DataNSCLSEP 21, 2026

Nscale S-1 discloses $103.4B backlog, locks compute cost floor beneath AI sales tools

Anthropic and Microsoft take-or-pay contracts totaling $88.4B anchor the input costs behind every AI customer-acquisition tool small-business owners are evaluating.

BY LINH VINH · DATA INFRASTRUCTURE

Nscale, the London-based AI cloud provider, filed its Form S-1 with the SEC on September 18, 2026, disclosing $103.4 billion in active and contracted total contract value as of August 31 and setting up a New York Stock Exchange listing under the proposed ticker NSCL. Goldman Sachs, J.P. Morgan and Morgan Stanley are lead bookrunners.

Buried in the risk factors is the single most consequential disclosure the AI industry has produced this year: the actual dollar shape of the compute contracts anchoring frontier model economics. Microsoft’s agreements provide for payments of up to $43.8 billion. Anthropic’s provide for up to approximately $44.6 billion. Combined, that’s $88.4 billion of take-or-pay obligation, with Anthropic’s tranche requiring dedicated NVIDIA Vera Rubin infrastructure at Nscale’s Monarch Compute Campus in West Virginia.

Take-or-pay is the operative phrase. These aren’t consumption estimates. They’re contractually fixed floors, running for the better part of a decade, that price the compute layer beneath every AI application a small business is currently evaluating.

The financials underneath the backlog are, by conventional standards, alarming. Revenue for the six months ended June 30 was $140.6 million against a net loss of $1.02 billion. Remaining performance obligations sit at $56.4 billion. One customer accounted for 52% of first-half revenue. Nscale carries more than $8 billion in debt excluding a Dell financing arrangement, and per Techstrong.ai’s read of the filing, hasn’t yet obtained binding commitments for the financing required to fulfill the contracts.

Read as an equity story, this is a coin flip. Read as a structural disclosure, it’s the clearest picture yet of why AI tooling prices at the application layer aren’t going anywhere fast: the compute suppliers can’t discount what they’ve already contractually promised to charge. The 2019 WeWork S-1 taught the market to read backlogs skeptically. The Nscale filing invites the opposite read. The backlog is the point. The $30B NYSE valuation math sits on top of it, but the floor is what matters.

For anyone waiting on cheaper AI before committing, the S-1 quantifies the wait.

Sources

About the author

Linh Vinh

DATA INFRASTRUCTURE

Linh Vinh covers vector databases, retrieval, feature stores, and the plumbing layer of enterprise AI. She has been writing about data infrastructure for several years and files on infra vendor consolidation, pricing pressure, and technical architecture shifts.