The Justice Department is investigating whether Nvidia structured its Groq deal to avoid antitrust review

The US Justice Department is investigating whether Nvidia deliberately structured its December licensing agreement with AI inference-chip designer Groq to evade antitrust review. The DOJ has issued Nvidia a formal information request, creating regulatory and potential legal risk around Nvidia's AI-related dealmaking. The inquiry could delay, alter, or challenge the transaction and raises broader scrutiny risks for Nvidia's competitive strategy in AI chips.

Analysis

The principal exposure is not a near-term earnings impairment but a regulatory discount on Nvidia's AI-adjacency strategy. If DOJ characterizes licensing-plus-hiring or asset-access arrangements as de facto acquisitions, Nvidia loses a flexible route to secure scarce inference IP without the delay and remedy risk of a formal merger process. That would raise the cost and duration of building a broader inference stack, while increasing the strategic value—and potential acquisition premiums—of independent inference and custom-silicon vendors.

Over the next 1-3 months, the key catalyst is whether the request escalates into a Hart-Scott-Rodino-related enforcement action, civil investigative demand, or broader inquiry into Nvidia's partnerships. A formal challenge would likely compress NVDA's premium multiple more than consensus estimates imply because it validates that regulators are targeting the company's ecosystem control rather than only its GPU market share. Conversely, a limited document review or no-action outcome would make the initial weakness a buyable headline event, provided hyperscaler capex and Blackwell deployment commentary remain intact.

Second-order beneficiaries are AMD and Intel, whose accelerator and inference alternatives become more strategically relevant if customers and startups view Nvidia's commercial arrangements as politically constrained. The more important structural beneficiary may be hyperscaler custom silicon—GOOGL TPU, AMZN Trainium, and MSFT's internal silicon efforts—as regulatory friction encourages customers to diversify away from Nvidia-controlled software, networking, and inference pathways. The contrarian view is that Groq's commercial scale is too small to alter Nvidia's earnings trajectory; the market should distinguish transaction-process risk from a finding that restricts CUDA bundling, exclusivity, or supply allocation, which would be materially more consequential over 6-18 months.

Key Decisions for Investors

• Maintain NVDA core exposure but hedge the next 60-90 days with a modest NVDA put spread financed by selling a lower-strike put; use a 10-15% downside band to protect against a formal DOJ escalation while avoiding excessive premium spend. Remove the hedge if the inquiry remains procedural and management reaffirms accelerator supply, gross-margin, and Blackwell ramp expectations.
• Initiate a 1-3 month relative-value basket: long AMD and/or INTC versus short NVDA in equal beta-weighted notional, sized small. The thesis is regulatory diversification demand rather than an immediate share shift; exit if DOJ does not escalate or if AMD/Intel fail to show incremental inference design wins or improved accelerator supply commentary.
• Watch GOOGL and AMZN for evidence that internal accelerator capacity is being pulled forward. Do not buy solely on this inquiry; upgrade to a long only if upcoming cloud disclosures show AI workload migration, improving cloud margins, or explicit third-party availability expansion for TPU/Trainium.
• Avoid treating this as a standalone short-NVDA catalyst. The thesis is falsified by a no-action resolution, unchanged partner-access economics, and continued hyperscaler AI capex growth; under that outcome, regulatory multiple compression should reverse faster than any fundamental impact emerges.