Trump Readies AI Order Amid Industry Boom and Regulatory Pushback

Rapid advances in artificial intelligence have placed new pressure on the federal government to define its role before market forces and foreign competitors set the terms. On Wednesday afternoon, OpenAI reported that one of its reasoning models solved a geometry problem unsolved by mathematicians for 80 years. The same day, Anthropic projected first-quarter revenue above $10 billion and signed a $1.25 billion monthly compute agreement with SpaceX through 2029. Nvidia posted $81.6 billion in quarterly revenue, nearly all from data-center chips. These developments arrived hours before President Trump was scheduled to sign an executive order on AI with chief executives from OpenAI, Anthropic, Microsoft, Google, Meta, Amazon, Apple and SpaceX in attendance.

The order is expected to create a voluntary system under which leading labs would share new models with the government 90 days before public release. White House AI czar David Sacks has described the step as a way to address cybersecurity risks without imposing mandatory licensing. The approach differs from earlier proposals that would have required pre-approval or imposed liability rules on developers. Tech executives have offered mixed reactions, with some viewing the timeline as manageable and others warning that any formal channel risks favoring companies already inside the room.

Anthropic has separately backed tighter export controls on advanced chips and new disclosure requirements. Sacks called those positions a regulatory-capture effort that would raise costs for smaller rivals. The company has not released a detailed rebuttal tying its recommendations to specific national-security assessments. Meanwhile, public polling cited in multiple reports shows roughly 70 percent of Americans believe AI is moving too fast, while nearly two-thirds doubt the economic gains will reach everyone.

The central unresolved question is whether the coming order will function mainly as an information-sharing bridge or will evolve into broader constraints on model development and distribution. Historical precedents such as the failed Solyndra and Fisker investments illustrate the risk that government selection of technology winners can produce poor returns. At the same time, the absence of any coordinated U.S. framework leaves open the possibility that standards will be written elsewhere, either by foreign governments or by the largest firms acting alone.