The U.S. Justice Department is investigating Nvidia’s $17 billion licensing arrangement with AI chip startup Groq, raising a larger question about how regulators should treat deals that transfer technology and talent without formally buying the company.
According to Reuters, citing a New York Times report, the department is examining whether Nvidia structured the transaction in a way that avoided the antitrust scrutiny normally triggered by an acquisition.
Nvidia announced the arrangement in December as a $17 billion non-exclusive license to Groq’s chip technology. The company also hired several Groq executives, including founder Jonathan Ross.
A deal can look different on paper than in practice
The structure reflects a broader pattern in artificial intelligence: major technology companies are increasingly securing strategic technology, researchers and leadership through licensing and hiring agreements that stop short of buying an entire startup.
That can offer speed and flexibility. It can also complicate the regulatory framework, which was built around clearer categories such as acquisitions, mergers and minority investments.
The DOJ reportedly opened its investigation shortly after the Groq deal was announced and sent Nvidia a formal request for information. Nvidia has defended the arrangement, saying it reflects a system that rewards innovation and benefits consumers. Groq and the Justice Department had not commented publicly to Reuters.
Why the Groq structure matters
Groq has built specialized chips designed for AI inference—the stage where trained models generate answers in real time. That market is becoming strategically important as AI use shifts from model training toward billions of daily interactions involving chatbots, coding systems, voice agents and enterprise software.
Nvidia already dominates the broader AI accelerator market. Access to Groq’s technology and senior team therefore raises an obvious competition question: when a dominant company gains the practical benefits associated with acquiring a startup without technically completing an acquisition, should regulators evaluate the arrangement based on its legal structure or its economic effect?
The answer could shape future AI transactions well beyond Nvidia. The frontier AI market is moving quickly enough that companies have strong incentives to invent new deal structures faster than regulators can classify them.
If the DOJ concludes that such arrangements require greater review, licensing-plus-hiring deals could become the next major front in technology antitrust.