U.S. equities fell sharply on Monday as artificial intelligence infrastructure stocks plummeted across the board. The Nasdaq declined more than peers, driven by deep losses in semiconductor and AI-related holdings. Nvidia, the dominant player in AI chips, dropped 16%, marking one of its worst single days in recent memory.
The sell-off stems from China's release of DeepSeek, a new AI model that challenges assumptions about the capital intensity required to develop competitive large language models. DeepSeek reportedly achieves performance levels comparable to leading U.S. models like OpenAI's o1 while using significantly fewer computational resources and lower training costs. This development threatens the narrative that has driven trillion-dollar valuations across the AI sector.
The revelation spooked investors who had priced in years of sustained demand for advanced chips from Nvidia, Advanced Micro Devices, and other semiconductor makers. If companies can build effective AI systems more cheaply and efficiently, the case for massive ongoing chip expenditures weakens materially. Broadcom, Marvell Technology, and other firms supplying infrastructure to data centers also suffered steep declines.
The market's reaction reflects a fundamental repricing of AI economics. For months, investors assumed that building state-of-the-art AI required ever-greater computational power and thus ever-greater chip purchases. DeepSeek's emergence suggests efficiency gains can substitute for raw compute, potentially reducing the total addressable market for premium processors.
This does not mark the end of AI demand. Rather, the sector is experiencing a correction after perhaps moving too far ahead of fundamental realities. Many AI companies still require substantial infrastructure investments. But the uniformity of that demand, and its growth trajectory, now faces real questions.
Broader market indices sold off on the tech weakness. The S&P 500 and Dow Jones Industrial Average also declined, though less severely than the Nasdaq. Bond yields fell as investors rotated into defensive positioning, lifting Treasury prices.
The DeepSeek development arrives amid ongoing scrutiny of AI's return on investment. Companies have spent hundreds of billions on AI infrastructure without generating commensurate revenue growth. If efficiency improves faster than expected, that gap widens, adding pressure to already-questioned valuations across the space.