Lambda Borrows $1B More to Keep Nvidia Chips Flowing to Microsoft

Lambda has taken on another $1 billion in short-dated private debt, this time to buy Nvidia chips it plans to lease directly to Microsoft. JP Morgan Chase arranged the facility, adding to a borrowing spree that has made the AI cloud provider one of the more aggressive users of debt financing in the sector this year.

The deal follows a $926 million loan Lambda closed to fund Nvidia GB300 GPUs under a separate supply arrangement with Nvidia itself. Combined with a $1 billion secured credit facility the company closed back in May 2026, Lambda has now stacked roughly $3 billion in debt within a matter of months, all tied to acquiring the physical hardware that underpins its cloud business.

The pattern reflects a broader shift in how "neoclouds" (companies that rent out GPU capacity rather than build consumer-facing AI products) are financing growth. Rather than diluting equity further, Lambda is leaning on lenders to front the capital for chip purchases, then recouping the cost through long-term leasing contracts with hyperscalers like Microsoft.

A financing pattern with sector-wide reach

Lambda is not acting in isolation. Banks and technology companies have collectively raised more than $400 billion in AI-related debt globally so far in 2026, a scale that reflects just how capital-intensive the GPU leasing business has become.

Lambda's own balance sheet has grown quickly. The company raised $1.5 billion in venture capital in November 2025 at a $5.43 billion post-money valuation, and it is now reportedly in talks for a $3 billion pre-IPO round, on top of the debt it has piled up to fund chip purchases.

That combination, equity to fund growth and debt to fund hardware, has become the standard playbook for neocloud operators racing to lock in Nvidia supply before larger rivals do. For Lambda, leasing that hardware straight to Microsoft offers a predictable revenue stream to service the new loan, but it also ties the company's near-term finances closely to the pace at which hyperscalers keep signing multi-year GPU contracts.

Whether that model holds up depends largely on demand from Microsoft and other cloud customers staying as strong as it has been through 2026. If leasing commitments slow, the debt load Lambda has taken on to buy chips ahead of demand becomes a much heavier obligation to carry.