U.S. AI dominance rests on infrastructure, not models—and that's harder to challenge

The article challenges a common misconception in AI discourse: that dominance is determined by model performance or cost. The emergence of low-cost models from Chinese companies like DeepSeek has led many observers to conclude the competitive gap is narrowing. However, the author argues that beneath the headline competition over model capabilities lies a deeper structural reality—one rooted in the capital-intensive infrastructure that all AI systems ultimately depend on. Nvidia's 85% control of AI chip production, combined with its proprietary CUDA software layer, creates what the article calls 'high switching costs' for anyone attempting to shift to a competitor, effectively locking the entire ecosystem into American technology. This infrastructure advantage is reinforced by a "commercial-military symbiosis" between Silicon Valley and Washington: the Pentagon's FY2027 budget earmarks more than $54 billion for autonomous warfare systems, and major U.S. tech firms have signed binding contracts providing 'all lawful use' access to their AI tools for defense purposes. The scale of this integration is enormous—ten of the world's largest companies by market cap are American tech firms that comprise this ecosystem, representing commercial concentration in the trillions of dollars.

The control over undersea fiber-optic cables adds a second layer of structural dominance. Silicon Valley firms own 70% of usable undersea cables in 2026, and Washington has demonstrated willingness to use this control: in 2020, U.S. regulators blocked the Hong Kong segment of the Pacific Light Cable Network, forcing Google and Meta to abandon a direct U.S.-Hong Kong link, leaving 13,000 kilometers of cable unused on the ocean floor. Even Chinese frontier labs, despite training on domestically-hosted infrastructure, remain dependent on American undersea cables for global operations. The article also highlights the CLOUD Act—a legal mechanism that allows U.S. authorities to access data held by American cloud providers anywhere in the world, providing Washington with 'incredible leverage' on the AI ecosystem and access to detailed digital trails of AI activity. While the article acknowledges that Taiwan and South Korea manufacture critical semiconductors and that China is rushing to build parallel networks through its 'digital silk road,' it concludes these efforts are unlikely to fully offset American dominance in the near term. The real structural shift would come only when a competing, alternative AI ecosystem—likely Chinese-led—fully emerges with its own standards, infrastructures, and governance, but until that happens the article argues the U.S. maintains firm control over the global AI landscape.