of the EU AI Act. They agreed to speak on the condition of anonymity, which allowed a level of candor that is rare in policy conversations about AI.

We started with the question that opens the Sovereign AI essay: why did every major economy suddenly start building national AI stacks in the same eighteen-month window? The official answer — national security, data sovereignty, economic competitiveness — is true but incomplete. The unofficial answer involves a series of private briefings in late 2023 that changed how several European capitals thought about AI dependency.

The export controls did not just restrict chips. They demonstrated that the supply chain is a lever, and everyone on the wrong end of a lever starts looking for alternatives.

Key Moments

• 08:14 — Why GPU export controls reshaped the alliance map overnight
• 17:30 — The data-residency clause in the AI Act that nobody is reading carefully
• 29:45 — What a “national champion” model actually looks like in practice
• 41:20 — How small countries should bet — the Singapore and UAE playbooks
• 48:55 — The bilateral compute agreements that will matter more than any regulation

Why This Matters

This is a long episode — nearly an hour — and it rewards close listening. The policy details matter because they will shape the product landscape for the rest of the decade. If you are building anything that touches multiple jurisdictions, this is required context.

The guest’s framing of AI regulation as infrastructure regulation, not product regulation, changed how I think about the entire space. I suspect it will change your thinking too.