Beijing wants you to believe the self-reliance story. Semiconductor supply chains that answer to no one in Washington. Artificial-intelligence models built on homegrown chips and homegrown code. An economy that innovates on its own terms, answers to its own rhythm, and needs nobody’s permission to lead. When DeepSeek made its debut, the party treated it as proof — vindication of a decade of strategic investment. When Moonshot AI’s Kimi followed, the message sharpened: China doesn’t just compete, it does so with cheap, open-source models built to win.
Except the stock market didn’t get the memo. When Chinese equities sold off hard this month, regulators reached for a familiar scapegoat: “risk imported from abroad.” Not a domestic policy failure. Not a reckoning on valuation. Foreign risk — the kind that, by definition, a truly self-reliant economy would be insulated from. The admission is remarkable precisely because it was involuntary. Beijing’s own regulators conceded, however obliquely, that the fate of Chinese equities now runs through Wall Street.
The chain of dependency is even more embarrassing than regulators let on. Chinese investors have little money in Korean stocks. Korean capital is not a major force in Shanghai or Shenzhen. Yet when South Korean chipmakers — riding the same memory-chip windfall as China’s own ChangXin Memory Technologies — ran through a speculative frenzy and then cracked, they rattled Chinese tech too. Not because any money moved. Because sentiment did. Chinese tech valuations, some analysts argue, are more stretched than Korea’s, resting on a domestic-substitution story that assumes China’s champions deserve to trade like Nvidia’s ecosystem without Nvidia’s numbers behind them. That is not independence. That is dependence wearing a different flag.
Consider CXMT. The chipmaker went from years of losses to blowout profits almost overnight — not because it executed some business transformation, but because a global supply shortage handed it a windfall that everyone else is also scrambling to exploit. Stellar earnings now barely move the stock. The rally already priced in the blowout. This is what happens when valuation runs on narrative rather than fundamentals: the market buys the story before the company can write it.
Huijin, the state-investment arm that functions as China’s plunge-protection team, stepped back into the market last week, buying broad-based ETFs to arrest the slide. That is what sovereign wealth does when market fundamentals fail to hold. But the real repair work — stabilizing the tech names that actually cratered — depends on sentiment holding steady in the United States and South Korea. The state can prop up indices. It cannot prop up the foreign market psychology that Chinese tech valuations have quietly, and increasingly, anchored themselves to.
None of this means the AI story out of China is fabricated. The talent is real. The ambition is real. DeepSeek’s Liang Wenfeng told investors that AI could eventually account for 10% of global GDP and warned that any company hoarding that prize would be “abandoned by history.” That framing — AI as too large for one company or country to monopolize — lines up with Beijing’s push to flood the world with cheap, open-source Chinese models. It is a coherent strategy. The problem is not the strategy. The problem is the price tag.
The real irony of Beijing’s technological-independence push is this: the more successful it appears — soaring valuations, IPO plans for DeepSeek and Moonshot, talk of overtaking the United States — the more it is being financed and validated by a market logic developed in America and tested in South Korea. China wants the world to look at its AI industry the way it looks at America’s. But if China’s own regulators are to be believed, China’s market is still the one doing the looking. Independence, it turns out, is easy to declare. It is much harder to trade on.