The market had a rocky start to the week as the artificial intelligence (AI) industry experienced a substantial setback on Monday. The tech-focused Nasdaq Composite index fell 3.1% on the day, but many AI-related companies experienced a much larger drop. At the heart of the selloff was DeepSeek, a Chinese AI company that has burst onto the scene.

Here is an extremely high-level summary of what has transpired:

• DeepSeek has recently launched a series of models that are said to be on par with U.S. rivals in common performance tests.
• The company reportedly achieved this substantially faster and cheaper than their U.S. counterparts and without access to the leading-edge processors due to trade restrictions that prevent Chinese access.
• DeepSeek was able to leverage open-source data, allowing it to focus on creatively iterating others' prior advances rather than going through the expensive process of building from scratch.
• DeepSeek reportedly spent $5.6 million developing their model, compared to the $100 million OpenAI spent developing ChatGPT 4.0.

CNBC published a good explainer video on their YouTube channel if you’d like to dig deeper. 

The key item driving the market’s response is the question of whether we REALLY need to be funneling hundreds of billions of investment dollars into AI infrastructure. Can DeepSeek’s approach, which required very little capital and no access to the most advanced computing power, simply be replicated?  If the answer to that question is yes, it spells big trouble for companies like NVIDIA that have been riding that wave of investment.

In response to the development, pretty much any company involved in the AI industry came under significant pressure. NVIDIA specifically closed 17% lower, shaving $600 billion from its market value on its way to becoming the largest single-day stock plunge in history. That number is so large it’s hard to really wrap your mind around. To put it into better context, consider the table below. NVIDIA’s loss yesterday would be the equivalent of all eleven of these stocks abruptly ceasing to exist.

Source: Google Finance as of 1/28/2025.

What is most interesting about this story is the fact that we have seen this type of thing play out before. In the late 1990s, everyone knew that the widespread adoption of high-speed internet would be an absolute game changer. Investment dollars poured into any entity that claimed some sort of involvement. Many of those were ‘.com’ stocks with silly business models and no revenue that unsurprisingly went to zero.

The more interesting comparison for NVIDIA is to look at what happened to those profitable firms that facilitated the infrastructure needed to move the masses online. The market priced those businesses as if the massive initial infrastructure buildout would need to continue forever. By the bursting of the Technology Bubble, it was clear that there was substantial excess capacity. The demand for things like fiber optic cables waned, and the business values of those companies cratered.  

This all begs an important question for investors: Is it better to own an exciting business whose value is largely dependent on the potential for future growth or a more stable business whose value is supported by fundamentals today? As the table below illustrates, investor excitement about the potential growth for NVIDIA has shot its valuation to the moon. Even the Magnificent 7 stocks (which includes NVIDIA) look cheap by comparison.

Source: Morningstar Direct, as of 12/31/2024. P/E ratios are based on the trailing 12 months. S&P weights based on the iShares S&P 500 ETF (IVV).

Hot stocks like NVIDIA and the rest of the Magnificent 7 generate significant excitement. The excitement leads investors to compete for the opportunity to buy. The buying pressure pushes prices higher. Higher prices generate even more investor excitement, and the cycle repeats.

The risk for investors is that the stock’s valuation, and therefore the hurdle the company’s operations must clear, rises with the stock’s price. If the fundamentals don’t eventually justify the excitement, a large correction will ensue. Conversely, as investors give too much attention to certain ‘loved’ stocks, they simultaneously ignore some ‘unloved’ stocks. The lack of excitement can depress the valuation of these other companies and ultimately lower the bar that their operations must clear. This is a common explanation for the ‘value effect.’ Historically, investors have been better off buying businesses that don’t require massive future growth to justify their current valuation.

So how will the story play out for NVIDIA? The AI boom seems to have a lot of parallels to the internet boom. With such a high valuation on NVIDIA, there is a real risk that investors have been looking at explosive short-term revenue growth as if it will go on forever. The arrival of DeepSeek seems to have exposed this possibility, and investors have, at a minimum, taken a pause. Even if this is the case, don’t underestimate the ability of investors to ignore reality. The Nasdaq 100 Index lost 83% of its value during the bursting of the Technology Bubble from 2000 to 2002. During that period, there were half a dozen rallies in which stocks recovered about 15% on average.

It’s also possible that the DeepSeek story has been overblown. Perhaps chip demand will continue to explode as a larger consumer base interacts with AI-powered apps more frequently. I also wouldn’t rule out the possibility that DeepSeek was, in fact, able to get their hands on a large quantity of cutting-edge chips but is hiding that fact to ensure continued access despite the trade restrictions. 

As I write this on Tuesday, the stock has recovered some of its Monday losses. How things progress from here is anyone’s guess.

Source: MarketWatch

IMPORTANT DISCLOSURE INFORMATION