Nebius Secures $775M GPU-Backed Loan to Fund AI Cloud Expansion

Nebius (NASDAQ: NBIS) Group has secured approximately $775 million through its first senior secured debt facility, using deployed graphics-processing infrastructure and contracted customer payments as collateral as the AI cloud company looks to recycle capital into further expansion.

The Amsterdam-based company entered into the facility on July 10 through wholly owned subsidiaries in the US and Finland, according to a regulatory filing. Nebius announced the transaction on Friday, saying the proceeds would support the global buildout of its full-stack AI cloud platform.

The loan matures on Oct. 31, 2030, and carries a floating interest rate equal to one-month Term SOFR plus 2.5 percentage points, or about 6.83% effectively at the moment. Nebius said the facility, together with cash flows from the underlying customer contract, covers more than 100% of the capital expenditures required to deploy the GPU infrastructure backing the loan.

That structure allows Nebius to borrow against equipment that has already been installed and a customer contract that has moved into the servicing phase. Rather than using the proceeds to fund the same deployment, Nebius can redirect the capital into additional capacity for AI-native companies and enterprise customers using its cloud platform.

Nebius did not identify the investment-grade customer whose contract supports the facility. The company separately said it recently delivered the latest scheduled capacity tranche to Microsoft and remained on track to complete the remaining deployments under that agreement.

Nebius’s Microsoft contract, announced in September 2025, calls for the company to provide dedicated GPU capacity from its Vineland, New Jersey, data center over five years. Regulatory disclosures have valued the commitment at as much as $17.4 billion through 2031. Nebius said when announcing the contract that it expected to finance part of the associated capital expenditure using customer cash flows and debt secured by the contract and related infrastructure.

Asset-level debt limits parent-company exposure

The borrowers under the new facility are Nebius Compute II LLC in Delaware and Nebius Compute II Oy in Finland.

Substantially all of the borrowers’ assets, as well as the shares in the borrowing entities held indirectly by Nebius, are pledged as collateral. The loan agreement also requires the borrowers to maintain a debt-service coverage ratio of at least 1.15 times and comply with a minimum-liquidity requirement.

Nebius Group itself has not provided a full corporate guarantee of the debt. Its guarantees are generally limited to specified “bad acts” by the borrowers and the performance of certain management and data center colocation obligations by other Nebius subsidiaries.

That ring-fenced structure means lenders primarily rely on the pledged GPUs, the customer contract and cash flows generated by the financed deployment, rather than the broader balance sheet of the parent company.

The borrowers may voluntarily repay the facility with 10 business days’ notice without a premium or penalty, apart from customary breakage costs.

The transaction was led by MUFG as structuring agent, sole bookrunner and underwriter. ABN AMRO, Bank of America, Deutsche Bank and HSBC joined MUFG as mandated lead arrangers. Citi, Crédit Agricole CIB, ING and Morgan Stanley served as senior lead arrangers, while Goldman Sachs also participated. Nebius said the facility was significantly oversubscribed.

A new layer in Nebius’s financing strategy

The facility adds asset-backed lending to a capital structure that has so far relied heavily on equity and convertible debt.

Nebius raised $4.34 billion in March through convertible senior notes due in 2031 and 2033, with interest rates of 1.25% and 2.625%, respectively. Those notes are unsecured and can ultimately be converted into Nebius shares under certain conditions. The proceeds were earmarked for data center construction, GPU purchases and expansion of the company’s AI cloud platform.

The new facility is smaller and carries a floating interest rate, but it does not create the potential shareholder dilution associated with convertible securities. Its pricing also reflects the credit quality of the customer supporting the deployment rather than relying solely on Nebius’s corporate credit.

Nebius said the arrangement could serve as a repeatable financing framework for other long-term customer deployments. The company has more than $40 billion of additional contracted revenue involving investment-grade customers including Microsoft and Meta, providing a potential pool of contracts against which similar facilities could be raised.

In March, Nebius announced a separate five-year agreement with Meta worth as much as $27 billion. The contract includes $12 billion of dedicated capacity scheduled to begin delivery in early 2027 and a commitment from Meta to purchase as much as $15 billion of capacity left available across certain future Nebius clusters.

The ability to finance individual deployments against contracted revenue is becoming increasingly important as AI infrastructure providers race to install expensive GPUs and secure power and data center capacity before customers begin making regular payments.

Nebius spent about $2.47 billion on property, equipment and intangible assets during the first quarter, more than four times the amount a year earlier. Quarterly revenue rose to $399 million from $50.9 million, while adjusted EBITDA reached $129.5 million compared with a loss of $53.7 million in the prior-year period.

The company is simultaneously developing owned data centers, leasing colocation capacity and introducing partnerships in which third-party infrastructure owners provide buildings and power while Nebius supplies its software, systems architecture and customer relationships.

Chief Operating Officer Ophir Nave said the financing supported that diversified approach, spanning owned facilities and more asset-light partnerships. The central test will be whether Nebius can repeatedly obtain similarly priced project-level debt while meeting the delivery schedules and performance requirements attached to its large customer contracts.