Trump AI Action Plan (Winning the Race) + 2025-07 EOs Package
Summary
Section titled “Summary”“Winning the Race: America’s AI Action Plan” is a 25-page strategy document released by the White House on 2025-07-23, with 3 companion executive orders signed the same day.
Core objective (responding to Section 3 of EO 14179’s 180-day Action Plan mandate): “sustain and enhance America’s global dominance in AI through deregulation, ideologically neutral AI, infrastructure investment, and international competition.”
Three pillars
Section titled “Three pillars”Pillar 1 · Accelerate Innovation
Section titled “Pillar 1 · Accelerate Innovation”- Prioritize federal AI R&D funding
- National-scale AI infrastructure (compute, data, energy)
- Encourage open-source model ecosystems (particularly against the pressure from China’s DeepSeek)
- AI for Science applications
Pillar 2 · Build AI Infrastructure
Section titled “Pillar 2 · Build AI Infrastructure”- Data centers: accelerated permitting, energy supply, land allocation
- Domestic semiconductors + targeted chip export controls
- AI-ready grid and nuclear energy investment
Pillar 3 · Lead International AI Diplomacy
Section titled “Pillar 3 · Lead International AI Diplomacy”- Promote U.S. AI models as the international default
- Push back against the EU’s Brussels Effect
- Technology decoupling from China’s AI ecosystem
Three companion executive orders (same day)
Section titled “Three companion executive orders (same day)”EO · Preventing Woke AI in the Federal Government
Section titled “EO · Preventing Woke AI in the Federal Government”- OMB to issue “Unbiased AI Principles” implementation guidance within 120 days
- LLMs under federal contracts (new or existing) must comply with the Unbiased AI Principles
- “Decommissioning costs” — LLM vendors bear replacement costs if non-compliant
- Impact: Anthropic, OpenAI, Google, Meta, and others need compliant variants for federal LLM contracts
Controversy:
- What counts as “neutral”? Who decides?
- Potential First Amendment (free expression) conflicts
- Industry consensus that “ideological neutrality is technically unrealistic to achieve”
EO · Accelerating Federal Permitting of Data Center Infrastructure
Section titled “EO · Accelerating Federal Permitting of Data Center Infrastructure”- “Qualifying projects”: AI training / inference / simulation / synthetic-data data centers with compute needs ≥ 100 MW
- Streamlined NEPA (National Environmental Policy Act) review
- Accelerated land / energy / water approvals
- Grid interconnection priority
Impact: accelerates mega-cluster projects such as Stargate, xAI Colossus, and the Meta Llama Cluster.
EO · Exporting American AI
Section titled “EO · Exporting American AI”- Coordinates Commerce, Treasury, and State
- Allied nations (EU, Japan, Korea, UK) get preferential access to U.S. AI models and compute
- Non-allies (China, Russia, Iran, North Korea) face tightened controls
- Links with BIS chip export controls (2023-10, 2024-12, etc.)
State-law preemption preview (landed 2025-12 as EO 14365)
Section titled “State-law preemption preview (landed 2025-12 as EO 14365)”The Action Plan explicitly directs:
- The Attorney General to establish an AI Litigation Task Force before 2026-01-10
- The Commerce Secretary to publish a “burdensome state-laws” assessment before 2026-03-11
- BEAD broadband funds to be conditioned on state AI regulatory “stringency”
→ All of these landed in EO 14365.
Comparison with Biden’s EO 14110
Section titled “Comparison with Biden’s EO 14110”| Dimension | Biden EO 14110 (2023, revoked) | Trump AI Action Plan (2025) |
|---|---|---|
| Framing | ”Safe, secure, trustworthy" | "Global dominance” |
| Frontier models | 10²⁶ FLOP reporting | No mandatory reporting |
| AI Safety Institute | Established AISI | Reorganized as CAISI, focused on standards |
| Civil rights | Blueprint AI Bill of Rights | Removed / revoked |
| State-law relationship | Default compatibility | Active preemption |
| Open source | Neutral | Explicit support |
| International approach | Multilateral + G7 | Bilateral + allies-first |
Comparison with other countries’ strategies
Section titled “Comparison with other countries’ strategies”- China’s “New Generation AI Development Plan” (2017): three-step pathway to global dominance by 2030; state-led industrial policy
- EU AI Act (2024): Brussels Effect + ex ante compliance
- UK AI Safety Institute (2023): frontier safety risk evaluation
The Trump AI Action Plan seeks to reshape international AI discourse, responding to the twin pressures of China’s industrial policy and the EU’s ex ante regulation.
Text and archives
Section titled “Text and archives”| Source | Link |
|---|---|
| Official landing page | ai.gov |
| 25-page full-text PDF | download from ai.gov |
| Sidley Data Matters deep dive | datamatters.sidley.com/…/trump-ai-action-plan |
| Covington Inside Privacy | insideprivacy.com/artificial-intelligence/trump-administration-issues-ai-action-plan |
Version history
Section titled “Version history”| Date | Event |
|---|---|
| 2025-01-23 | EO 14179 launches the 180-day Action Plan mandate |
| 2025-07-23 | Action Plan + 3 EOs released same day |
| 2025-12-11 | EO 14365 preempts state laws (Action Plan’s third step) |