Nvidia disclosed for the first time that it beneficially owns roughly 9.3% of AI cloud provider Nebius Group, in a Schedule 13G filed with the U.S. Securities and Exchange Commission on Monday, July 20, 2026. Nebius shares closed up 2.76% on Nasdaq and rose as much as 3.9% in after-hours trading once the filing became public.
The disclosure landed three days after Nebius announced a $775 million senior secured debt facility backed by GPU infrastructure it already operates — the first financing of its kind from the company, and a template it says it intends to repeat.
Key Highlights
- Nvidia beneficially owns approximately 22.25 million Nebius Class A ordinary shares, or about 9.3%.
- Only 1.19 million of those shares are held directly; roughly 21.07 million sit in pre-funded warrants acquired in a $2 billion investment in March 2026.
- The warrants cannot be exercised until September 11, 2026, with comparable restrictions on selling the underlying shares.
- Separately, Nebius closed a $775 million secured facility on July 17, priced at SOFR plus 2.50% and maturing October 31, 2030.
- Nebius says the same structure can be applied to more than $40 billion of contracted revenue from investment-grade customers including Microsoft and Meta.
Details
The 13G is a disclosure, not a transaction. Nvidia's position dates back to a $2 billion strategic investment made on March 11, 2026, largely structured as pre-funded warrants rather than common stock. Monday's filing simply formalised the resulting ownership percentage — which is why the market reaction was measured rather than dramatic.
That formalisation still matters. A Schedule 13G puts Nvidia on the record as Nebius's largest outside institutional holder, a fact institutional analysts can now cite directly in research coverage rather than infer from a press release.
The financing announced on July 17 is arguably the more structurally novel development. Nebius secured $775 million against deployed GPU infrastructure plus contracted cash flows from a single investment-grade customer. According to the company, the facility together with those contracted cash flows covers more than 100% of the capital expenditure required to deploy the underlying hardware.
The syndicate was broad. MUFG acted as structuring agent, sole bookrunner and underwriter, joined as mandated lead arrangers by ABN AMRO, Bank of America, Deutsche Bank and HSBC. Citi, Crédit Agricole CIB, ING and Morgan Stanley served as senior lead arrangers, with Goldman Sachs participating. Nebius described the transaction as significantly oversubscribed.
"This financing is an important step in that strategy, and reinforces our confidence that our disciplined, diversified approach will enable us to build a sustainable AI cloud business with strong and durable margins," said Ophir Nave, Nebius Chief Operating Officer, in the company's announcement.
Impact
Taken together, the two events describe how the so-called neocloud sector — AI-specialised infrastructure providers competing with the hyperscalers — is now being funded.
Historically, neoclouds have grown by issuing equity or raising unsecured debt to buy GPUs. Nebius's facility inverts that: the GPUs it already runs, and the contracts they serve, become the collateral. That converts an operating asset into comparatively cheap growth capital without further diluting shareholders, months after a $4.3 billion convertible raise in March 2026.
If the structure proves repeatable at the scale Nebius projects, financing terms — not chip allocation alone — start to determine which providers can build fastest. Rivals including CoreWeave, which has leaned heavily on debt raises to fund GPU purchases, are the obvious candidates to copy it.
There is a second-order implication for buyers of AI compute. When a provider pledges contracted cash flows as collateral, an enterprise signing a multi-year AI cloud agreement is indirectly supporting someone else's credit structure. Payment reliability on those contracts becomes systemically relevant in a way it was not before.
Background
Nvidia's stake in Nebius is one strand of a broader pattern: the chipmaker has taken positions across the AI infrastructure stack rather than limiting itself to selling processors. Backing the cloud providers that deploy its hardware helps ensure demand, capacity and ecosystem alignment further downstream.
Nebius, for its part, operates an asset-light model relative to peers, with outside partners financing and owning parts of the physical infrastructure while Nebius supplies the architecture and runs the platform. The company reports a backlog in the region of $46 billion, anchored by partnerships with Nvidia, Microsoft and Meta.
What's Next
The near-term marker is September 11, 2026, when Nvidia's warrant exercise restrictions lift. Beyond that, the question is whether Nebius can execute additional asset-level facilities at comparable terms against its stated pipeline of more than $40 billion in contracted revenue — and whether other neoclouds follow it into GPU-backed debt markets.
Source: CNBC