I was born and raised near Cleveland, and have witnessed my city by the lake go from economic powerhouse to also-ran. In 1965, Cleveland was the fifth-wealthiest metro in America. Today it ranks 36th. What happened? Many factors, but the most important? Foreign competition.
For a while, geography was destiny. Cleveland sat perfectly between the auto hub of Detroit and the steel hub of Pittsburgh, feeding both. Then Toyota and Honda ate the lunch of a slovenly, complacent Big Three. Foreign steel dumping rewrote the economics of the mills, and the middle of the value chain simply collapsed. Nobody in 1965 believed it could happen. But that's the thing about incumbents; the ground gives way slowly, then all at once.
Which brings me to the AI wars, because the pattern is starting to rhyme.
Cheaper Chinese models are flooding the market and the price gap is not subtle. On output tokens, OpenAI's flagship runs around $30 per million; DeepSeek's comparable rate is roughly $0.28: over 100 times cheaper. And this isn't just a developing-world story. Chinese providers went from under 2% of OpenRouter's traffic a year ago to more than 45% by April 2026. Vercel saw DeepSeek's share of its token usage leap from under 1% to 17% in a single month. One report found that roughly 80% of U.S. AI startups have already wired at least one Chinese open-source model into production. This is happening in Silicon Valley's own backyard: when the startup Lindy found its inference bill had grown larger than its payroll, it moved its agent traffic from Anthropic to DeepSeek, saving millions a year.
Now, brace yourself, because in the coming months we’ll hear a great deal of FUD on this front. Security! (Your data, routed through Shanghai.) Sovereignty! (Do you really want your national infrastructure running on the CCP’s weights?) Loyalty! (We can't outsource the industry of the future.) These concerns are super-legitimate, and every Western government and global enterprise will have to work through them. And this would be an easy argument for the US labs to win; but for the 100-to-1 price difference.
Thus, this cuts to the fundamentals. As Anthropic and OpenAI prepare to go public, we should all be asking the Cleveland question: what happens to the middle of the value chain when someone offers 90% of the capability for a fraction of the price?
The week's most important AI news, curated for you:
