Nebius arranges $775M secured facility for AI-cloud expansion

Two indirect Nebius Group subsidiaries entered an approximately $775 million senior secured term-loan facility on July 10, 2026. Nebius disclosed the agreement in a Form 6-K filed July 17 and said the financing is intended primarily for expansion of its full-stack AI cloud.

The filing identifies the borrowers as Nebius Compute II, LLC and Nebius Compute II Oy. The facility matures on October 31, 2030. Its annual interest rate is based on one-month Term SOFR, with a 0% floor, plus a 2.50% margin. The borrowers may voluntarily prepay after ten business days' notice without a premium or penalty, apart from customary breakage costs.

The security package is broader than a simple loan against GPUs. Subject to stated exceptions, the obligations are secured by substantially all assets of the two borrower subsidiaries and their indirectly held shares. Nebius Group provided limited guarantees for specified bad acts and performance obligations rather than a broad parent-company repayment guarantee.

The distinctions matter. Public documents do not disclose how much of the facility has already been drawn, the full fee schedule, the minimum-liquidity threshold, the collateral valuation or the identity of the investment-grade customer supporting this deployment. Microsoft and Meta are named elsewhere in Nebius's customer pipeline, but readers should not infer that either company is the unnamed customer behind this facility.

AI-cloud providers must spend heavily on accelerators, power, networking and facilities before new capacity produces revenue. A secured facility tied to deployed infrastructure and contracted cash flows can provide buildout capital without relying only on new equity. That does not make the expansion risk-free or guarantee faster delivery.

For customers evaluating an AI-cloud provider, the relevant questions remain operational: when clusters become available, what service-level commitments apply, how upgrades are handled and how concentrated capacity is among a small number of contracts. For the provider, utilization, accelerator obsolescence, energy costs, customer concentration and refinancing determine whether the financing remains economical.

This transaction shows how AI infrastructure is becoming a capital-engineering problem as well as a systems-engineering problem. Nebius has completed one asset-backed financing structure that it says may be reused for other deployments. The stronger long-term signal will be whether the financed capacity arrives on schedule, stays utilized and generates enough cash to service the facility while preserving flexibility for the next hardware generation.