Nscale Files for US IPO After AI Cloud Revenue Surges 1,252%

Nvidia-backed Nscale reported $140.6 million in first-half revenue and more than $103 billion in contracted value as it prepares for a US public listing.

The AI boom is creating a new class of cloud infrastructure companies built specifically around the enormous compute requirements of frontier models.

Nscale is now preparing to test public-market appetite for that thesis. Reuters reported on September 18 that the Nvidia-backed British AI cloud provider filed for a U.S. IPO after revenue surged 1,252% in the first half of 2026.

The company reported $140.6 million in revenue for the six months ended June 30, compared with $10.4 million a year earlier. Its net loss, however, reached $1.02 billion.

Nscale’s filing captures the defining economics of the AI infrastructure race: extraordinary demand is producing extraordinary growth, but supplying that demand requires extraordinary capital.

The neocloud model

Companies such as Nscale, CoreWeave, Nebius, Crusoe and Lambda have emerged as alternatives to traditional hyperscale clouds for AI workloads.

Their pitch centers on access to high-end accelerators, specialized networking, data-center capacity and power. For AI laboratories, securing compute can be as strategically important as hiring researchers.

Nscale says it operates across 14 regions and has a power pipeline exceeding 10 gigawatts. Reuters reported that its total contracted value has grown beyond $103 billion.

Concentration remains a risk

Rapid infrastructure expansion often comes with customer concentration. According to the filing, Nscale’s largest customer represented 52% of first-half revenue.

Large contracts can accelerate growth, but dependence on a small number of customers creates risk if spending plans change.

For AI infrastructure providers, the challenge is not only building capacity. It is converting giant contracts into diversified, durable cash flow.

Why the IPO matters

A successful listing could become an important signal for the broader AI infrastructure market. Investors have already funded enormous data-center projects, chip purchases and power agreements.

Public markets will now be asked to value those commitments against losses, debt requirements and long construction timelines.

What to watch

CodaDaily will be watching the final valuation, customer concentration disclosures, capital expenditure plans and how investors respond to the company’s path toward profitability.

The filing also reinforces a broader reality: AI competition is increasingly constrained by physical infrastructure.

Models may be software, but the AI economy is being built with chips, cables, buildings and electricity.