(Bloomberg) — The U.S. Justice Department has sent subpoenas to Nvidia Corp. and other companies as it seeks evidence the chipmaker violated antitrust laws, an escalation of an investigation into the largest supplier of AI processors.
Most read from Bloomberg
The DOJ, which previously sent questionnaires to companies, is now sending legally binding requests requiring recipients to provide information, people familiar with the investigation said, bringing the government a step closer to filing a formal complaint.
Competition authorities are concerned that Nvidia is making it harder to switch suppliers and penalizing buyers who don’t exclusively use the company’s artificial intelligence chips, according to the sources, who spoke on condition of anonymity because the discussions are confidential.
Nvidia shares, which suffered a record-breaking defeat on Tuesday, fell further in late trading after Bloomberg reported on the subpoenas. Still, the stock has more than doubled this year, fueled by explosive revenue growth at the Santa Clara, California-based chipmaker.
As part of the probe, which Bloomberg reported earlier in June, investigators have reached out to other technology companies to gather information. The DOJ’s San Francisco office is leading the investigation, the sources said. A DOJ representative declined to comment.
In response to questions about the research, Nvidia said its market dominance stems from the quality of its products, which deliver faster performance.
“Nvidia wins on merit, as demonstrated by our benchmark results and the value to customers, who can choose the solution that best suits their needs,” the company said in an emailed statement.
Nvidia has faced regulatory scrutiny since it became the world’s most valuable chipmaker and a major beneficiary of the AI spending boom. Sales have more than doubled every quarter, outpacing former chip leaders such as Intel Corp.
In the Justice Department investigation, regulators are examining Nvidia’s acquisition of RunAI, a deal announced in April. The company makes software for managing AI computing, and there are concerns that the partnership will make it harder for customers to switch from Nvidia chips. The sources said regulators are also questioning whether Nvidia is giving preferential shipping and pricing to customers who exclusively use its technology or buy its complete systems.
Founded in 1993, Nvidia made its name selling graphics cards to computer gamers. But its approach to chipmaking ultimately proved useful for building AI models, a process that involves bombarding the software with data. The company has also quickly expanded its offerings to include a suite of software, servers, networking and services — all aimed at accelerating the adoption of AI, Nvidia argues.
Nvidia CEO Jensen Huang said he is prioritizing customers who can use his products in ready-to-use data centers as soon as he ships them. The policy is intended to prevent hoarding and accelerate the broader adoption of AI.
The success of its products — coupled with the struggles of rivals to supply alternative chips — has made Nvidia a crucial part of the supply chain for some of the world’s largest companies. Microsoft Corp. and Meta Platforms Inc., for example, spend more than 40 percent of their hardware budgets on the chipmaker’s equipment. During the height of the shortage of Nvidia’s H100 accelerator, individual components were selling for as much as $90,000 apiece.
Analysts predict that Nvidia will generate $120.8 billion in revenue in calendar year 2024, up from $16 billion in 2020, with the bulk of that money coming from its data center unit. In fact, Nvidia is expected to make more profit this year than the total revenue of its nearest rival, Advanced Micro Devices Inc.
There are also broader regulatory questions about Nvidia’s practices. Access to AI capabilities has become a major focus for governments around the world, with the technology increasingly important to economic strength and national security.
(Updates with Nvidia response from the sixth paragraph.)
Most read from Bloomberg Businessweek
©2024 Bloomberg LP
