Antitrust and Securities Risks in Nvidia's Nebius Investment

The legal issue in Nvidia’s Nebius investment starts in two filings that should be read together, not in the market’s enthusiasm for AI infrastructure. Nebius disclosed in its FY2024 Form 20-F that one supplier accounted for 73% of its total capital expenditures and warned that this concentration created exposure to supply disruption and pricing dependency.[1] Nvidia then filed a Schedule 13G on July 20, 2026 reporting 9.3% beneficial ownership of Nebius Class A shares, consisting of 1,190,476 shares held and 21,065,936 shares underlying a pre-funded warrant, with a six-month restriction on exercise or sale until September 11, 2026, and sole voting and dispositive power reported.[2]

That pairing is the reason analysis of Nvidia’s Nebius investment cannot stop at “strategic investment.” Nvidia is not just a capital provider. On the public record available today, it is also the supplier whose chips appear to drive the largest disclosed capital-expenditure dependency at Nebius. The legal question is not whether that structure is automatically unlawful. It is whether the same company’s position as supplier, equity holder, and possible allocator of scarce AI-compute inputs gives regulators and investors enough concrete facts to ask harder questions.

The loop is visible before anyone argues about intent

The basic loop is easy to overstate and just as easy to understate. Nvidia invests in Nebius. Nebius needs Nvidia GPUs to build AI-cloud capacity. Nvidia’s equity upside may improve if Nebius expands successfully. Nebius’s expansion, in turn, can require more Nvidia supply. That is the circularity. It does not prove an antitrust violation by itself, and it does not prove that Nebius bought GPUs on anything other than commercial terms. But it does put supply economics and equity economics into the same frame.

Reuters reported Nvidia’s $2 billion investment in Nebius on March 11, 2026.[3] Four months later, the 13G made the ownership mechanics more precise: a disclosed 9.3% beneficial ownership position, a pre-funded warrant, a six-month restriction, and sole voting and dispositive power.[2] Those are not atmospheric signals of support. They are the kind of details that matter when a disclosure lawyer, regulator, or audit committee asks who has economic exposure, who can vote, and when the position can be exercised or sold.

Nebius’s own supplier concentration disclosure does equal work here. A 73% capital-expenditure concentration in one supplier is not merely a procurement footnote when the same ecosystem participant later appears as a material equity holder. It tells investors that the company’s growth plan depends heavily on continued access to a supplier’s products, pricing, delivery schedule, and commercial willingness.[1]

Why antitrust law sees more than one possible problem

The antitrust analysis should not be compressed into a single label. “Circular financing” describes a structure; it is not a cause of action. The more useful exercise is to separate the Sherman Act questions from the Clayton Act questions, because they look at different conduct and different competitive harms.

Sherman Act Section 1: tying, exclusivity, and preference

Under Section 1 of the Sherman Act, the relevant risk would be concerted conduct that restrains trade. In this setting, the obvious theories are tying, exclusive dealing, or preferential access arrangements. The question would not be whether Nvidia invested in Nebius. It would be whether capital, supply, pricing, customer priority, or capacity allocation were conditioned in a way that foreclosed rivals or distorted customer choice.

A tying theory would require more than a supplier investing in a customer. Regulators would look for evidence that access to a desired product, financing term, or capacity arrangement was conditioned on taking another product or accepting a related commercial obligation. In the AI infrastructure context, that might mean asking whether access to scarce GPUs was linked to cloud-capacity commitments, preferred purchasing behavior, or other terms that made rival suppliers or rival cloud providers less viable. The public Nebius materials cited here do not establish that kind of condition. They do, however, identify the commercial setting in which such a question would be asked.

Exclusive-dealing analysis would ask a slightly different question: whether Nebius, because of supply dependency and Nvidia’s investment position, had practical or contractual incentives to concentrate purchases with Nvidia in a way that substantially foreclosed competing accelerator suppliers. Nebius’s 20-F supplier concentration figure is important because it shows that the dependency was already economically significant in 2024.[1] It does not show exclusivity. It does show why a regulator would care about the difference between a procurement preference, a supply bottleneck, and a contractual restraint.

Preference is the harder and more modern version of the problem. If a dominant supplier also holds equity in a customer, the enforcement question may shift from formal exclusivity to allocation: who gets chips first, who receives better terms, whose buildout is enabled, and whose is delayed. Reuters reported that the DOJ had been investigating Nvidia since at least September 2025, including subpoenas concerning alleged tying, exclusive dealing, and customer-prioritization practices.[5] The American Action Forum also discussed antitrust scrutiny of Nvidia in the context of AI-chip market power and related distribution concerns.[6] Public reporting does not reveal the full scope or current status of the investigation as of July 22, 2026, so the investigation should be treated as context, not proof.

Clayton Act Section 7: the equity stake is its own issue

Section 7 of the Clayton Act looks at acquisitions whose effect may be substantially to lessen competition or tend to create a monopoly. That inquiry can reach minority stakes. The practical question is whether Nvidia’s 9.3% beneficial ownership in a GPU-dependent AI-cloud company could soften competition, influence purchasing, give Nvidia competitively sensitive visibility, or entrench Nvidia’s supply position.[2]

A minority position is not automatically competitively dangerous. The analysis would turn on rights, incentives, information flows, market structure, and the degree of dependence. The 13G reports sole voting and dispositive power, but the public materials summarized here do not establish board rights, veto rights, information rights, or contractual commitments that would make the Clayton Act case straightforward.[2] That absence matters. Enforcement risk is not the same thing as liability.

Still, the equity stake cannot be separated from the supply market. If Nvidia were only a passive investor in an unrelated software company, the Clayton Act concern would look different. Here, the company taking the equity position is also the supplier associated with Nebius’s largest disclosed capital-expenditure concentration.[1][2] The economic incentives are therefore not incidental to the transaction. They are part of the competitive fact pattern.

The securities question is narrower, but not secondary

The securities-law issue is not whether Nvidia and Nebius chose an inefficient capital structure. It is whether investors can see enough of the related-party and supply-dependency picture to understand the risk they are underwriting. That includes the magnitude of supplier concentration, the identity and importance of the supplier if ascertainable from the filings and surrounding record, the investment mechanics, and any material commercial arrangements that connect the two.

Nebius did disclose a significant supplier concentration in its Form 20-F and warned that dependency could affect supply and pricing.[1] Nvidia separately disclosed its beneficial ownership in the July 2026 Schedule 13G.[2] Those disclosures are important. The open question is whether they are sufficiently connected for investors to understand the full commercial relationship: Nvidia as supplier, Nvidia as investor, Nebius as purchaser, and Nebius as a company whose growth depends on securing advanced GPU capacity.

That question should be framed carefully. A disclosure gap is not established merely because two facts appear in separate filings. Public-company disclosure is often distributed across annual reports, ownership filings, exhibits, risk factors, and transaction announcements. The risk is that the related-party character of the arrangement can become functionally harder to evaluate if the supply dependency and the equity economics are not presented in a way that investors can read together.

For investors, the materiality question is concrete. If Nebius’s capital buildout depends heavily on Nvidia GPUs, and Nvidia holds a meaningful economic position in Nebius, then supply interruptions, pricing changes, warrant exercise limits, voting power, and any preferential or restrictive commercial terms all become part of the same risk map. The law does not require every commercial dependency to be turned into a legal conclusion. It does require material risks to be described without making investors assemble the central structure from fragments.

CoreWeave shows the pattern more plainly

CoreWeave is the closest parallel because it makes the loop easier to see. I/O Fund’s June 2026 analysis described Nvidia’s $2 billion investment in CoreWeave and a $6.3 billion backstop for unsold GPU capacity, creating a structure in which Nvidia invests, CoreWeave buys Nvidia GPUs, and Nvidia supports unsold capacity.[4] The Nebius transaction, as presented in the cited materials, does not include the same explicit $6.3 billion backstop. That distinction matters.

The comparison is useful precisely because it prevents overclaiming. CoreWeave appears to include a more explicit capacity backstop structure, while Nebius presents a supplier-equity loop anchored by capital expenditure concentration and Nvidia’s beneficial ownership.[1][2][4] Both structures can attract antitrust attention, but they do not create identical legal records.

That is also why vendor-financing analogies only go so far. Vendor financing is not new, and it is not inherently suspect. Suppliers often help customers buy expensive inputs. The difference in AI infrastructure is scarcity and market position. When the supplier is associated with the bottleneck input and also takes equity exposure in the customer buying that input, standard financing logic begins to overlap with foreclosure, preference, and disclosure concerns.

The DOJ backdrop increases the cost of ambiguity

The DOJ investigation matters because it tells companies and investors which facts regulators may already be testing: bundling, tying, exclusive dealing, and customer prioritization.[5][6] It does not mean the Nebius investment is unlawful. It does mean that transaction documents, supply agreements, allocation policies, side letters, board materials, and investor disclosures would likely be read against an enforcement environment already focused on Nvidia’s role in AI infrastructure.

The more disciplined antitrust question is not “Is Nvidia too big?” It is whether Nvidia used its position in scarce GPU supply to shape downstream AI-cloud competition in ways that agreements, investments, or preferential practices made durable. The Nebius filings do not answer that question. They supply the factual architecture that makes the question serious.

Some critics have compared AI circular deals to dot-com-era round-tripping, while others characterize the structures as vendor financing rather than circular financing.[7] The analogy is useful only up to a point. The legal work is not done by naming the loop. It is done by identifying the cash flows, purchase obligations, capacity commitments, voting rights, supply constraints, and investor disclosures that make the loop consequential.

What would matter next

If regulators or private plaintiffs examined the Nebius structure, the most important evidence would likely sit in the transaction exhibits and commercial agreements rather than in public commentary. The public record identifies supplier concentration and ownership mechanics. It does not disclose the full set of supply commitments, pricing terms, allocation rules, information rights, or operational understandings that would determine whether the structure crosses from exposure into liability.

• Whether GPU access, pricing, or delivery priority was conditioned on equity, purchase commitments, or other commercial obligations
• Whether Nebius had practical or contractual limits on using rival accelerators or alternative suppliers
• Whether Nvidia received information rights, governance rights, or visibility into Nebius plans beyond ordinary investor protections
• Whether capacity allocation favored Nvidia-backed customers over similarly situated rivals
• Whether Nebius’s disclosures connect supply dependency and Nvidia’s ownership position clearly enough for investors to evaluate material risk

Those are not accusations. They are the document requests that naturally follow from the public facts. The 73% supplier concentration explains why supply terms matter.[1] The 9.3% beneficial ownership position explains why investment terms matter.[2] The DOJ investigation explains why allocation and bundling issues cannot be dismissed as academic.[5][6]

A bounded legal-market judgment

Nvidia’s Nebius investment is not a clean morality play about Big Tech, and it is not a self-proving antitrust violation. It is a foundational example of AI infrastructure circular financing risk because the same public record places Nvidia in overlapping roles: supplier linked to Nebius’s largest disclosed capital-expenditure concentration, investor with a reported 9.3% beneficial ownership position, and market participant operating under reported DOJ antitrust scrutiny.[1][2][5]

That combination creates antitrust exposure under Sherman Act theories of tying, exclusivity, and preference, and under Clayton Act scrutiny of minority equity acquisitions that may entrench supply power. It also creates a securities-law disclosure question about whether investors can understand the related-party and supplier-dependency picture without reconstructing it from separate filings. The current public record supports those risks as risks, not outcomes.

References

1. Nebius Form 20-F for FY2024, SEC EDGAR, April 2025, link
2. Schedule 13G filing, StockTitan/EDGAR, July 20, 2026, link
3. Nvidia to invest $2 billion in AI cloud firm Nebius, Reuters, March 11, 2026, link
4. Nvidia, CoreWeave, Nebius: Circular Financing in the GPU Boom, I/O Fund, June 2026, link
5. DOJ antitrust investigation into Nvidia, Reuters, September 23, 2025, link
6. Antitrust scrutiny of Nvidia, American Action Forum, October 2024, link
7. Should we worry about AI’s circular deals?, Noahpinion, 2026, link