Nasdaq sinks as DeepSeek-linked AI rout crushes Nvidia, down 16%

The Nasdaq led U.S. stocks lower in a broad AI rout, and Nvidia’s slide was the headline you could feel in your portfolio. Nvidia was down 16% as AI infrastructure makers got hit, with many stocks falling by double digits. In other words, this was not a calm, orderly “rotation.” It was a rapid repricing of what investors think the next wave of AI spending will reward, and how quickly those expectations can change.

The spark, per the report, was China’s DeepSeek. That matters because DeepSeek sits in the same conversation as the models, tooling, and compute demand that AI infrastructure companies depend on. When a significant AI entrant from China becomes a focal point for performance and capability, markets tend to ask a brutal question: are we pricing the winners correctly, or are we paying upfront for a future that might not play out on schedule? This time, the market’s answer showed up immediately in the trading tape, pulling the Nasdaq down as AI infrastructure names collapsed.

To understand why a single catalyst can drive an entire AI complex lower, you have to zoom out to how these companies are valued. AI infrastructure businesses, especially those tied to accelerators, software stacks, and the enabling ecosystem, often trade on expectations of sustained demand. That demand expectation is partly technical, but it is also financial and operational. If investors believe frontier model progress will reduce the “effective compute needed” per unit of output, or if they see alternative supply chains and training approaches gaining traction, the market can quickly move from optimism to skepticism. Even if nothing breaks operationally, the valuation math can shift overnight.

This is where “broad rout” is more than just a dramatic phrase. When “many” AI infrastructure makers fall in the double digits, it signals contagion across the sector, not a one-off dip tied to a single earnings miss or product delay. Investors are treating DeepSeek as an update to the competitive landscape rather than a niche development. That kind of repricing can be especially painful for companies that have already been priced for continued leadership and for boards that have to manage the gap between long-range strategy and near-term market expectations.

There is also a policy and regulatory layer that tends to amplify these moves. U.S.-China tech competition has long been entangled with export controls, compliance risk, and shifting rules of engagement for where advanced chips and related technology can flow. Even without new regulations in this specific report, the market’s sensitivity to developments associated with major Chinese AI efforts is structurally high. That means executives in AI infrastructure are not just selling hardware or software. They are also navigating a world where geopolitics can reshape demand, supply, and timelines faster than a traditional product roadmap ever could.

Second-order implications are where boardrooms feel the tremor. When Nvidia is down 16% and the Nasdaq is leading declines, CFOs and investors have to revisit scenarios that seemed settled. For example: What portion of near-term revenue guidance relies on assumptions about global AI deployment speeds? How much of the business is insulated versus exposed to shifts in model efficiency and training approaches? If the market believes alternative pathways are compressing the compute advantage that traditionally accrues to the most dominant training stacks, then pricing power, utilization expectations, and even customer procurement timing can all come under pressure.

For executives at other AI infrastructure makers, the message is not that AI will stop. It is that the market can change its mind fast, and it can do so based on competitive signals from outside your immediate customer base. A DeepSeek-linked narrative is enough to move multiples across the sector, which means leadership teams should be prepared for volatility not just around earnings, but around the perceived trajectory of AI capability and the resulting demand for compute. In this environment, board-level oversight of strategy, risk, and investor communications becomes as important as product execution, because expectations can reset in a single session.