
Has China just burst an AI bubble?
You gotta love it when a party ends and the lights come on. Blinking and adjusting to the bright lights, you begin to see clearly that which was previously blurry and obscured.
One of the most repeated mistakes equities investors make is the assumption that their poster-child industry leader will always remain so. Nobody ever seems to assume that competitors will emerge and take a significant share from the first mover. That failure leads to sky-high price-to-earning (P/E) ratios and unrealistic pricing.
The recent emergence of DeepSeek, a small but innovative Chinese artificial intelligence (AI) startup, has just shaken the U.S. AI leadership narrative. DeepSeek’s revolutionary reasoning model, R1, is reported to have stunned the industry with its comparatively excellent performance over leading American AI models like OpenAI’s ChatGPT, Anthropic’s Claude, and Meta’s (NASDAQ:META) Llama.
If the Chinese demonstration is to be taken at face value (there is always the possibility of Chinese deception and subterfuge), the development raises serious questions about the sustainability of the U.S.-dominated AI boom, particularly for companies like Nvidia (NASDAQ:NVDA), whose fortunes are deeply intertwined with the sector’s growth.
Whenever financial market performance and valuations hinge on a handful of high-performing stocks, even a minor disturbance to the narrative can lead to dramatic consequences.
Monday’s global AI sell-off wiped out over US$1 trillion in market value, and Nvidia, a central character in the AI-driven growth story, saw its valuation plummet by nearly 17 per cent. Such volatility is a wake-up call for investors to reassess their assumptions about the AI market’s future. And keep in mind, as we have reported here previously, the biggest use-case for AI at the moment appears to be companion bots.
We wrote here in 2023 questioning AI revenue models and whether downstream companies can meet optimistic investor assumptions about monetization.
The DeepSeek disruption
DeepSeek’s R1 model has drawn attention not just for its performance but for its cost-effectiveness. Unlike U.S.-based companies that have poured billions into building proprietary AI models, DeepSeek reportedly achieved groundbreaking results with a mere US$6 million investment. Moreover, the startup’s decision to release its model as an open-source platform threatens to democratise AI in a way that could destabilise the current market dynamics upon which the current lofty valuations are based.
The development poses a direct challenge to Nvidia, whose high-end, US$ 40,000-per-unit graphics processing units (GPUs) have become the backbone of AI training worldwide. While Nvidia claims its hardware will remain essential for advanced AI processes, the emergence of cheaper, more efficient alternatives could disrupt demand for its products.
A familiar rivalry
This isn’t the first time the U.S. has faced a technology race. There was the Soviet Union’s success in launching the first satellite into space and frequent challenges to U.S. military superiority. Today, China may have taken a similar but perhaps temporary lead in AI development, potentially beating the U.S. to a more cost-effective and scalable AI solution.
For investors, the outcome probably doesn’t matter. What matters is the narrative has changed, and so growth assumptions need to include some possibility of competition.
The economic stakes
The Trump administration’s ambitious US$500 billion Stargate initiative – a five-year plan to bolster AI infrastructure – relies heavily on the assumption that U.S. companies will continue to dominate the AI sector. If China’s DeepSeek can outperform at a fraction of the cost, it casts doubt on the rationale for such massive investments and forces a reevaluation of America’s competitive strategy. Investors, too, must grapple with the potential for a fundamental shift in market dynamics, as DeepSeek’s approach could render existing models and hardware less relevant.
Market fragility and investor reactions
The U.S. stock market gains have been concentrated in a handful of high-performing companies. Any reappraisal of their prospects leaves the market vulnerable to rapid downturns, as demonstrated by Nvidia’s sharp sell-off following concerns about its long-term growth prospects. Some analysts argue that the market may simply be adjusting to more rational valuations after a period of exuberance. Others, however, warn that this could mark the beginning of a more significant shift in the AI industry’s trajectory.
What lies ahead?
DeepSeek’s rise introduces both challenges and opportunities. While its cost-effective approach could undermine demand for Nvidia’s high-end GPUs, it might also expand the overall AI market by making the technology more accessible.
Over 30 years, I have warned investors that predicting winners in technology is nearly impossible, and the current duality underscores the complexity of predicting the long-term impact on U.S. AI stocks.
While one bad day in the markets doesn’t spell the end of the AI boom, it highlights the vulnerabilities of a highly concentrated market.