Jan. 27 was a wild day in the stock market as investors reacted to Chinese start-up DeepSeek's R1 "reasoning" model that competes with OpenAI's o1 model.
Semiconductor stocks Nvidia (NVDA +1.68%), Broadcom, and Taiwan Semiconductor Manufacturing -- which constitute over $5 trillion in combined market cap -- fell between 13% and 17% in a single session.
But outside of chip stocks and some artificial intelligence (AI)-focused companies, the rest of the market largely held up. In fact, Apple (AAPL +2.65%) jumped 3%. And Meta Platforms (META +1.81%), which is one of the largest investors in AI, gained 2%.
Here's a look at how markets are reacting to DeepSeek's impact on AI and why Meta Platforms stands out as a top AI stock to buy now.
Image source: Getty Images.
AI efficiency improvements
Reports indicate that DeepSeek was able to build its model at a far lower cost than OpenAI's solution, which led to fears of slowing demand for AI infrastructure like Nvidia graphics processing units (GPUs) or Broadcom's application-specific integrated circuits. The news marks a potential step change in the way AI investments are viewed.
The Donald Trump-backed $500 billion Stargate AI project is tasked with building out data centers to support AI workflows. Microsoft is on track to spend around $80 billion on AI in fiscal 2025. If highly sophisticated AI tools can be built fairly inexpensively, more scrutiny may be warranted on AI spending. And investors won't applaud a company just because it is throwing money at AI.
However, a subsequent 9% rebound in Nvidia the following day on Jan. 28 came as some investors and analysts viewed efficiency improvements as a net positive, arguing that overall compute demand could increase if AI became more affordable for smaller companies.
A similar comparison would be the price of an iPhone over time. When the iPhone first came out in 2007, it cost $499 for the 4 GB version and $599 for the 8 GB model. Inflation-adjusted, that cost is closer to $780 for the 4 GB and $930 for the 8 GB. But the new iPhone 16 starts at just $799, so the cost of an iPhone has stayed about the same since its release despite product innovation and exponential storage increases. Rather, Apple has grown by several folds since then, both in terms of earnings and the stock price, because the iPhone became a consumer staple and Apple expanded its ecosystem across accessories and services. The same could occur with Nvidia. If the market for its products grows, Nvidia won't need to rely so much on pricing power, which could benefit the company long-term.
As of market close on Jan. 24, Apple was the worst-performing component of the Dow Jones Industrial Average year to date. So a rebound in Apple, which may be viewed as less of a pure AI play than companies like Nvidia, makes some sense. But what may come as a surprise is Meta, which blasted to a new all-time high on Tuesday.
NASDAQ: META
Key Data Points
Meta is a compelling AI play
In October, I predicted that Meta would be worth more than Alphabet and Amazon by 2026, which I still think will happen.
Meta is one of the clearest examples of AI fueling rapid growth and margin expansion. AI is instrumental in Meta's algorithms for delivering engaging content to users so they stay on the platform longer and, therefore, see more ads. Meta AI is also helpful for creators to customize content, engage with their audiences, and more.
So although Meta is spending a substantial amount of money on research and development (R&D), it is still growing its revenue and operating income at a breakneck pace. Meta has operating margins around 40%, even considering around 27 cents for every dollar in revenue goes toward R&D.
META Revenue (TTM) data by YCharts
The results are even more impressive once factoring in steep (and growing) losses from Meta's Reality Labs division, which makes hardware and metaverse software, and invests in AI.
Meta has a price-to-earnings (P/E) ratio of just 31.8 and a forward P/E of 26.5, which is a reasonable valuation for a high-octane growth stock. But again, the valuation is even more attractive once considering Reality Labs' impact on earnings.
In the first three quarters of 2024, Meta lost $12.76 billion on Reality Labs and reported net income of $41.52 billion. So if you take out Reality Labs, the amount that investors are paying for the rest of Meta (the advertising business) is actually dirt cheap.
Investing in AI stocks
The sell-off in Nvidia is a painful reminder of the importance of knowing what you own and why you own it, and aligning your investments with your risk tolerance.
Nvidia is a phenomenal, highly innovative company. But its value is also based on continued rapid earnings growth. Meta isn't expected to grow earnings as quickly as Nvidia and is a better value.
Still, Nvidia could be worth owning if you have a high risk tolerance and believe in the long-term investment thesis. But the stock could continue to be highly volatile while investors re-evaluate Nvidia's role in the future of AI.
