Lambda borrows $917M to buy chips from its own investor

Lambda is raising $917 million in leveraged debt to finance a major hardware acquisition, with the funds earmarked for a contract involving Nvidia chips and related infrastructure, according to a Monday report from Bloomberg. Morgan Stanley is leading the transaction, which was first detailed by reporters Jeannine Amodeo, Paula Seligson, and Gowri Gurumurthy.

The borrowing entities are Lambda Compute II LLC and Lambda Cloud Canada Inc. A lender call kicked off at 10:30 a.m. New York time on Monday, with commitments due by Thursday. Pre-marketing efforts have already drawn interest totaling nearly $2 billion in orders, signaling robust appetite for the paper.

Bloomberg positions this deal as the latest escalation in the debt-fueled race to build out artificial intelligence infrastructure. Its data shows global AI-linked borrowing has reached nearly $600 billion since last year. Lambda sits squarely in the category of “neoclouds,” firms that lease access to GPUs and other compute resources rather than selling hardware outright.

A closed loop with Nvidia

The transaction’s most striking feature is who sits on the other side. Nvidia already holds an equity stake in Lambda, and it supplies every chip the company runs, since Lambda operates exclusively on Nvidia silicon.

The relationship deepened in September 2025. Nvidia agreed to lease GPUs back from Lambda under a $1.3 billion, four-year deal covering 10,000 servers, plus a separate $200 million contract for 8,000 additional servers. RCR Wireless reported that this arrangement made Nvidia Lambda’s largest customer, with the combined agreements worth roughly $1.5 billion across 18,000 servers.

That means Nvidia plays four roles at once: investor, supplier, biggest tenant, and counterparty to the very contract this loan exists to fund. The circularity is not unprecedented. Google built a similar loop around its own TPUs, guaranteeing rental income for data centres that buy its chips to serve Anthropic. The Bank for International Settlements flagged this pattern in June, warning that a collapse in AI investment could ripple through credit markets with severity comparable to 2008.

The BIS specifically called out chipmakers taking equity stakes in firms that then commit to purchasing chips from those same investors. The terms of such arrangements, it noted, “are typically poorly disclosed, with risks of the same asset being pledged multiple times.”

The venue matters more than the number

At first glance, $917 million looks modest next to the sums flowing through AI infrastructure. The venue, however, is the story. This deal lands in the institutional leveraged loan market, where investors buy debt issued by companies rated below investment grade.

CoreWeave opened that door in April, with Bloomberg calling its $3.1 billion deal the first of its kind to finance chips in the institutional loan market. CoreWeave followed up in May by selling debt backed by customer contracts, including one with OpenAI. A later deal tied to different contracts forced the company to accept a hefty yield, pushing its borrowing costs sharply higher.

Compare that with an alternative route. Nebius raised a $775 million GPU-backed facility in July at SOFR plus 2.50 percentage points, roughly 6.8%. Banks syndicated that deal, and an investment-grade customer contract stood behind it, generating cash flows that covered more than 100% of the capital expenditure.

Lambda, by contrast, pays as much as 3.75 percentage points over the benchmark rate, at a discounted price of 99 cents on the dollar. Same collateral class, different market, meaningfully higher cost.

Terms reveal the nervous party

The loan structure itself tells you what lenders demanded. Maturity runs 4.4 years, while institutional loans typically stretch to seven. It also amortises fully, meaning the debt repays gradually over roughly four years rather than hitting as a lump sum at maturity. Bloomberg notes that amortisation ranks among the protections investors have been seeking, and here it eliminates refinancing risk entirely.

A call-protection clause adds a penalty if Lambda redeems early. Bloomberg describes the combination as closer to a bond deal than a traditional loan.

These features serve one specific purpose: they shorten the window in which a GPU must keep earning its keep. TNW raised the open question when Nebius borrowed in July. The entire asset class rests on residual value assumptions that nobody in this industry has yet been forced to test. A four-year amortising schedule sidesteps that uncertainty rather than resolving it.

What Lambda actually is

Machine learning engineers founded the company in 2012, working out of the Noisebridge hackerspace in San Francisco’s Mission District. It now brands itself as an AI-only company, building modular AI factories with power, liquid cooling, and high-bandwidth interconnects at what it calls gigawatt scale. Its mission statement promises to “make compute as ubiquitous as electricity and give everyone in America the power of superintelligence.”

A management overhaul followed a $1.5 billion Series E in November 2025. Lambda named Michel Combes chief executive in May, after his stints at Sprint, SoftBank International, and Alcatel-Lucent. Former AT&T Communications boss John Donovan became chairman. Co-founders Stephen and Michael Balaban shifted to chief technology officer and chief product officer roles. The company targets 3GW of AI compute under management by 2030.

Microsoft signed a multibillion-dollar agreement with Lambda last November to deploy tens of thousands of Nvidia GPUs, CNBC reported. Lambda has also held preliminary discussions with bankers about a public listing. It sells only in North America, and it has started constructing its own data centres rather than leasing capacity from third parties.

What the deal does not settle

Investor demand is clearly there. Order books near $2 billion against a $917 million loan point to genuine appetite, and the amortising structure suggests buyers went in with full awareness of the risks.

Several questions remain open. Lambda and Morgan Stanley did not respond to Bloomberg’s requests for comment. The detail tying the debt to an Nvidia contract comes from an unnamed source. Lambda’s revenue and loss figures sit behind paywalled reporting, so no audited numbers are publicly available.

The broader market has also grown pickier. Loan investors began pushing back this month as anxiety rose, Bloomberg reported. Big Tech alone has lined up $350 billion of AI debt over five years, and private credit keeps piling on more. Blue Owl’s Stack Infrastructure is pursuing a $5.9 billion data centre loan of its own.

Lambda leaves lenders with one question, and it has nothing to do with chips. Nvidia sits on both sides of this trade. If AI demand holds, that reads as alignment. If it slips, the same fact reads as concentration, and the loan matures long before anyone finds out which.