Lambda has raised approximately $1 billion in private short-dated debt, arranged by JPMorgan, to purchase Nvidia GPUs that Microsoft will lease. The transaction, marketed to private placement investors, represents a growing trend in AI infrastructure financing where debt is secured not by the borrower's own cash flows but by the contractual obligations of a major technology tenant.
The structure is notable because Lambda is borrowing against Microsoft's willingness to keep paying, rather than against its own revenue. Microsoft gains access to high-demand AI compute without having to carry the debt on its balance sheet. This arrangement lets Lambda scale its GPU fleet while transferring much of the credit risk to investors who are betting on Microsoft's long-term commitment to AI workloads.
Second major raise this month
This is the second significant debt raise for Lambda in a short period. Two weeks earlier, the company borrowed $917 million against a contract with Nvidia, which plays a triple role as Lambda's investor, its supplier, and a customer. That unusual triangular relationship highlights how deeply intertwined AI hardware vendors, cloud providers, and specialized compute companies have become.
The terms of the latest deal have not been independently verified. Reports attribute the information to people who were not authorised to speak publicly, and JPMorgan declined to comment when approached. Despite that lack of official confirmation, the deal fits a broader pattern of AI-related lending that has accelerated sharply in recent months.
Growth of AI-linked private credit
According to industry tracking, more than $400 billion of AI-related debt has been raised globally in 2026 alone. This includes a wide range of structures, from corporate bonds to asset-backed loans tied to data centers, chips, and cloud contracts. Private credit funds and institutional investors have become major players, stepping in where traditional bank lending might be constrained by regulatory capital requirements.
Lambda is also reported to be in discussions to raise as much as $3 billion ahead of a possible public listing next year. A successful initial public offering would give the company access to equity markets and could help refinance some of the short-dated debt it has been using to fund GPU purchases. The company's aggressive expansion reflects the intense demand for Nvidia accelerators, which remain the backbone of most large-scale AI training and inference projects.
Similar structure at Nebius
Lambda is not alone in using this financing approach. Nebius, an Amsterdam-headquartered company, raised $775 million against its own GPUs earlier this year. That marked the first secured debt taken by the company. Nebius's anchor tenant is the same as Lambda's: Microsoft. Nebius holds a five-year contract with Microsoft worth $19.4 billion and has indicated that it has around $40 billion of additional contracts that could be securitised in the future.
The Nebius deal demonstrates that lenders are willing to accept AI hardware as collateral, but the value of that collateral depends heavily on the creditworthiness of the end customer. In both cases, Microsoft's commitment provides the comfort that investors need. The actual chips sit in data centers, operating and generating value, but ownership remains with the borrowing entity or its creditors.
The collateral chain and its risks
These transactions create a relatively short collateral chain. A European lender is ultimately underwriting one American company's continued appetite for compute it does not own. If Microsoft were to cancel or stop paying under the lease, the chips would remain in place, but the expected cash flows would disappear. The lender would be left with physical assets whose value could decline rapidly in a market downturn or if newer, more powerful chips enter the market.
This risk has not gone unnoticed by financial regulators. The European Central Bank warned in May about the opaque valuation practices and limited liquidity in private credit markets. It specifically pointed to portfolios concentrated in a few US issuers whose valuations are closely tied to the AI narrative. A sudden shift in sentiment about AI profitability could leave these lenders exposed to losses that are difficult to quantify.
The Bank for International Settlements was blunter in June. It warned that an AI investment collapse could disrupt credit markets on the scale of the 2008 financial crisis. The BIS also noted that poor disclosure of deal terms makes it hard to tell whether the same asset has been pledged twice in separate financing arrangements. Double pledging could multiply losses and create hidden interdependencies among lenders.
European regulatory and policy dimensions
The situation lands awkwardly in Brussels. The companies that the European Union points to when it discusses sovereign compute capabilities are financing themselves against an American lease. The EU has talked about building strategic autonomy in AI infrastructure, but the financial structures underpinning some of its champions rely heavily on US technology giants as the ultimate source of repayment.
This dependency undermines the narrative that European AI companies are building independent capacity. If Microsoft were to reduce its commitments or renegotiate contracts, the economic foundation of these companies would weaken, and the EU's aspirations for homegrown compute power would suffer a significant setback.
There is also a broader question about where the risk ultimately resides. The chips will sit somewhere and someone will own them. Under the current structure, the party carrying the loss in a downturn is not the one using the compute. The user, whether Microsoft or another tenant, gets the benefit of the hardware without bearing the financing risk. That risk sits with the lenders, and by extension, with the broader private credit market.
As AI infrastructure financing grows in scale and complexity, regulators are paying closer attention. The combination of short-dated debt, rapidly depreciating hardware, and concentration in a small number of large technology customers could create systemic vulnerabilities. The deals involving Lambda and Nebius are innovative, but they also test the limits of how much risk can be safely transferred through collateralized lending.
For now, the appetite for AI debt remains strong. Investors are drawn to the high yields and the perceived safety of contracts backed by companies like Microsoft. But the warnings from the ECB and the BIS suggest that the true risk profile of these transactions may be poorly understood. Without better disclosure and more thorough stress testing, the market could be building a fragile edifice on the assumption that AI demand will continue growing indefinitely.