Nvidia attracts antitrust attention as enforcers signal early interest in AI

The Justice Department’s antitrust division has already contacted Nvidia, which has an estimated 80% share of the AI ​​chip market, to ask questions about the terms of the agreements and partnerships, according to people familiar with the matter.

The investigation is in its early stages, and government lawyers have not issued a subpoena to Nvidia seeking internal documents, the people said. The company said last week that it had not received a subpoena after media reports that it had. The department could issue a subpoena in the coming months if it determines a more detailed investigation is necessary, the people said.

The Federal Trade Commission, which shares antitrust powers with the Justice Department, this year launched its own investigation into the investments that Microsoft, Amazon.com and Google-owner Alphabet have made in the most compelling AI startups. The FTC review, which is ongoing, is examining whether the tech giants gained any special advantages over their competitors by investing billions of dollars in companies that make the most popular generative AI products.

The involvement of the Justice Department and the Federal Trade Commission in overseeing the early days of AI stands in contrast to the approach those agencies took toward Google, Facebook and other companies before they became the internet giants they are today.

Federal regulators have historically been timid about intervening in fast-growing technology markets with products that have appealed to consumers. Liberal and conservative critics have argued that a hands-off approach over the past two decades—across administrations—has allowed a handful of tech giants to amass too much power. The backlash has helped spur a shift toward aggressive antitrust enforcement that the Biden administration is embracing.

The Justice Department “would say, ‘We’ve watched enough of Big Tech to know we want to kick the tires,’” said Randal Picker, a law professor at the University of Chicago. “You can have market power, you can have it. The question is always, how do you use it?”

A handful of the largest tech companies are fiercely competing for market power in AI, which requires massive capital investments that smaller competitors can’t match. Antitrust enforcers say a tipping point could come when incumbents gain control over critical aspects of AI—similar to the rise of internet search and social media.

FTC Chairwoman Lina Khan worries that dominant technology companies could gain new monopolies in the emerging field of generative artificial intelligence, or systems that have human-like abilities to communicate, create multimedia, write computer code and more.

“There are now very legitimate concerns that critical inputs to this technology may already be controlled by a handful of companies,” Khan told venture capital and startup advisory firm Y Combinator last year. “We may already be seeing bottlenecks that could impede innovation and hinder competition.”

Jonathan Kanter, the head of the Justice Department’s antitrust division, warned in May that the competitive dynamics in AI are similar to those in other technology industries, where a single company dominates. “AI relies on vast amounts of data and computing power, which can give dominant firms significant advantages,” Kanter said in a speech at Stanford University. “Powerful network effects can enable dominant firms to control these new markets.”

Nvidia’s chips are essential for training AI models developed by OpenAI and others. The department’s investigation comes as the company faces growing antitrust scrutiny around the world, from the European Union and the U.K. to China and South Korea. Its French offices were raided by antitrust regulators last year. Nvidia said in an August securities filing that it had received requests for information from various enforcement agencies about its sales, how it allocates limited supplies of chips and partnerships with AI companies.

Nvidia’s high market share isn’t inherently illegitimate. The company’s growth exploded after customers discovered that its specialized chips, once used primarily for graphics in video games and then for computers that handled cryptocurrency transactions, were particularly well-suited to the computational demands of AI.

To bring antitrust claims, the government would need to find instances where Nvidia took actions (such as restrictive contract terms) that harmed competition and artificially maintained or increased its market power.

Some of Nvidia’s rivals complain that it uses its massive share of the AI ​​computing market to its advantage.

Rodrigo Liang, CEO of AI chip startup SambaNova Systems, said competition is growing and there is a strong demand for alternatives to Nvidia. But he added that some customers are concerned that Nvidia could stop selling its AI chips, known as GPUs, to customers who are exploring alternatives like his.

“It’s more of a commercial question of if the world has a supply constraint on GPUs, do I want to take the risk of not having supply?” he said.

An Nvidia spokeswoman said the company does not require exclusivity and supports all its customers and partners.

“Our customers know they have the freedom to choose any combination of suppliers they feel are best suited for them, and we will work with them no matter what decision they make,” she said.

Because Nvidia’s chips are so heavily used in AI development, decisions about who gets them can affect the success of companies. Nvidia’s most advanced AI chips have long been limited in supply, with customers sometimes waiting months for orders to be filled. The company said demand will outpace supply for the next-generation chips, codenamed Blackwell, which are set to begin shipping in the coming months.

Nvidia said in a statement that it owes its market position to the high computing performance of its AI products and added that it is happy to answer questions from the government on how it conducts its business.

“Nvidia wins on the merits as reflected in our benchmark results and customer value, and customers can choose the solution that is best for them,” the company said.

In addition to investigating companies’ business practices, antitrust enforcers can challenge mergers that remove too much competition from a market. The government has been slow for years to challenge the deals that created today’s tech giants. The FTC, for example, allowed Facebook to acquire Instagram in 2012.

The Justice Department is investigating Nvidia’s deal in April to buy Run:ai, a startup that offers software that organizes and streamlines the use of clusters of AI chips, making them more efficient.

The department began investigating Nvidia’s Run:ai deal after reaching a power-sharing agreement with the FTC on how to divide oversight of AI companies, according to people familiar with the matter. Politico previously reported on the investigation into the Run:ai deal.

Agency efforts to investigate AI companies can last months or years. Merger investigations can last up to a year, and investigations into broader company conduct often take longer.

Write to Dave Michaels at [email protected] and Asa Fitch at [email protected]