The question credit-minded institutional investors are asking about Nscale is not whether the demand for AI compute is real. It almost certainly is. The question is sequencing: who pays for the infrastructure before the contracts start generating cash?
That tension sits at the center of the S-1 Nscale filed with the SEC on September 18, 2026, and it is the reason the prospectus deserves more scrutiny than a headline customer list alone would suggest.
The Gap That Matters
Driven by long-term take-or-pay contracts, active and contracted total contract value reached $103.4 billion as of August 31, 2026, up from $38.0 billion at the end of 2025. That acceleration is genuinely striking. What is less prominent in the coverage is the denominator. Only $2.6 billion of that total contract value was active by August 31. Microsoft and Anthropic accounted for up to about $88.4 billion of contractual payments. The rest is committed but not yet generating revenue, because the physical infrastructure to deliver it does not fully exist yet.
A substantial portion of the company’s contracted business is concentrated among a small number of very large customers. Nscale said its largest customer accounted for 52% of revenue during the first half of 2026.
The Financing Condition Hidden in the Anthropic Deal
The Anthropic agreement anchoring the prospectus carries a clause that institutional buyers will read carefully. Under the Anthropic Services Agreements, Nscale must use its best efforts to obtain qualifying financing for the required GPU equipment and data center infrastructure. As of the prospectus date, Nscale had not obtained binding commitments for any financing required to perform the agreements.
The contracts allow termination under specified circumstances if qualifying financing remains unavailable. That is not a technicality. It does not erase the demand. It makes financing part of product delivery. Nscale needs lenders and equity investors to fund the hardware and campuses before the contracts can generate their advertised value.
What the Numbers Actually Show
Revenue for the six months ended June 30, 2026, surged 1,252% year-over-year to $140.6 million, compared to $10.4 million in the prior-year period. However, heavy capital commitments and capacity build-outs weighed on near-term profitability, with net loss for the first half of 2026 expanding to $1,020.1 million from $368.9 million a year earlier. Part of that widening loss reflects accounting rather than pure cash burn: $457.1 million of the net loss came from losses on fair value adjustments tied to financing instruments. Even stripping those out, operating losses reached $492.0 million during the latest six-month period.
Nvidia will take at least $1 billion of Nscale’s $3.1 billion in convertible loan notes; the company said it has raised more than $4.6 billion in debt since the start of 2026. That debt load, combined with an equity raise targeting up to $3 billion, frames the real ask: Nscale is not merely selling investors a share of future AI revenues. It is asking them to fund the delivery mechanism for contracts that do not fully activate until it does.
What Investors Are Missing
The conventional read on Nscale is that elite customers validate the business. That is not wrong, but it undersells the circular nature of the capital structure. Nvidia is a key supplier and a major financing counterparty. That means the chipmaker sitting at the center of AI demand is simultaneously a vendor, a creditor, and an equity-linked participant in one of its largest customers. If Nscale’s buildout slows, Nvidia’s GPU order pipeline slows with it.
Stocks to Watch
- Nscale (NSCL): The pricing amendment, not the S-1, is the document that resolves the valuation argument. Nscale said its March 2026 Series C round valued the company at $14.6 billion. Whether public investors accept a premium or discount to that mark depends on how confidently they price the financing close.
- Nvidia (NVDA): Its position as supplier and financing counterparty makes Nscale’s IPO outcome a direct signal for GPU demand assumptions embedded in Nvidia’s own forward estimates.
- CoreWeave (CRWV): The comparable that went public first. Nscale’s prospectus gives analysts a second data set to test whether hyperscale AI infrastructure can justify its loss profile on contracted backlog alone.
- Iris Energy (IREN) and Nebius (NBIS): Smaller AI infrastructure players that benefit if Nscale’s listing validates the sector and attracts institutional capital flows, but face compression if Nscale prices aggressively and absorbs available demand.