London-based AI infrastructure provider Nscale is preparing for a potential initial public offering in the United States, aiming to raise as much as $3 billion, according to sources familiar with the matter. The company is reportedly targeting a launch as early as September and is collaborating with Goldman Sachs and JPMorgan on the planned listing.
Nscale, headquartered in the UK, has been aggressively expanding its AI data center operations, with current construction projects underway in Norway and West Virginia, among other regions. As global demand for AI computing power surges, the company is seeking to tap public markets to fund its large-scale infrastructure buildout, which includes extensive investments in chips, facilities, and power resources.
Before the IPO, Nscale has communicated to potential investors that its contracted revenue backlog totals approximately $51 billion, providing a substantial foundation for the offering. However, discussions are still ongoing, and both the final capital raised and the timeline remain subject to change, with a potential delay possible. Nscale has declined to comment on the matter.
If completed at the discussed scale, the IPO would rank among the larger recent fundraising events in the AI infrastructure sector. The company's board has also drawn attention, featuring notable figures such as Sheryl Sandberg, former COO of Meta Platforms (META.US), and Nick Clegg, former president of global affairs at Meta and former UK deputy prime minister.
Nscale's business model centers on building large-scale AI data centers, procuring substantial volumes of AI chips and related infrastructure, and leasing computing capacity to clients training and running artificial intelligence models. With the rapid advancement of generative AI, tech enterprises worldwide have seen a sharp spike in demand for GPUs, data centers, and electricity, prompting Nscale and similar firms to deploy tens of billions of dollars in expanding computational capabilities.
Plans include adding roughly 10 gigawatts (GW) of new power capacity for its AI computing centers, on top of approximately 831 megawatts (MW) already operational or contracted. For context, one gigawatt of power can supply around 750,000 US homes at any given time, underscoring the scale of the proposed expansion if the 10 GW target is achieved.
This immense power requirement highlights the capital-intensive nature of current AI infrastructure development. Companies must not only acquire costly AI chips but also invest heavily in data center construction and secure sufficient electricity and related resources.
Beyond scaling up data centers, Nscale is also branching into AI software and compute management. In July, the company agreed to acquire software startup Anyscale for $1.65 billion, leveraging its technology to help clients use AI computing resources more efficiently. By integrating infrastructure with a software platform, Nscale aims to deliver not just raw compute power but also improved utilization efficiency, broadening its footprint across the AI infrastructure value chain.
As Nscale gears up for its listing, a wave of capital market activity is sweeping through the AI infrastructure sector. In May, Blackstone Digital Infrastructure Trust raised $2 billion via an IPO, while Brookfield-backed Csquare went public in July, securing $1.21 billion. Meanwhile, data center operator Switch has filed confidentially for a listing, potentially hitting the market as early as November, with earlier reports suggesting a valuation near $50 billion including debt.
These fundraising efforts reflect the rapidly escalating capital needs of global AI infrastructure development. As large AI models demand ever-greater computational resources, data center operators must commit vast sums to chip purchases, facility construction, and power procurement, driving more companies toward equity and debt markets for financing.
Comments