Lambda’s IPO case depends on contracts funding the GPUs

Lambda is seeking as much as $4 billion at a $14.5 billion pre-money valuation ahead of a planned 2027 IPO. The question is whether the customer commitments supporting Lambda’s valuation can also support the debt required to bring GPU clusters online.

The company’s reported backlog rose from $15 billion in June 2026 to $50 billion in September 2026. But $35 billion of that September total appears to come from an Anthropic commitment signed in late August 2026. The agreement adds a major demand signal, but it also makes Lambda’s near-term case dependent on one large customer relationship and the economics of delivering its capacity.

Coatue Management and Blackstone are leading the proposed equity round. Lambda, Coatue, and Blackstone did not provide comment on the financing. The round would be Lambda’s last major private financing before the proposed public listing, which had reportedly been targeted for 2026 before shifting to 2027 amid market uncertainty.

×

Lambda is using a structured debt model rather than simply holding more cash on its balance sheet. On October 1, 2026, it closed a $1.008 billion delayed-draw term loan with a 6.78% fixed interest rate. The facility is structured so that capital is drawn as clusters reach commissioning milestones, rather than all being funded at closing. This connects the cost of the loan to infrastructure entering service—the point at which GPU systems can begin fulfilling contracted work.

The financing was marketed to insurance companies and fixed-income investors, received an A (low) rating from Morningstar DBRS and Baa1 from Moody’s, and matures on May 30, 2033. It is secured by the GPU servers and related infrastructure it finances, as well as the contracted cash flows attached to those deployments. Lambda says the proceeds will support three committed customer deployments across multiple data centers, backed by two investment-grade offtakers.

Recommended reading

Nscale’s $3.36B note includes $1B due in November

Sergey Kuznetsov • Sep 25, 2026 • 3 min read

Lambda has not named those two customers in its financing announcement or specified how the three deployments map to its reported Anthropic contract. The company’s public statement establishes that the debt is supported by two investment-grade customers; it does not establish whether Anthropic is one of them.

A delayed draw is not a blank check

The official financing announcement calls this Lambda’s first U.S. fixed-rate financing and its first institutional debt transaction above $1 billion. It says the loan was oversubscribed and priced inside its target range. The structure says more than the marketing language.

A delayed-draw loan does not make physical infrastructure cheaper. Lambda must still purchase, deploy, and commission GPU capacity, and the debt is ultimately repaid under a fully amortizing schedule. The design keeps borrowing aligned with delivery milestones. That limits the time Lambda pays for capital before an associated cluster is in service, while giving lenders security in the servers and customer cash flows tied to each deployment.

The 6.78% rate is fixed and paid through semi-annual coupons. That gives Lambda certainty on this facility’s interest cost through its May 2033 maturity, but it also creates an obligation that does not flex with GPU utilization, customer demand, or the timing of an IPO. The company has not disclosed the total amount it expects to draw at each commissioning milestone, the depreciation assumptions behind the collateral, or revenue and margin figures for the deployments. Those omissions matter when assessing whether a fast-growing backlog becomes cash flow on schedule.

The October facility is Lambda’s second institutional credit facility. The company says it has raised approximately $1 billion in each of its major 2026 financings, including the broadly syndicated loan that closed on August 27, 2026. The new loan also represents Lambda’s third newly accessed credit market in 18 months, according to Chief Executive Officer Michel Combes.

“The capital in this offering underwrites infrastructure in decades, not quarters, and has funded us as a private company on the strength of our customer contracts.”

Lambda is attempting to finance GPU infrastructure through contracted demand rather than speculative capacity construction. But customer contracts are not the same as cash already collected, and the company’s description shows why credit investors are central to the model. The loan is secured by GPU hardware and by the cash flows that those customer contracts are expected to generate.

The $35 billion concentration problem

The backlog increase is the central fact prospective IPO investors will have to examine. Lambda added $35 billion to a $50 billion reported backlog in a period that coincided with the Anthropic agreement. Even if the commitment is firm and economically attractive, it shifts the company’s story from broad customer diversification toward execution against a single unusually large deployment relationship.

That does not mean the Anthropic deal is weak. Its precise terms determine far more of Lambda’s valuation case than the aggregate backlog figure alone suggests. The available reporting does not state the agreement’s duration, the timing of capacity deliveries, cancellation protections, payment schedule, service-level requirements, or how much capital Lambda must deploy before revenue is recognized. It also does not establish which party bears the risk if hardware delivery or data-center commissioning slips.

These contract details matter for a neocloud. Lambda is raising equity while layering secured, fixed-rate debt behind physical GPU deployments. A backlog number can support that strategy only if contractual milestones, customer payments, and cluster commissioning stay aligned. The delayed-draw structure is built around that alignment, which is why the $1.008 billion facility is more informative than a generic statement that the company has raised another $1 billion.

The primary document adds another constraint absent from the equity-round headline: the loan funds three committed deployments for two investment-grade offtakers, rather than an unrestricted expansion pool. That may lower credit risk for lenders, but it means the financing is tied to designated infrastructure and contracted cash flows. Investors in the proposed $4 billion equity round are funding a broader company whose public valuation will depend on whether such project-level debt can be repeated at scale.

Nscale offers a nearby IPO comparison, not a clean one

Lambda is not the only neocloud using private capital to reach public markets. In September 2026, we reported that British competitor Nscale raised $3.36 billion ahead of a planned NYSE IPO, including $1 billion from Nvidia that was not due until mid-November. Nscale had claimed contracts exceeding $103 billion, a planned IPO valuation of $35 billion, and a target raise of $3 billion.

The comparison shows the same financing pressure: large reported contracts require substantial infrastructure capital before all related revenue is realized. But the structures disclosed so far are materially different. Nscale’s reported package included a convertible note and a delayed Nvidia tranche. Lambda’s October financing is a secured, fixed-rate, delayed-draw term loan with named ratings, a semi-annual coupon, GPU-server collateral, and a final maturity date.

The numbers are not interchangeable. Lambda calls its $50 billion figure backlog, while Nscale’s previously reported number was “claimed contracts.” Neither company’s supplied disclosures provide enough common detail to compare contract duration, capacity obligations, recognized revenue, or expected margins. Both companies are using reported demand to obtain the equity and debt needed to build supply.

Debt-market access could also differentiate the companies. Lambda’s ability to obtain investment-grade ratings and fixed-rate institutional funding could reduce its reliance on repeated equity raises for specific deployments. Those benefits depend on commissioned systems and contracted payments supporting the secured loan. The company’s October announcement explicitly warns that its forward-looking statements are not guarantees and may differ materially from actual results.

What the proposed round would actually finance

A $4 billion round at a $14.5 billion pre-money valuation would give Lambda more than a valuation marker before an IPO. It would supply another equity buffer while the company commits capital to GPU infrastructure and continues using credit facilities for specific deployments. The reporting does not disclose the round’s final amount, the post-money valuation, a closing date, or the ownership stakes of Coatue and Blackstone. It also does not state whether the new equity is earmarked for particular data centers or customer commitments.

That missing detail prevents a simple conclusion that the equity round solves Lambda’s buildout needs. The company’s latest debt instrument covers GPU infrastructure attached to three deployments; the proposed equity raise is much larger but less specifically described. The operational test is whether Lambda can convert its reported backlog into commissioned clusters quickly enough that secured debt remains covered by customer cash flows and fresh equity is used for growth rather than schedule slippage.

AI infrastructure planning requires coordinated decisions across compute, networking, software, and memory. Lambda’s financing adds contract terms, construction timing, collateral, and the cost of capital. A GPU order without an energized, commissioned cluster does not activate a delayed loan draw or produce the cash flow pledged to lenders.

The proposed IPO will test more than appetite for GPU exposure. Lambda has disclosed a mechanism for funding commissioned deployments, but its valuation rests heavily on a $35 billion customer commitment whose operating and payment terms remain undisclosed. Until those terms—or the resulting revenue and utilization figures—are public, the $50 billion backlog is a promise that still has to finance the hardware behind it.