US AI export controls and Pentagon reshuffles: Who’s winning the new power game?

In President Donald Trump’s Washington, AI policy is becoming a battlefield where companies and defense leaders try to anticipate shifting export-control rules. Politico reports that Anthropic CEO Dario Amodei has not “played the game” of aligning messaging with the White House, even as the administration’s approach to AI governance and security rapidly evolves. The article frames export controls as a political fight, not just a technical compliance exercise, with the White House and the Defense Department positioned as key arbiters of what can be shipped and to whom. Separately, reporting on the Pentagon highlights internal friction under Defense Secretary Pete Hegseth, including secret vetting and blocked promotions tied to a “war on diversity,” underscoring how personnel policy may affect institutional risk tolerance and operational culture. Geopolitically, the cluster points to a tightening linkage between AI industrial policy, alliance management, and dollar-centered leverage. One post argues that America’s allies are already exposed to Trump’s bargaining on trade, alliances, and the dollar system, and that AI could become the most powerful lever of all—effectively turning compute, models, and export permissions into bargaining chips. Another piece notes that Japanese politicians now market their role in building an AI “tech stack” intended to outpace China, signaling that alliance cooperation is increasingly framed as strategic industrial competition rather than values rhetoric. Meanwhile, the Pentagon diversity controversy suggests the US defense establishment may be reshaped in ways that influence how quickly it adopts or restricts emerging AI capabilities, potentially affecting interoperability with allies and the credibility of US commitments. Market implications concentrate in AI infrastructure, export-control-sensitive software, and defense-adjacent technology spending. If Anthropic-style model providers face stricter export controls or more politicized licensing, investors may reprice the probability of delayed international revenue, raising risk premia for frontier-model developers and AI platform vendors. The “AI as leverage” narrative also implies that semiconductors, cloud capacity, and data-center buildouts in allied jurisdictions could see accelerated demand, while China-linked AI supply chains may face tighter constraints and higher compliance costs. On the defense side, internal personnel vetting and promotion slowdowns can translate into procurement and adoption delays for AI-enabled systems, affecting defense tech contractors and cybersecurity/ISR vendors that rely on stable institutional pipelines. What to watch next is whether export-control enforcement becomes more explicit—through licensing guidance, enforcement actions, or changes in interagency review timelines involving the White House and the Defense Department. Track signals such as Anthropic’s public policy posture, any new compliance disclosures, and whether other frontier AI labs adjust messaging to match administration priorities. For alliance dynamics, monitor trade and dollar-related bargaining language alongside concrete AI cooperation announcements from Japan and other partners, looking for conditionality tied to compute access or model deployment. Finally, watch Pentagon personnel policy outcomes under Hegseth—especially promotion patterns, vetting procedures, and any resulting morale or capability gaps—because these can quickly translate into operational risk and procurement timing shifts that markets will price in within weeks.

Geopolitical Implications

• 01

  The US may use AI export permissions as leverage over allies’ trade and alliance behavior, deepening transactional alliance management.

• 02

  Japan’s framing of AI cooperation as industrial competition against China suggests a shift toward capability blocs rather than values-based alignment.

• 03

  Defense institutional restructuring under Hegseth could affect US credibility on rapid AI capability integration and interoperability with partners.

• 04

  Politicized export-control enforcement can fragment global AI supply chains, increasing compliance costs and accelerating regionalization of compute and model deployment.

Key Signals

• —New or updated AI export-control guidance, licensing criteria, or enforcement actions involving frontier model providers.
• —Public statements and compliance disclosures from Anthropic and peer labs about policy alignment and export readiness.
• —Evidence of conditionality in ally AI cooperation announcements (compute access, deployment restrictions, or procurement-linked commitments).
• —Pentagon promotion and vetting outcomes under Hegseth, including any reported impacts on readiness, retention, or program execution.