In discussions about global rivalry in the field of artificial intelligence, the focus is often on comparing specific algorithms—whether it be developments from OpenAI, Anthropic, or Chinese labs like DeepSeek and Z.ai. However, as I see it, the real advantage of the United States rests not on this. The foundation of leadership is control over capital-intensive infrastructure: from graphics processing units to underwater data highways.
The Ecosystem as a Weapon
The key asset is Nvidia, which holds about 85% of the global market for chips used to train advanced AI models. But it's not just about the hardware. The CUDA software platform has become the de facto standard for developers, creating a high barrier to switching to alternatives. This forms an entire "American AI stack," including hardware manufacturers (Broadcom), cloud giants (AWS, Google Cloud, Azure), and model developers at the level of OpenAI and Meta.
The Defense Circuit and Budget Flows
Equally important is the symbiosis between the technology sector and the defense establishment. A telling example is the Defense Autonomous Warfare Group (DAWG)—a Pentagon structure for autonomous military systems, for which $54.6 billion is allocated in the fiscal year 2027 budget. The potential value of federal AI contracts from the Department of Defense reached $90.7 billion in 2026, although only $7.2 billion has actually been reserved. These figures demonstrate a strategic scale that extends far beyond the civilian market.
Digital Geography and Data Control
A separate front is underwater fiber-optic cables. American corporations are actively building their own routes, gaining leverage over global traffic. The example of the Pacific Light Cable Network is telling: the segment to Taiwan was approved, while a direct connection to Hong Kong was blocked due to risks associated with the PRC. Meanwhile, Chinese labs, even when training models domestically, remain dependent on the global internet for data collection and user services. The CLOUD Act of 2018 additionally gives U.S. authorities the right to request data from American cloud providers, wherever it is stored.
The Limits of Dominance
It would be a mistake to consider this advantage absolute. The production of critical components—chips and servers—is concentrated in Asia (TSMC, SK Hynix, Samsung, Foxconn). However, Taiwan and South Korea are deeply integrated into a supply chain oriented toward the United States, which reduces the risk of them using this position against Washington. In the long term, we are likely moving toward the formation of two competing AI ecosystems—American and Chinese—with their own standards and rules.
My conclusion: Washington wins not through the brilliance of individual algorithms, but through systematic control over the entire chain—from silicon to legal norms. This is a more sustainable advantage than any single model, and it is this that will determine the balance of power over the coming decade.