TheChek News Flow

What to Know About China’s DeepSeek AI

The Chinese upstart says it has trained high-performing AI models cheaply, without using the most advanced chips

DeepSeek has Silicon Valley in awe and investors in a frenzy.

The Chinese artificial-intelligence upstart has shot to prominence after saying it had trained high-performing AI models cheaply, without the most advanced chips.

Tech stocks sank Monday as investors fretted about the implications, wiping some $1 trillion from the stock market’s value. Nvidia, which makes the chips at the heart of the AI boom, closed down 17%. Nvidia and other stocks that swooned recovered some ground Tuesday.

Here’s what you need to know about DeepSeek:

DeepSeek is a Chinese AI company, which just over a week ago launched its latest AI model, which it calls R1. The company said the model was particularly good at problem solving, performing on par with OpenAI’s o1 reasoning model—but at a fraction of the cost per use. A DeepSeek app is currently top in iPhone download rankings for the U.S.

The conventional thinking was that AI companies needed expensive, leading-edge computer chips—such as those made by Nvidia—to train the best systems. That has justified huge spending by the biggest U.S. tech companies, such as Alphabet and Meta Platforms.

Just last week, companies including SoftBank, Oracle and OpenAI pledged to spend $500 billion to build new AI infrastructure in a venture they call Stargate.

DeepSeek’s use of less advanced chips—combined with innovative model-training techniques—is now raising questions about the investment case for stocks seen as big winners from AI.

In addition, DeepSeek released its R1 model as open source. That means other companies can pick up and adapt the model for their own use, potentially opening the door for other cheap AI alternatives.

It is hard for DeepSeek to buy cutting-edge chips because of U.S. export controls, intended to hinder Chinese organizations from developing innovative AI for military purposes.

That DeepSeek appears to have been able to achieve state-of-the-art performance suggests that those export controls may be ineffective—either because U.S.-designed chips aren’t necessary to make the best AI models, or because those chips are somehow making it to China in sufficient quantities anyway.

DeepSeek grew out of a hedge fund co-founded by Liang Wenfeng that uses AI to find profitable trades in financial markets. Liang, a math geek who caught the investing bug, started writing AI algorithms to pick stock as a student. He founded his hedge fund, called High-Flyer, with two college friends in 2015. It now manages some $8 billion, making it one of China’s largest quantitative funds.

No. DeepSeek is a closely held Chinese company led by Liang, which isn’t traded on the stock market.

The researchers behind DeepSeek say that they tested R1 against some of the top AI models from OpenAI—and found that it was very competitive. Those evaluations include one developed by OpenAI itself that includes computer-programming tasks that an AI model must complete on its own, such as patching a bug in a given piece of software. R1 performed on par with a version of OpenAI’s reasoning-focused model, called o1, and outperformed an earlier one called o1-mini.

DeepSeek published costs for using R1 that were an order of magnitude below those charged by U.S.-based companies for their most sophisticated models.

Users have said R1’s writing and problem-solving skills are impressive but some note that the model performed worse than rivals on specific types of problem solving. In coming weeks, more third-party testing should give a better understanding of how well R1 really performs. And users can test it out themselves, too.

In a paper published last week, the Chinese researchers behind DeepSeek said their new model would sometimes suddenly stop and realize it should re-evaluate its initial approach to a problem, and allocate more thinking time to do so. They described the behavior as the model having an “Aha!” moment.

Advertisement

“Rather than explicitly teaching the model on how to solve a problem, we simply provide it with the right incentives, and it autonomously develops advanced problem-solving strategies,” the researchers wrote.

What have U.S. AI companies said about DeepSeek?

OpenAI CEO Sam Altman on Monday called R1 “an impressive model, particularly around what they’re able to deliver for the price,” in a post on X. He also said that it was invigorating to have a new competitor and that his company would move up some of its product releases.

You may also like

Checkbox

Copy LinkCopy Link

Copy EmbedCopy Embed

Share on FacebookFacebook

Share on TwitterTwitter

0:27ADVERTISEMENT

Tap For Sound

President Trump said the launch of a low-cost Chinese AI model should be seen as a wake up call for U.S. industries. Trump also announced plans to impose new tariffs on semiconductor imports. Photo: elizabeth frantz/Reuters

Why is Nvidia taking such a big hit?

Nvidia has been one of the biggest winners in the AI boom because its chips have almost exclusively powered the training and in many cases the day-to-day running of the most powerful existing AI models. Nvidia—and its investors—have bet heavily that new generations of those cutting-edge chips will be necessary to develop the most powerful AI models. DeepSeek’s success suggests that Nvidia’s lead on AI chip development may not be as big as thought, or as crucial to developing new AI models.

In a statement, Nvidia called DeepSeek “an excellent AI advancement” and said the work required for it to come up with answers, called inference, “requires significant numbers of Nvidia [chips] and high-performance networking.”

Is DeepSeek a disaster for stocks linked to the AI boom?

Not everyone thinks DeepSeek has upended the AI-infrastructure industry. While DeepSeek might have found a way to cut AI training costs, AI demand keeps surging, and tech companies still need more computing power, wrote Stacy Rasgon, a Bernstein semiconductor analyst.

“Is DeepSeek doomsday for AI buildouts?” Rasgon and his colleagues wrote in a report on Monday. “We don’t think so.”

What does DeepSeek mean for the global AI race?

DeepSeek’s success building an AI model could rebalance the global playing field when it comes to AI development—and that has cheered some countries outside the U.S.

Government officials in France, for instance, said Monday that DeepSeek shows that agile companies with clever techniques still compete in the AI race, even if they have less money or limited access to the best AI chips. In other words, opportunities remain for those outside of the U.S. to catch up to Silicon Valley.

“The message is that we can compete,” said an official at France’s Élysée Palace, noting that raw computing power may no longer be the determinant of who wins in AI.

To be sure, DeepSeek also is a warning to other parts of the world. French startup Mistral AI has made its name on being a smaller, more efficient competitor to U.S. companies like OpenAI. Now it will have to keep up with DeepSeek and others that use its models, too.