# What Exactly Did Nvidia and Groq Agree To?

According to Bloomberg Technology, Nvidia Corp. entered a $20 billion licensing agreement with AI chip startup Groq Inc. — a structure that transfers intellectual property rights and, reportedly, key personnel, without a formal acquisition. Bloomberg reported on September 10, 2026, that the Justice Department is investigating whether that structure was chosen deliberately to avoid antitrust review under the Hart-Scott-Rodino Act. The distinction matters enormously. An outright acquisition of Groq at $20 billion would almost certainly have triggered an HSR filing, a lengthy review, and potentially a Second Request from the DOJ or FTC. A licensing agreement — even one worth $20 billion — can fall outside HSR thresholds if it does not constitute a transfer of voting securities or substantially all assets of a operating business. Nvidia, the dominant supplier of AI accelerators, has every incentive to structure the deal this way. Groq makes LPU (Language Processing Unit) inference chips that compete directly with Nvidia's GPUs in low-latency AI inference. Absorbing Groq's IP and talent neutralizes a credible inference competitor without the regulatory friction of a merger.

# Why Is the DOJ Focused on Deal Structure Rather Than Deal Size?

According to Bloomberg, the DOJ's concern is not the $20 billion price tag — it is the deliberate choice of a licensing vehicle over an acquisition. That is a structural enforcement question, and it is the more dangerous one for Nvidia. Antitrust law has long recognized that substance can override form. If a licensing agreement effectively transfers control of a competitor's core technology and team, regulators can argue it is a de facto merger subject to review. The DOJ's investigation suggests it is testing exactly that theory. If the agency concludes Nvidia chose licensing specifically to evade HSR, the precedent would be far broader than one deal: it would cast doubt on every "acqui-hire," IP license, and strategic partnership that AI incumbents have used to consolidate capability without formal merger filings.

# Who Actually Competes With Nvidia in AI Inference?

Groq is one of a small number of credible alternatives to Nvidia in AI inference. Its LPU architecture targets deterministic, low-latency inference workloads — a segment where Nvidia's general-purpose GPUs are less efficient. Other competitors include AMD's Instinct line, Google's TPUs (internal only), and a handful of startups. If Nvidia absorbs Groq's IP and key engineers through a licensing deal, the competitive landscape narrows materially. That is the economic harm the DOJ would need to establish. The table below compares the deal structures and their regulatory exposure.

Deal Structure	What Transfers	HSR Filing Required?	Antitrust Exposure	Example
Full acquisition	Voting securities, control	Yes, above threshold	High — full merger review	Microsoft/Activision
Asset purchase	Substantially all assets	Often yes	Moderate to high	Various
Licensing agreement	IP rights, possibly personnel	Usually no	Now under DOJ scrutiny	Nvidia/Groq
Acqui-hire	Talent, small IP	Usually no	Low, but rising	Common in AI
Verdict	Licensing is the most regulatorily exposed structure if the DOJ establishes that economic substance, not legal form, controls.

# What Does the DOJ Need to Prove — and How Hard Is That?

The DOJ would need to show that the Nvidia-Groq licensing agreement functioned as an acquisition in substance. Evidence could include exclusive licensing terms, transfer of key Groq engineers to Nvidia, non-compete provisions, or a price that reflects the entire enterprise value of Groq rather than the value of specific IP. Bloomberg's reporting does not detail the specific terms the DOJ is examining, but the $20 billion figure is striking: it is large enough to suggest Nvidia paid for the whole company's worth, not a narrow license. That is the evidentiary hook. The harder question is remedy. If the DOJ finds the structure evasive, it can seek to unwind the deal, require HSR filing retroactively, or impose conduct remedies. Unwinding a completed licensing deal is legally difficult but not impossible — and the threat alone could chill similar transactions across the AI sector.

# What Happens Next for Nvidia, Groq, and the AI Chip Market?

According to Bloomberg, the investigation is ongoing and no charges have been filed. Nvidia has not publicly commented on the probe in the source material. Groq's position is more precarious: if the deal is unwound or restructured, the startup's independence is compromised either way. For the broader market, the signal is clear: the DOJ is willing to look through deal structures to assess competitive effects. That is a shift from the traditional HSR threshold analysis, and it will force AI companies to reconsider how they structure partnerships, licenses, and talent acquisitions.

# Predictions

1. By Q1 2027, the DOJ will issue civil investigative demands to Nvidia and Groq seeking internal communications about the decision to structure the deal as a licensing agreement rather than an acquisition.
2. Nvidia will restructure or voluntarily file for HSR review by mid-2027 rather than risk a precedent that treats licensing agreements as de facto mergers.
3. At least two other AI chip startups will cite the Nvidia-Groq probe in negotiations by Q3 2027, demanding acquisition structures rather than licensing deals to preserve regulatory scrutiny and higher valuations.

# Article Summary

• The DOJ probe into Nvidia's $20 billion Groq licensing deal is a structural antitrust case, not a size case — and that distinction will define AI consolidation rules.
• Groq's competitive independence is compromised regardless of the probe's outcome; its IP and talent are already entangled with Nvidia.
• Licensing agreements are now the most regulatorily exposed deal structure in AI, reversing the conventional wisdom that smaller, non-merger deals face less scrutiny.
• The DOJ's theory — that economic substance overrides legal form — could apply to every acqui-hire and IP license in the AI sector.
• Nvidia's best legal strategy is likely voluntary restructuring, not litigation, because a loss would set a precedent that constrains its future dealmaking.