U.S. AI Dominance Rests on Infrastructure, Not Models

The global conversation around artificial intelligence typically centers on which company builds the most capable model. Yet this framing misses the structural reality of how AI power actually works.

According to analysis first reported by Fortune, the United States maintains its AI leadership not through algorithmic superiority but through ownership and control of the physical and digital infrastructure that makes modern AI possible. This includes hyperscale data centers, cloud computing platforms, specialized servers, and the undersea fiber-optic cables connecting them—all subject to American export controls, data extraction laws, and deep integration with defense agencies.

The Nvidia ecosystem as blueprint

Nvidia exemplifies how infrastructure dominance functions. The company controls roughly 85% of the global market for graphics processing units required to train frontier AI models. Its CUDA software platform has become the default development environment, creating substantial switching costs that lock customers into its ecosystem.

That ecosystem extends through hardware manufacturers like Broadcom and cloud hyperscalers—Amazon Web Services, Google Cloud, and Microsoft Azure—which provide the infrastructure powering foundational AI developers including Anthropic, OpenAI, Meta, and Alphabet.

These companies form what amounts to a new U.S. AI industrial complex, bound together through extensive defense and intelligence contracts. The Department of Defense has signed operational agreements with Google, OpenAI, Microsoft, Amazon Web Services, Oracle, and Nvidia to deploy frontier AI tools onto classified military networks. The Pentagon's FY2027 budget allocates more than $54 billion for autonomous warfare and drone systems through a newly formed Defense Autonomous Warfare Group.

Federal AI contracting surged by $90.7 billion in 2026 alone, according to the Brookings Institution. OpenAI, xAI, and Google have signed contracts allowing the Pentagon "all lawful use" for defense purposes, including autonomous weapons and mass surveillance. Anthropic has reportedly embedded engineers inside the National Security Agency to adapt its Mythos model for offensive cyber operations.

Physical chokepoints

American cloud hyperscalers now own and operate roughly 70% of usable undersea cables in 2026. Meta plans to build a 40,000-kilometer around-the-world fiber-optic subsea cable at a projected cost of $10 billion. Google's Pacific Connect Initiative will spend over $1 billion connecting Japan to the South Pacific.

Washington retains authority to restrict where these networks extend and who accesses them. In 2020, U.S. regulators blocked the Hong Kong segment of the Pacific Light Cable Network backed by Google and Meta, forcing abandonment of the direct U.S.-Hong Kong link over espionage concerns. Some 13,000 kilometers of already-laid cable was left unused on the ocean floor.

Even Chinese frontier labs training on domestic infrastructure remain dependent on American undersea cables when their models access the global internet for training data or serve international users.

The U.S. CLOUD Act requires U.S.-headquartered cloud providers to produce data within their possession, custody, or control even when stored outside the United States. This grants authorities access not only to stored data but detailed digital trails of AI activity—user prompts, model responses, access patterns, and behavioral records.

Supply chain dependencies

U.S. dominance faces limits. Asian manufacturers produce the semiconductors, AI servers, and essential components. Nvidia's supply chain runs through TSMC, SK Hynix, and Samsung, alongside system integrators like Foxconn, Quanta, and Wistron.

Nvidia CEO Jensen Huang announced billions in new investments and contracts in Taiwan and South Korea last month, including advanced chips and packaging from TSMC and billion-dollar deals with SK Hynix and Samsung. Yet Taiwan and South Korea are unlikely to weaponize their position in the AI value chain, preferring to maintain earnings in what amounts to a friend-shored U.S. technology stack.

China is building parallel undersea fiber-optic networks along its digital silk road to reduce exposure to American infrastructure control.

Why it matters

While attention focuses on model capabilities and chip performance, the physical and legal infrastructure underlying AI ecosystems determines who ultimately controls the technology. Companies and nations building on U.S.-dominated infrastructure remain subject to American regulatory reach regardless of where they operate or whose models they deploy. This structural advantage persists even as competitors develop technically impressive alternatives.

These details were first reported by Fortune.