1 — Origins & Stated Objectives

United States. The plan explicitly casts AI as a national‑security race and cites the goal of “unquestioned and unchallenged global technological dominance.” It is organized around three pillars: Accelerate AI Innovation, Build American AI Infrastructure, and Lead in International AI Diplomacy & Security, and follows Executive Order 14179 (“Removing Barriers to American Leadership in AI”). [PDF] (EO 14179)

China. The plan presents AI as an international public good guided by principles including benefiting people, respecting sovereignty, development‑orientation, safety and controllability, fairness and inclusiveness, and open cooperation, and situates implementation within the UN’s Future Summit and its Global Digital Compact. It also calls for “AI+” integration across industries. (Gov.cn full text)

Takeaway. Washington starts from competition and speed; Beijing starts from order and inclusion.

2 — Scope & Architecture

United States (3 pillars → c.~90 actions). The plan details scores of near‑term policy steps: deregulation, streamlined federal procurement, expanded AI evaluation, defense adoption, permitting reform for data centers and fabs, grid upgrades, and tightened export controls. (Reuters) [PDF]

China (13 action areas). The framework sets out thirteen domains: accelerating innovation; empowering industries; interoperable compute and data‑center buildout; open‑source ecosystems; high‑quality data; green/energy‑efficient AI; standards; public‑sector adoption; safety governance; UN‑anchored multilateralism; global capacity building; and multi‑stakeholder participation. (Gov.cn full text)

Note on compute targets: China’s 300+ EFLOPS by 2025 target derives from a separate October 2023 national compute‑infrastructure plan (not the July 2025 governance plan). (Reuters, Oct. 9, 2023) (English.gov.cn)

3 — Governance Philosophy & Ideological Posture

United States. Agencies are directed to remove “onerous regulation,” review FTC actions for undue burdens, and revise the NIST AI Risk Management Framework to eliminate references to misinformation, DEI, and climate change. Procurement guidance favors systems that are “objective” and free from top‑down ideological bias. [PDF]

China. Emphasis falls on standards, fairness, sovereignty and risk management (classification, testing, threat‑information sharing, traceability) within interoperable frameworks grounded in international law. (Gov.cn full text)

Related U.S. executive orders (July 23, 2025): Preventing “Woke” AI in the Federal Government; Promoting the Export of the American AI Technology Stack; Accelerating Federal Permitting of Data Center Infrastructure.

4 — Open‑Source & Access

United States. Strong support for open‑source/open‑weight models (including NAIRR pilots, compute‑access market concepts, and NTIA‑led SME adoption) to spur startups and research. [PDF]

China. Calls for cross‑border open‑source communities, compliant open‑source ecosystems, and openness of non‑sensitive resources to improve stack interoperability—especially benefiting developing countries. (Gov.cn full text)

Example. Open‑weight releases like DeepSeek‑R1 illustrate China’s growing presence in open ecosystems. (DeepSeek‑R1 GitHub) (DeepSeek announcement)

5 — Compute, Energy & Physical Infrastructure

United States. The plan prioritizes expedited permitting for data centers, fabs, and energy infrastructure; grid expansion; secure facilities for defense; and onshoring semiconductors. [PDF]

China. The governance plan urges clean power, intelligent compute, unified compute standards, and support for Global South infrastructure access. (Gov.cn full text) (For the separate national compute target see: English.gov.cn, Oct. 2023.)

Investment baseline. U.S. private AI investment reached $109.1B in 2024 vs. $9.3B in China. (Stanford HAI AI Index 2025) [Report PDF]

6 — Data, Privacy & Safety

United States. Focus on evaluation ecosystems, interpretability/robustness, and biosecurity; expand export‑control enforcement; and evaluate PRC frontier models for alignment with censorship/propaganda. [PDF]

China. Proposes risk classification, testing, incident response, data protection, and traceability to curb misuse. (Gov.cn full text) (Analysts continue to document censorship concerns in some Chinese models.) (arXiv, June 2025) (USCC, 2024)

7 — International Strategy

United States. Export the “American AI tech stack,” counter Chinese influence in standards bodies, tighten controls on compute/semiconductor technologies, and align protections with allies. [PDF] (EO: Export of AI Stack)

China. Proposes establishing a global AI cooperation organization, with Shanghai under consideration for the headquarters, to coordinate development and governance and reduce fragmentation—especially for the Global South. (Reuters) (English.Shanghai.gov.cn)

8 — Workforce & Domestic/Global Adoption

United States. Jobs and living‑standards focus; centers of excellence and sandboxes; sector pilots (e.g., healthcare, energy, agriculture); infrastructure skills. [PDF]

China. Capacity building as a centerpiece of international cooperation (joint labs, shared testing, curated datasets) and public‑sector first‑mover deployments in health, education, and transport. (Gov.cn full text)

9 — Standards & Multilateralism

United States. Lead through innovation and alliances; resist norms perceived to constrain frontier progress; scrutinize PRC models. [PDF]

China. Accelerate standards via ITU/ISO/IEC, reduce algorithmic bias, and establish transparent, interoperable frameworks; supports UN‑anchored mechanisms. (Gov.cn full text)