What Actually Happened

On July 23, 2025, the Trump White House released "Winning the Race: America's AI Action Plan" — a document mandated by Executive Order 14179. According to Morrison Foerster's legal analysis of the plan, it identifies over 90 federal policy actions across three pillars: accelerating AI innovation, building American AI infrastructure, and leading in international AI diplomacy and security.

The same day, Trump signed three additional Executive Orders: one banning "woke AI" in the federal government, one fast-tracking environmental permitting for data center construction, and one promoting exports of American AI technology. Four major AI policy moves arrived in a single day.

The Three Pillars — What They Mean

Pillar One: Deregulate. Morrison Foerster's breakdown shows the plan calls for eliminating "red tape and onerous regulation" and explicitly targets what it calls "ideological biases" in AI systems. The plan promotes open-source AI development and directs agencies to build "world-class datasets" for American research. The executive order on "woke AI" directly targets federal systems built around DEI-driven outputs.

Pillar Two: Build. The administration wants to fast-track energy projects and modify the environmental permitting system to accelerate data center construction. The plan acknowledges that AI development requires physical infrastructure — power, land, and chips — and that current permitting timelines are slowing deployment.

Pillar Three: Export and Block. The plan directs the U.S. to actively promote American AI exports and coordinate with "like-minded" countries on international AI governance. This establishes a coalition around American AI standards rather than Chinese ones.

The plan was co-signed by Michael Kratsios (Assistant to the President for Science and Technology), David Sacks (Special Advisor for AI and Crypto), and Marco Rubio (National Security Advisor). The signatures indicate this functions as a national security document as much as a technology policy.

The Export Control Gap

The U.S. cannot win this race alone, and the allies it needs lack some critical tools.

A March 2025 report from the Center for Strategic and International Studies (CSIS), authored by Gregory C. Allen and Isaac Goldston, documents this plainly. Countries like the Netherlands, Germany, South Korea, Japan, and Taiwan control critical chokepoints in the AI and semiconductor supply chain. But many lack legal equivalents to U.S. tools like the Foreign Direct Product Rule and the Entity List — mechanisms America uses to cut off Chinese access to advanced chips.

CSIS identifies the problem as enforcement capacity and political will. Some allies possess the legal authority. They are not consistently using it. The AI Action Plan addresses this through language about "like-minded countries," but the diplomatic work of securing Dutch and German enforcement of American-style controls remains incomplete.

DeepSeek Changed the Calculus

The Hudson Institute's Nury Turkel documented how China's DeepSeek AI launch forced urgent reconsideration of U.S. strategy. Trump called it a "wake-up call." New York, the Pentagon, and Capitol Hill all moved to ban it from their systems.

China exploited loopholes in U.S. export control laws to build a competitive frontier AI model, according to Hudson Institute analysis. The new plan explicitly directs agencies to "identify and eliminate loopholes" — language that traces back to Trump's first-day executive order on export controls.

American Companies Already Moved

While Washington wrote policy, some American companies were already operating under the assumption that domestic-only AI infrastructure is essential.

Headwall, a Maryland-based startup, builds AI-powered virtual command centers for military, law enforcement, hospitals, and rescue crews. According to the Daily Signal, co-founder Geoff Bundt emphasized the system "can be run in a completely off-network environment" — meaning zero foreign dependency. Co-founder Adam Weiner said the company launched specifically to advance American innovation for national defense.

The U.S. Department of Defense has spent at least $55 billion over the past decade on domestic AI production, according to the Daily Signal. Sen. Rick Scott (R-FL), a senior member of the Senate Armed Services Committee, stated: "It's foolish to rely on a country that hates us for national security."

What Gets Overlooked

Most coverage frames the AI Action Plan as a tech deregulation story. This is a national security mobilization that involves technology. The three co-signatories include the National Security Advisor. The export control strategy depends on allied governments unprepared for enforcement. And the domestic infrastructure buildout requires energy permitting reforms that will face legal challenges from environmental groups.

What Happens Next

If this plan succeeds, American AI systems become the global standard rather than Chinese ones. If it fails — if allies don't enforce export controls, if permitting reforms stall, if the defense AI investment produces bureaucratic bloat instead of capability — China fills the gap.

The plan is on paper. Execution begins now.