The first big casualty of the stock market's DeepSeek scare — aside from a few hundred billion dollars in frothy Nvidia valuation — is the AI industry's religion of scale.

State of play: Ever since the advent of ChatGPT two years ago, U.S. tech firms, led by OpenAI, have shared the belief that AI will keep improving as long as we keep throwing more chips, money, power and data at it.

OpenAI CEO Sam Altman preached this gospel last year, writing that machine learning gets "predictably better with scale."

• "To a shocking degree of precision," Altman wrote in his "The Intelligence Age" essay, "the more compute and data available, the better it gets at helping people solve hard problems. ... AI is going to get better with scale, and that will lead to meaningful improvements to the lives of people around the world."

The rest of the industry largely agreed.

• Nvidia's high-end chips became coveted and scarce as its stock price skyrocketed.
• OpenAI's backer Microsoft, its chief rival Google, and cloud giant Amazon all got busy with colossal infrastructure investments.
• A macho race to build bigger, more energy-hungry data centers began. Elon Musk's xAI built a big computing cluster in Tennessee in what he claimed was record time.

But over the past year, the payoffs from this race to scale up have grown elusive.

• Some of the most significant advances in pushing the boundaries of AI have been made not by making it bigger but by building it differently.
• That includes OpenAI's most recent leap forward in the form of its reasoning model, o1.
• This new generation of AI models can solve more complex problems than their predecessors but not because they're bigger or have been trained more. Instead, they strategically use extra time and computing resources while figuring out their answers to users' questions.

Over the past months, DeepSeek, a China-based research lab, has been steadily matching many of OpenAI's achievements using what it says is a fraction of OpenAI's budget.

• The DeepSeek R1 model, released last week, performs comparably to OpenAI's o1. The company has publicly released all its models for free download and use.

Why it matters: Maybe we can get what we want from AI without spending hundreds of billions of dollars on infrastructure or choking the planet with CO2.

The bottom line: As businesses and investors rethink the future, they're squeezing some air out of the AI bubble — and no one likes to see their portfolios lose value.

• But it's not as though OpenAI, DeepSeek and everyone else aren't still going to need to buy Nvidia's chips and build more data centers.
• They just may not need as crazy much as everyone assumed until yesterday.

Early yesterday it looked like DeepSeek might have set off a broad-based market rout, but by the close of trading the carnage was much more selective.

• While Nvidia lost $600 billion of market value in a single day, for instance, Apple gained more than $100 billion.

The bottom line: What's bad for the companies looking to sell AI products is likely to be good for the companies looking to buy them.

• And there are many more of the latter than there are of the former.

Silicon Valley's overnight obsession with DeepSeek has renewed questions about whether it's safe for U.S. companies to even interact with China-linked technology.

Why it matters: Security concerns alone are enough for the biggest U.S. companies to pump the brakes on using DeepSeek's models right now.

• Prompt Security CEO Itamar Golan told Axios that in recent days his U.S. customers — many of which are Fortune 500 companies — were asking how to keep their proprietary data safe from DeepSeek.
• But corporate anxieties won't stop employees from downloading the app on their own to explore its usefulness — just look at how corporate ChatGPT bans went.

The big picture: The U.S. often doesn't take too kindly to companies from adversarial nations gaining popularity on its home field — but DeepSeek's free, open-source options and advanced capabilities could make it too appealing for U.S. companies to completely ignore.

What they're saying: "It's another tool owned by a Chinese entity that will get plenty of the U.S. data," Golan told Axios.

• "People talk with AI about everything, what they like, what they do, what they plan — all will be incorporated in the back end of this model, which is huge."

The intrigue: DeepSeek's open-source approach makes security concerns a bit different than those surrounding TikTok, Huawei or even Russian cybersecurity company Kaspersky, Golan said.

• Typically, the U.S. worries about the potential for a backdoor in products that would let adversaries collect sensitive data about Americans. China-linked companies also face a litany of laws that could force them to hand over key information to Beijing.
• But anyone can inspect R1's model weights, along with the training code from DeepSeek's V3 model, which R1 is trained on.
• Large companies could also use this open-source information to build their own DeepSeek-based apps that are walled off from the startup's servers, Levi Gundert, chief security officer at cyber firm Recorded Future, told Axios.

Zoom in: DeepSeek's links to China also raise a few other unique cybersecurity challenges, experts say.

• Unless they're using a walled version of the product, inquiries sent to DeepSeek are processed on the company's servers in China, according to its privacy policy.
• That would subject the data collected to local Chinese data privacy laws — one of the main concerns for Washington's China hawks.

Yes, but: China is already capable of lurking inside U.S. companies, even without a clear backdoor.

• They've hacked the Treasury Department and several major U.S. telcos just in the last year. Officials have also warned they're lurking on water systems, ports and other critical infrastructure.
• U.S. AI companies have also been on watch for employees who could leak information to Chinese companies.

What we're watching: Whether DeepSeek gets the TikTok or Huawei treatment depends on how many companies actually make the choice to ditch U.S. AI companies.

• President Trump said yesterday that DeepSeek is a "positive" and that the company's advancements are a "wake-up call for our industries that we need to be laser focused on competing to win."

• Apple Intelligence AI is turned on by default in the iPhone's latest software update. (CNBC)
• Open AI's Sam Altman called DeepSeek's R1 "impressive," but promised better models. (Axios)

I watched four really good movies on the plane: "This is Where I Leave You," "Thelma," "Goodrich," and "Summer Camp."