The United States maintains its edge in the AI race not so much through individual models but by controlling infrastructure—chips, clouds, data centers, and undersea cables. This opinion was expressed by Alex Capri, author of the book Techno-Nationalism, in a column for Fortune.
According to Capri, discussions about AI competition often focus on comparing models, such as those from OpenAI, Anthropic, DeepSeek, Z.ai, or Moonshot. However, advanced systems require a broader and more capital-intensive base: graphics processors, servers, cloud infrastructure, large data centers, and data transmission networks.
Capri highlighted the role of Nvidia, stating that the company controls about 85% of the global market for graphics processors used in training advanced AI models. Its advantage is not only in chips but also in the CUDA software ecosystem, which has become the standard environment for many developers, creating high switching costs to alternatives.
Around this ecosystem, Capri built a broader “American AI stack”: hardware manufacturers like Broadcom, cloud hyperscalers such as Amazon Web Services, Google Cloud, and Microsoft Azure, as well as model developers and big tech firms like OpenAI, Anthropic, Meta, and Alphabet.
Defense Framework
Capri also linked the U.S. advantage to the collaboration between tech companies, the Pentagon, and intelligence agencies. He noted that major AI infrastructure providers serve both the commercial market and participate in developing systems for the defense sector.
As an example, he cited the U.S. Department of Defense’s structure for developing autonomous military systems, the Defense Autonomous Warfare Group (DAWG). The Pentagon’s budget request for the 2027 fiscal year allocates $54.6 billion for DAWG.
Another indicator is federal AI contracts. According to Brookings, in 2026, their potential value at the U.S. Department of Defense rose to $90.7 billion, with $7.2 billion actually reserved for this direction.
Undersea Cables and Data
Another level of infrastructure is undersea fiber-optic cables. Capri noted that major American tech companies are increasingly building their own backbone networks, thereby gaining influence over data transmission routes. He cited the Pacific Light Cable Network as an example: in 2020, authorities supported the segment between the U.S. and Taiwan but restricted direct connection with Hong Kong due to risks associated with China.
Capri assessed that control over such routes is important not only for the U.S. but also for Chinese labs. Even if models are trained within China, they remain dependent on the global internet for data collection, user service, and external API operations.
The author also reminded of the CLOUD Act. The law, enacted in the U.S. in 2018, allows authorities to request data from American cloud providers through legal procedures, even if stored outside the country.
Limitations of U.S. Advantage
Capri acknowledged that U.S. dominance is not absolute. The production of key components for the AI stack depends on Asia: TSMC, SK Hynix, Samsung, Foxconn, Quanta, and Wistron.
However, he assessed that Taiwan and South Korea are unlikely to use this position against Washington, as they are integrated into a U.S.-friendly tech chain. In the future, Capri suggested the emergence of two competing AI ecosystems—American and Chinese—with different standards, infrastructure, and governance rules.
Earlier, Nvidia, Meta, Microsoft, and 22 other companies urged U.S. authorities not to restrict open-weight AI models. The signatories stated that the country’s leadership depends on an open ecosystem, not a single closed model.
In November 2025, Anthropic announced a $50 billion investment in American AI infrastructure.
