The stock market was down Monday, Jan. 27, 2025, with an especially sharp sell-off in the stock of the AI-chip making giant Nvidia (NVDA), which many investors suddenly perceive to be under pressure from the Chinese company DeepSeek. The company claims to have developed a rival to AI chatbots, like OpenAI’s ChatGPT, but at a fraction of the cost. Below are some quick, preliminary thoughts from me and our team on this rapidly developing story.
Early thoughts
It’s hard not to see the growing AI race between the U.S. and China through a geopolitical lens. Many of today’s headlines certainly do. Whether this is a “Sputnik moment” or not remains to be seen. It’s ultimately unknowable, presently, whether China is catching up to, on par with, or ahead of the U.S. in the realm of artificial intelligence. And while U.S. technological hegemony may be important in terms of our national security, especially in the realm of AI, it’s unclear what the investment implications would be of a new leading entrant from China — good, bad, or somewhere in between.
What we do know is this
First, we should be careful not to take anything coming out of China at face value — certainly anything having to do with financial statements or economic data is highly questionable. Claims that DeepSeek has built a more efficient LLM for only $6 million should be taken with a huge grain of salt. It’s possible, even probable, that DeepSeek isn’t properly accounting for prior R&D investment.
Second, whether DeepSeek’s new model (its first release, “R1”) truly outperforms the best models available from U.S. companies (such as OpenAI’s ChatGPT) needs to be pressure-tested by experts. Even if it does, the race isn’t over — in fact, DeepSeek’s technological advancements are likely to be quickly vetted, replicated, and improved upon by U.S. companies. Such is the nature of competition. This is not the end of history.
Third, is that the lesson in today’s sell-off has nothing to do with DeepSeek’s new open-source AI model and everything to do with proper diversification. While Nvidia declined more than 15% on Monday, the S&P 500 was down only 1.3%. Consider that on Monday, both value stocks and bonds were up while small cap and non-U.S. stock losses were relatively mild. Diversification worked. (As an aside, so much for the claims that bonds no longer provide diversification benefits to stock portfolios.)
The big takeaway
Leaders in all industries rise and fall. Rarely does any one company remain an industry leader for long. IBM, General Electric, Sears, Yahoo, Netscape, Intel, Sun Microsystems, WorldCom and so many others have all been surpassed (or permanently relegated to the dustbin of history) by newer, more competitive entrants.
We know from real-world historical data that the average company in the market has significantly underperformed the market by quite a bit. Said differently, most individual companies perform poorly compared to the overall market and the way we obtain better returns is by broad diversification, rather than trying to cherry pick the winners. (The majority of U.S. stocks – 55.2% – underperformed the one-month treasury bill for the 30 year period ending December 2020.)1
Therefore, the real lesson to be gleaned from Monday’s headlines is that investors should diversify beyond the market’s most popular companies — including those that are well-loved like Nvidia — no matter how great their respective investment thesis may sound. Today reminds us, yet again, that diversification can be our friend.
Visit the Mercer Advisors website for past insights from our CIO about major market movements and other interesting topics. Not a Mercer Advisors client but interested in more information? Let’s talk.
1 Bessembinder, Henrik; Te-Feng Chen, Goeun Choi; and Wei, John, K.C., “Long-Term Shareholder Returns: Evidence from 64,000 Global Stocks”, Financial Analysts Journal, April 2023.
