DeepSeek just shook up the artificial intelligence (AI) world in the biggest way since OpenAI launched ChatGPT in late 2022. The Chinese company's new R1 large language model (LLM) reportedly matches or beats OpenAI's o1 model on some benchmarks. DeepSeek is much more cost-effective, too.
This news caused upheaval among many AI stocks. Nvidia's (NVDA 0.40%) share price plunged 17% on Monday. Interestingly, though, not every AI stock took a beating. Shares of Meta Platforms (META -2.76%) even moved a little higher. Is Meta a better AI stock to buy than Nvidia after DeepSeek's stunning R1 debut?
The case for Meta
There's a simple explanation for why Nvidia investors panicked in the wake of the DeepSeek tsunami while Meta shareholders remained calm. DeepSeek's R1 model doesn't seem to present nearly as much of a threat to Meta's business as it does Nvidia's.
NASDAQ: META
Key Data Points
Meta's chief AI scientist, Yann LeCun, posted on Threads on Jan. 24, 2025, explaining his view of DeepSeek's R1 launch. He said the correct interpretation is that "open source models are surpassing proprietary ones." LeCun added:
DeepSeek has profited from open research and open source (e.g., PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people's work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.
If LeCun's take is right, Meta could benefit from DeepSeek's advances. The company has its own open-source AI model, Llama. Meta CEO Mark Zuckerberg posted on Facebook on the same day as LeCun's Threads post, "In 2025, I expect Meta AI will be the leading assistant serving more than 1 billion people." He also predicted that "Llama 4 will become the leading state of the art model." Importantly, Zuckerberg's optimistic comments came after DeepSeek's R1 release.
Advances that lower the cost of training and deploying AI models should help Meta, which is spending a lot of money on data centers for its internal AI use. Zuckerberg's Facebook post even mentioned his company is "building a 2GW+ datacenter that is so large it would cover a significant part of Manhattan." Incorporating good ideas from other open-source AI models could also make Llama better.
Meanwhile, Meta stock is cheaper than Nvidia's using one key valuation metric. Meta's shares trade at a forward price-to-earnings ratio of 24.5 versus a multiple of 33.3 for Nvidia.
The case for Nvidia
One argument for Nvidia is that there's been a major overreaction to DeepSeek's news. LeCun made this case himself, posting on Threads on Monday that the market reactions to DeepSeek "are woefully unjustified."
NASDAQ: NVDA
Key Data Points
Gabelli Funds portfolio manager John Belton also noted that some reports about DeepSeek's R1 model are "misleading." For example, he dismissed the idea that it would take only $6 million to "re-create DeepSeek." Belton pointed out the unknowns related to the Chinese company's efforts, including whether it "took shortcuts potentially without license."
There's plenty of discussion about the Jevons paradox, too. This economic principle maintains that a resource will be used more rather than less when it becomes more efficient. With this in mind, it could be likely that the demand for Nvidia's GPUs will increase instead of decrease as a result of DeepSeek's innovations.
It's highly unlikely that the DeepSeek news is causing any organization to cancel its orders of Nvidia's new Blackwell chips. Nvidia will almost certainly post spectacular results in its quarterly updates this year.
When the initial shock of DeepSeek's R1 launch wears off, investors could again flock to Nvidia. Buying the GPU stock on the dip has proven to be a smart move in the past.
A gut call
So is Meta the better AI stock to buy now after DeepSeek's R1 debut or is Nvidia? My gut instinct leans toward Meta -- for now, at least.
Gabelli's Melton thinks that the DeepSeek story "creates a new bear case" for Nvidia over the near term. He also believes that it could take a while for the stock to come back. I suspect Melton is right. I also fully agree with him that companies such as Meta that use AI extensively will benefit if the costs of inferencing decline as a result of R1.
However, sometimes a shot across the bow is just what a well-run company needs to step up its game even more. I predict that's exactly what Nvidia will do. Nvidia remains a great long-term pick, in my view.