Trillion-dollar tumble: How China’s DeepSeek app rattled Wall Street
Trillion-dollar tumble: How China’s DeepSeek app rattled Wall Street
The Nasdaq plunged as DeepSeek’s low-cost AI questions Nvidia’s stranglehold on AI chips.
The Chinese could not have wished for a more satisfying form of revenge on the U.S. than to strike where it hurts most: the stock market. And the timing couldn’t have been more impeccable—just one week after Donald Trump re-entered the White House. A “small” app, reportedly developed for only $5 million, managed to not only upstage the billions invested by American Big Tech in large language models but also wipe over a trillion dollars off Wall Street in mere hours.
If it weren’t so financially painful, it might almost be amusing. The drama was orchestrated by DeepSeek, a company founded by Chinese entrepreneur Liang Wenfeng, who initially aimed to develop an algorithm to identify patterns influencing stock prices, according to the Financial Times. The Nasdaq tech index suffered the heaviest blow, falling nearly 3% at market open, while the S&P 500, increasingly dominated by tech stocks, dropped 2%. Only the Dow Jones remained relatively stable, declining by just 0.2%. Midway through the trading day, the losses began to moderate.
Nvidia, the crown jewel of the AI boom and a major player with a large development center in Israel, emerged as the primary casualty of the Chinese "tsunami," plunging by almost 17%. Nvidia recently became the largest company by market value on Wall Street, overtaking Apple and reaching $3.5 trillion—a valuation largely based on the belief that generative AI cannot exist without its processors. Over the weekend, however, the Chinese challenged this notion by unveiling a language model they claim rivals ChatGPT and similar platforms, all developed in just a few months at a cost of $5 million. DeepSeek reportedly utilized around 2,000 Nvidia chips for training—compared to the tens of thousands used by companies like OpenAI or Google.
Yet, skepticism is warranted. This is, after all, China—a nation with a long history of concealing critical details about state involvement in private projects and even falsifying macroeconomic data. The U.S. and Israeli tech communities have already criticized DeepSeek’s app, citing concerns about giving it access to smartphones—this, ironically, comes just after a fierce battle in the U.S. to restrict TikTok.
The story of a young developer creating a breakthrough with minimal resources and without significant Nvidia hardware has captivated imaginations but also raised alarms. Over the weekend, Marc Andreessen, a co-founder of venture capital firm Andreessen Horowitz, called this development AI’s “Sputnik moment,” referencing the Cold War era when the Soviet Union’s successful launch of a satellite shocked the U.S. into accelerating its space program. However, by Monday, Andreessen and others had tempered their initial reactions, acknowledging that China’s complete dominance in AI is still far from certain. After all, the U.S. turned the Sputnik shock into an opportunity, ultimately landing on the moon in 1969.
For the stock market, this was likely more of a hysterical moment than a historic one. The markets are currently inflated, especially Nvidia, where high expectations around AI have driven valuations to extreme levels. It doesn’t take much to prompt corrections in such an environment. The AI revolution has also boosted non-tech companies, such as electricity providers expected to meet the increased energy demand from data centers. Consequently, stocks like Vistra and Constellation Energy fell sharply, dropping 15–18% in a single day.
What’s next? As users continue testing DeepSeek’s app, it may reveal limitations akin to cheap imitation products—impressive at first glance but incapable of sustaining performance under strain. The current enthusiasm seems centered around private users rather than the corporate market, where the big money in AI applications lies. Western corporations may remain hesitant to adopt a Chinese AI solution for mission-critical operations.
Even if DeepSeek’s breakthrough proves genuine and the model functions well, the impact isn’t necessarily all negative. Nvidia’s dominance has started to resemble a monopoly, with the company charging premium prices for its highly sought-after processors, often at the expense of customers’ profitability. If AI development becomes more affordable and accessible, smaller businesses—not just giants like Google, Microsoft, and Amazon—could benefit significantly.
For countries outside the U.S., particularly Israel, this moment underscores an important reality: the AI race is far from over. Israel has built its reputation as the "startup nation" on the ability to innovate under constraints. If a cheap, audacious model can challenge American dominance in generative AI, who better to rise to the occasion than Israeli entrepreneurs?
