Nscale’s IPO reveals $140.6m revenue against $1.02bn first-half loss

Nscale Limited has filed a Form S-1 with the U.S. Securities and Exchange Commission for an initial public offering on the New York Stock Exchange under the proposed ticker NSCL, bringing one of Europe’s fastest-growing artificial intelligence infrastructure companies closer to public markets. The filing does not yet specify the number of shares, offering price or formal IPO valuation, so reports that Nscale may seek a valuation around $30 billion should be treated as an expected target rather than a priced transaction. What the filing does establish is extraordinary operating growth: first-half 2026 revenue reached $140.6 million, up 1,252% from $10.4 million a year earlier, while the company simultaneously recorded a $1.02 billion net loss.

Nscale’s scale ambitions are even larger than its current income statement. As of August 31, the company disclosed approximately $2.6 billion of active total contract value and $103.4 billion of active and contracted TCV under long-term take-or-pay arrangements, supporting approximately 461,000 GPUs that were active or contracted. Its owned and controlled power pipeline has expanded beyond 10GW, while operations now span 14 regions and the workforce has grown from about 40 employees to more than 1,000 in roughly two and a half years.

How should investors interpret $103.4 billion of Nscale contract value against only $140.6 million of half-year revenue?

The difference illustrates how early Nscale remains in converting its infrastructure pipeline into recognized revenue. The $103.4 billion figure represents active and contracted total contract value across long-duration agreements rather than revenue already earned or cash already collected. Nscale says the weighted average contract life was approximately 5.7 years as of August 31, which means a substantial portion of the headline contract value is expected to be recognized over several future years.

The company’s largest disclosed agreements illustrate the scale. Between September 2025 and April 2026, Nscale entered into Microsoft statements of work providing for payments of up to approximately $43.8 billion through December 2033, subject to delivery and service-availability requirements. On August 25, Nscale entered Anthropic services agreements providing for aggregate payments of up to approximately $44.6 billion, again dependent on meeting specified infrastructure and availability obligations.

These are commercially powerful contracts because take-or-pay structures can give infrastructure developers greater revenue visibility and help support debt financing. They are not risk-free receivables. Nscale still has to construct the required capacity, deliver GPUs and associated infrastructure on schedule, maintain availability and satisfy contractual conditions before the entire value becomes economically realized.

That distinction will become central to the IPO debate. Public investors are being asked to value not only a cloud business producing $140.6 million of half-year revenue, but an enormous future infrastructure-delivery obligation tied to tens of billions of dollars of contracted demand.

Why did Nscale lose $1.02 billion while revenue was growing more than twelvefold?

The company’s net loss widened from $368.9 million in the first half of 2025 to $1.02 billion in the first half of 2026. The loss included an operating loss of approximately $492 million and $457.1 million of fair-value adjustments, meaning the billion-dollar headline does not consist entirely of cash operating expenses. Adjusted EBITDA was negative $199.2 million, equivalent to a negative 142% margin.

The operating economics still show how expensive the growth phase has become. Cost of revenue increased to $189.6 million from $7.8 million as Nscale expanded facilities and absorbed more rental and electricity costs, already exceeding the $140.6 million of reported revenue before other corporate and technology expenses were considered. Product and technology expenditure increased to $51.3 million as headcount and share-based compensation grew.

This is the central financial tension facing AI infrastructure companies. Customer demand can be extraordinary while current margins remain poor because billions of dollars must be committed to data centres, GPUs, power and networking before the underlying contracts reach mature utilization.

Nscale is attempting to bridge that period using long-term contracts, customer commitments, debt and equity capital. The model can become extremely powerful if utilization rises quickly enough because the same physical infrastructure begins producing recurring cash flow. It can become dangerous if construction and financing costs arrive faster than contracted workloads become operational.

How concentrated is Nscale’s customer base despite the enormous contract portfolio?

Nscale disclosed that its largest customer accounted for 52% of first-half 2026 revenue, down from 73% for full-year 2025. In 2024, one customer accounted for substantially all revenue, demonstrating that customer diversification is improving but remains a significant risk.

Large customer concentration is partly unavoidable in hyperscale AI infrastructure because only a limited number of companies currently purchase compute in tens of billions of dollars. Microsoft and Anthropic can support campus-scale developments that would be impossible to justify through hundreds of small enterprise customers.

The same concentration increases bargaining and execution risk. Delays, changes in deployment plans or reduced spending from even one major customer could affect a much larger portion of Nscale’s financial trajectory than equivalent customer changes would at a diversified hyperscaler.

The company is attempting to broaden that model through Nscale Cloud and the planned acquisition of Anyscale, which would bring roughly 200 employees focused on workload performance and infrastructure utilization. This can push Nscale toward enterprise software and managed AI services instead of remaining almost entirely dependent on extremely large physical infrastructure contracts.

Why does Nscale’s 10GW power pipeline matter as much as its GPU contracts?

AI accelerators are only valuable once they can be connected to electricity, cooling and networks. Nscale says its owned and controlled power portfolio has expanded from approximately 750MW to more than 10GW, providing a large pipeline from which future campuses can be developed.

This power control is an increasingly valuable competitive asset because grid access has become one of the principal limits on global AI infrastructure development. A provider can have financing and GPU allocations but still fail to deliver capacity if utility connections arrive several years late.

Nscale’s vertically integrated strategy is designed around controlling land, power, data centres and compute rather than purchasing each component only after a customer contract appears. That can accelerate customer deployment and potentially lower costs through scale.

It also creates substantial capital risk. A 10GW pipeline is very different from 10GW of operating capacity, and each additional gigawatt can require billions of dollars of construction and equipment investment before producing revenue.

The IPO can therefore be viewed partly as a capital-market tool for converting power rights and customer contracts into operating infrastructure. Investors need to determine how much shareholder capital will be required before that pipeline begins generating mature free cash flow.

Does a reported $30 billion target valuation make sense after Nscale’s March financing?

Reuters reported that Nscale is expected to pursue a multibillion-dollar listing and may target a valuation around $30 billion, while the company was valued at approximately $14.6 billion in a March financing. The IPO filing itself does not contain a price range, so no $30 billion valuation has yet been formally established.

A doubling of valuation within months would normally look aggressive for a company with only $140.6 million of first-half revenue. Nscale’s argument is clearly based on forward contracted infrastructure rather than conventional current-sales multiples.

The $103.4 billion of contracted and active TCV gives investors a much larger economic denominator to examine, but discounting that value requires assumptions around construction timing, financing expense, contract performance, GPU depreciation and customer concentration.

This is why Nscale could become one of the most revealing AI IPOs of the current cycle. It asks public investors to price a company where current revenue remains tiny relative to both losses and future contract commitments.

The investment case depends on believing that infrastructure under construction eventually converts the extraordinary contract book into recurring cash flow faster than the capital structure becomes burdensome.