The Department of Justice has opened an antitrust investigation into Nvidia's approximately twenty billion dollar licensing agreement with AI chip startup Groq, examining whether the deal was deliberately structured to avoid the kind of merger review that a traditional acquisition would trigger. The probe, first reported by The New York Times on September 9 and subsequently confirmed by Bloomberg, marks the most aggressive regulatory scrutiny yet of the "structured non-acquisition" loophole that has enabled AI industry consolidation without government oversight.

The License-Plus-Acqui-Hire Structure

Nvidia's arrangement with Groq, announced in December and valued at roughly twenty billion dollars according to published reports, involves licensing the startup's Language Processing Unit technology while hiring key personnel—a structure that lets Nvidia effectively gain control of Groq's innovation and talent without formally acquiring the company. By avoiding a traditional acquisition, the deal sidestepped Hart-Scott-Rodino premerger review, the federal process that scrutinizes large deals for anticompetitive effects before they close.

Jonathan Ross, founder of Groq and creator of the company's inference-optimized chip architecture, built the startup to challenge Nvidia's dominance in AI compute. The LPU represented one of the few credible alternatives to Nvidia's GPUs for running large language models in production. Now, with Nvidia holding the license and employing the key engineers, that competitive threat has effectively been neutralized—without regulators ever getting a chance to weigh in.

Bloomberg: Civil Investigative Demand Issued

Bloomberg subsequently reported that the DOJ has sent a civil investigative demand—a formal request for documents and information—to Nvidia as part of the investigation. While Nvidia, Groq, and the DOJ have made no public statements, the CID indicates the government is moving beyond preliminary inquiries into a more structured phase of fact-finding.

It's important to emphasize what this is and what it is not: this is an investigation, not a charge. No lawsuit has been filed, no court has ruled, and neither Nvidia nor Groq has been accused of breaking the law. The central theory under examination—that the structure was deliberately chosen to avoid HSR review—is exactly that: a theory being tested, not a finding. The deal may well survive full scrutiny.

The Broader Pattern: Structured Non-Acquisitions Across AI

The Nvidia-Groq probe is part of a broader pattern of regulatory focus on license-and-hire arrangements that have become the preferred M&A structure in artificial intelligence. Microsoft's deal with Inflection AI, Amazon's arrangement with Adept, Google's licensing agreement with Character.AI, and Meta's structure with Scale have all followed similar blueprints: billions of dollars changing hands, key founders and engineers moving to the acquirer, and the target company left as an empty shell—all without triggering formal antitrust review.

These deals share a common logic. By framing the transaction as a licensing agreement plus employee hiring rather than an acquisition, companies can argue they're paying for intellectual property and talent, not buying market share or eliminating competition. But regulators are now questioning whether that distinction holds when the economic substance of the deal looks identical to an acquisition.

Nvidia's Dominance Under the Microscope

The timing of the investigation is significant. Nvidia controls an estimated 80% to 90% of the market for AI training chips, a position that has made it one of the world's most valuable companies. Groq represented one of the few viable alternatives for inference workloads—the computationally intensive task of running AI models in production. By licensing Groq's technology and absorbing its team, Nvidia has consolidated its position at precisely the moment when regulators are most concerned about concentration in AI infrastructure.

The DOJ's willingness to probe this particular deal sends a clear signal: the government is no longer willing to accept creative structuring as a substitute for competitive markets. If the investigation leads to enforcement action—a lawsuit, a consent decree, or formal challenges to similar future deals—it could force companies across the AI stack to rethink how they pursue consolidation.

What Happens Next

Investigations of this kind typically take months, sometimes years. The DOJ will review internal documents, interview company executives and industry participants, and assess whether the deal's structure violated antitrust law. If prosecutors conclude the arrangement was designed to evade merger review, they could seek to unwind the deal, impose conditions on Nvidia's use of Groq's technology, or pursue civil penalties.

More broadly, the Nvidia-Groq probe will test the limits of structured non-acquisitions as an M&A strategy. If the DOJ succeeds in establishing that license-and-hire arrangements can be functionally equivalent to acquisitions for antitrust purposes, it would close a loophole that has powered billions of dollars in AI consolidation—and force the industry to operate under the same merger review standards that apply to every other sector of the economy.

For now, Nvidia, Groq, and the Department of Justice remain silent. But the question the government is asking—whether creative deal structures can be used to sidestep antitrust review—will have far-reaching implications for how AI companies grow, consolidate, and compete in the years ahead.