Microsoft’s Brad Smith on Washington’s AI policy: ‘Regulation without transparent or complete rules’

Microsoft President Brad Smith says the Trump administration’s AI policy is “regulation without transparent or complete rules,” arguing companies can’t plan and that export controls are the wrong tool for frontier AI delivered via APIs. He points to Commerce Department actions that forced Anthropic to pull Fable 5 worldwide and pushed OpenAI to delay early access to GPT-5.6, though both eased (Fable 5 returned earlier this month; GPT-5.6 is set for public launch Thursday). The episode is also cited by foreign governments as evidence of over-reliance on U.S. infrastructure, raising broader geopolitical and compliance uncertainty for frontier-model providers.

Analysis

The market is likely to misread this as a generic anti-AI headline; the real mechanism is a distribution tax on frontier models. That tends to hurt smaller model labs and any vendor whose monetization depends on frictionless public API rollout, while favoring incumbents that can absorb compliance overhead, negotiate with regulators, and sell through existing enterprise channels. For Microsoft, the bigger issue is not immediate revenue loss but a higher policy risk premium around AI attach rates and the cadence of Copilot/Azure AI monetization; if customers think access can be paused or gated, procurement cycles lengthen and multiples compress before earnings do.

Second-order, the "sovereign AI" reaction should push more spend toward private inference, regionally segregated clouds, and governance tooling. That is structurally better for platforms with compliance and identity strengths, and less favorable for pure frontier-model distribution. Salesforce can get a modest narrative tailwind because enterprise buyers under regulatory uncertainty prefer workflow software with auditability and controls over raw model access, but the read-through is still indirect; this is not an immediate demand inflection.

The contrarian point is that the consensus may be overpricing policy chaos for mega-cap incumbents and underpricing the moat effect. If Washington is effectively creating a trusted-partner regime, Microsoft is one of the few firms positioned to benefit from it. The real falsifier is a sustained slowdown in Azure AI usage or a formalized licensing process that materially delays model releases; absent that, this is mostly headline volatility rather than a fundamental earnings reset over the next 1-3 months.

Key Decisions for Investors

• Buy MSFT on 2-3% policy-driven pullbacks over the next 1-2 weeks; treat this as headline risk, not a thesis break, unless Azure AI monetization commentary deteriorates next quarter.
• Keep CRM on the watchlist as a relative beneficiary of enterprise governance demand; only add if management shows AI attach-rate improvement or if customers explicitly shift toward compliant workflow tooling.
• Avoid chasing frontier-AI pure plays on the assumption that policy ambiguity is temporary; if the government repeats these interventions, the valuation gap should widen for firms without a trusted-partner moat.
• Set an alert for any published 'trusted partner' criteria or formal pre-release process; if rules become explicit, rotate from policy-risk names into the platforms most likely to be approved, and trim exposure if Microsoft is excluded or delayed.