AI bubble burst? What's next
After reaching new record highs in recent weeks, markets plunged on worries that cheaper AI competition from China could pose risks to U.S. tech companies.1 What caused the market reaction? Let’s dive in.
What caused the tech market crash?
Much of the market run-up over the last two years has been driven by AI mania, with U.S. companies leading the charge. However, this narrative faced a test when DeepSeek, a Chinese tech startup, released an advanced (much cheaper) AI model. DeepSeek claims it was able to develop its advanced model for just $6 million using fewer hard-to-find computer chips, compared to the hundreds of billions collectively invested by U.S. rivals to develop their own AI models.1 If the buzz around DeepSeek R&D numbers turns out to be more than just hype, it raises questions about the efficiency and competitiveness of U.S.-based AI firms.
Why are AI models so expensive to train?
That is the trillion-dollar question. The rising costs of AI development have been a key focus for investors. Capital spending on AI model development is soaring, driven by the cost of the massive computing power needed to analyze data.2 The chart below shows the estimated costs to train some of the major models released over the last few years.
One major tech CEO estimated that training advanced AI models could cost anywhere from $10 billion to $100 billion.3 In contrast, DeepSeek claims to have developed its model for a fraction of the cost by using innovative ways to process data using fewer resources. If their approach is validated, it shows that there may be alternative paths to AI innovation that require less upfront investment.
What will this AI “space race” mean for U.S. AI companies?
While the long-term potential for AI could be massive, it’s still a very new technology with a rapidly evolving landscape and not a lot of return to show for substantial investments.4 If DeepSeek’s approach can be replicated, the disruption could benefit the sector by making it faster and cheaper to release new models. The new pressure would also force competitors to become more efficient in their operations.
Will the tech selloff trigger a bear market?
Market adjustments like this are not unusual, particularly after periods of strong performance. Stocks have been riding high on soaring tech firm valuations, and the pullback could be the reset the sector needs to return to Earth. Looking ahead, we can expect more volatility as investors adjust expectations and digest new data. Bear markets are always a risk, and it’s wise to stay flexible and be prepared for one to strike. That said, we see plenty of growth opportunities ahead despite the short-term uncertainty.
As always, we’re watching closely and monitoring trends. We’ll reach out directly with any specific strategy changes as needed.
Sources
1. CNBC, 2025 [URL: https://www.cnbc.com/2025/01/
2. Stanford, 2024 [URL: https://aiindex.stanford.edu/
3. IBD, 2024 [URL: https://www.investors.com/
4. Goldman Sachs, 2024 [URL: https://www.goldmansachs.com/
Chart sources:
Stanford, 2024 [URL: https://aiindex.stanford.edu/
CNN, 2025 [URL: https://www.cnn.com/2025/01/