Nscale Eyes $30B IPO Amidst AI Infrastructure Demand

European AI infrastructure provider Nscale has filed its registration statement with the U.S. Securities and Exchange Commission (SEC) for an initial public offering on the New York Stock Exchange under the ticker symbol NSCL. This move signals a significant ambition for the company, which has rapidly transitioned from a Bitcoin mining capacity lessor to a prominent player in the European AI compute power sector within a mere two and a half years. The filing underscores the substantial reliance on future contractually obligated revenue, highlighting a business model heavily weighted towards capacity yet to be delivered.

The company is reportedly targeting a valuation of approximately $30 billion, a substantial leap from its most recent private funding round where it secured $2 billion at a $14.6 billion valuation. This IPO, with Goldman Sachs, J.P. Morgan, and Morgan Stanley acting as lead underwriters, aims to capitalize on the intense demand for AI-tailored data center solutions. Notably, Nvidia, a critical hardware supplier, has also committed $3.1 billion in convertible notes, which will convert into non-voting shares during the IPO, underscoring its strategic importance and financial backing.

Nscale's financial trajectory shows explosive revenue growth, with first-half 2026 revenue reaching $140.6 million, a 1,252% increase from the prior year's $10.4 million. However, this top-line surge is accompanied by a significant net loss of $1.02 billion in the same period. A considerable portion of this loss stems from valuation adjustments on convertible instruments, which are expected to be resolved post-IPO. The operational loss stands at $492 million, with direct costs of services exceeding revenue even before accounting for substantial depreciation charges, indicating current operations are running below cost.

The core of Nscale's valuation narrative rests on its immense order backlog, valued at $103.4 billion. Of this, only $2.6 billion represents currently active contracts, with the vast majority committed for future capacity deployment. This future-oriented backlog is mirrored in its operational metrics: Nscale currently operates around 25,000 GPUs but has contracted for 461,000. Similarly, its data center capacity is projected to grow from 1.37 gigawatts to a much larger figure, with significant expansion planned, including the recently acquired Monarch Compute Campus in West Virginia.

Customer concentration remains a key characteristic, with a single client accounting for 52% of first-half 2026 revenue, down from 73% the previous year. However, the bulk of the order book is anchored by two major agreements: Microsoft has committed up to $43.8 billion through 2033, and Anthropic has agreed to up to $44.6 billion. These agreements are contingent on Nscale's ability to deliver the contracted capacity and availability. A significant caveat exists for the Anthropic deal, as Nscale has not yet secured binding financing commitments for the required capacity, only agreeing to use best efforts to obtain it.

The company faces substantial capital requirements, with open purchase obligations for technology equipment and construction services totaling over $27.5 billion, largely due in the current and next fiscal years. While Nscale held nearly $1.5 billion in liquidity at mid-year, its financial planning has previously raised concerns about its ability to continue as a going concern, a doubt it aims to mitigate through flexible capital expenditure management. The strong operating cash flow of $1.69 billion in the first half was primarily driven by $1.6 billion in customer prepayments, while capital expenditures consumed $3.29 billion.

Supply chain concentration is also a factor, with three suppliers dominating the procurement of Nvidia GPUs and related infrastructure. The company's reliance on rapid technological advancements and the potential for shorter-than-expected asset lifecycles present ongoing risks. The IPO proceeds are intended to fund its aggressive expansion plans and meet its significant contractual obligations in the rapidly evolving AI compute market.