September 25, 2026, (Inside AI) — London-based AI cloud provider Nscale has filed for a US initial public offering, seeking a valuation of about $30 billion. The move comes just six months after the company was valued at $14.6 billion in a private funding round. But the IPO filing reveals a stark gap between ambition and execution: Nscale plans to deploy 461,000 AI chips, yet only 5% are currently operational. To justify its target valuation, the company must complete 12 new server farms on schedule. With so much unproven, the decision to invite public market scrutiny now is unusual.
The filing, submitted on September 18, shows Nscale posted a net loss of $1.02 billion on revenue of $140.6 million for the six months ending June 30. A year earlier, the loss was $368.9 million on revenue of $10.4 million. The widening loss reflects heavy capital expenditure on data center infrastructure, but the revenue base remains small relative to the valuation sought. Public investors will need to believe that Nscale can convert its chip deployment pipeline into paying customers quickly.
The company’s pitch rests on surging demand for AI compute. Hyperscalers like Microsoft, Amazon, and Google have spent billions on data centers, and Nscale aims to serve smaller AI firms that cannot build their own. However, the AI infrastructure market is increasingly crowded. Competitors such as CoreWeave and Lambda Labs have also raised significant capital and are racing to deploy GPUs. Nscale’s differentiation is unclear from the filing, which does not disclose customer names or contract terms.
The IPO timing is particularly curious. Public markets have recently shown fatigue with money-losing tech listings. In 2025, several AI-related IPOs traded below their offer prices within months. Nscale’s $30 billion valuation implies a multiple of over 200 times annualized revenue, based on the latest six-month figure. That is far above established cloud providers like Equinix or Digital Realty, which trade at single-digit revenue multiples. The comparison underscores how much growth investors must price in.
Read: Nebius raises AI cloud prices again as demand for computing power soars
Nscale’s filing also reveals operational risks. The 12 new server farms are planned across multiple countries, including the UK and US. Each requires land, power, and cooling, plus regulatory approvals. Delays are common in data center construction. A missed deadline could cascade into revenue shortfalls and erode investor confidence. The company has not disclosed contingency plans or penalties for delays.
Moreover, the AI chip supply chain remains constrained. Nvidia, the dominant GPU maker, has long lead times for its latest chips. Nscale’s ability to procure 461,000 chips depends on Nvidia’s allocation and competing orders from larger buyers. If Nscale cannot secure chips on time, its deployment schedule slips further. The filing does not specify which chips it plans to use or whether it has binding supply agreements.
Critics argue that Nscale’s IPO is a test of investor tolerance for AI hype. The company’s $1.02 billion loss in just six months is larger than its cumulative revenue since inception. While revenue grew 13-fold year-over-year, the absolute number is tiny. To reach profitability, Nscale would need to scale revenue exponentially while controlling costs. That is a tall order in a capital-intensive industry.
Supporters counter that early-stage infrastructure plays often show losses before reaping scale benefits. Amazon Web Services lost money for years before becoming a profit engine. If Nscale can lock in long-term contracts with AI developers, its losses could narrow. But the filing does not disclose any such contracts, leaving investors in the dark.
The IPO also raises questions about corporate governance. Nscale’s largest shareholder is Blue Owl Capital, a private equity firm that led its March funding round. After the IPO, Blue Owl will retain a significant stake, potentially influencing strategic decisions. Public shareholders may have limited say. The filing does not detail dual-class share structures or other protections.
Market conditions add another layer of uncertainty. Interest rates remain elevated, making future cash flows less valuable today. High-growth tech stocks have underperformed the broader market in 2026. An IPO of a loss-making AI infrastructure firm could struggle to attract institutional buyers. Retail investors, however, might be drawn by the AI narrative.
Nscale’s filing is a litmus test for the AI investment cycle. If the IPO succeeds at $30 billion, it signals that public markets still have appetite for speculative AI bets. If it prices lower or trades down, it could chill further listings. The outcome will be watched closely by venture capitalists and startup founders alike.
The company has not announced a roadshow date or lead underwriters. Sources indicate that Goldman Sachs and Morgan Stanley are among the banks involved, but this could not be independently verified. Nscale declined to comment beyond its public filing.
For now, Nscale’s future hinges on execution. Completing 12 server farms and deploying 461,000 chips on time is a massive logistical challenge. The company’s ability to do so will determine whether its $30 billion valuation is visionary or delusional. Public investors will soon render their verdict.
