AI data center company Nscale plans to go public in the U.S. as early as September, aiming to raise up to $3 billion.

The UK artificial intelligence infrastructure company Nscale is planning to raise up to $3 billion through an initial public offering (IPO) in the US, joining the wave of AI data center companies intensifying their presence in the capital market. According to knowledgeable sources, the company could launch its listing as early as September this year and is currently collaborating with Goldman Sachs Group, Inc. and JPMorgan Chase & Co. to advance its plans. Headquartered in London, Nscale has been aggressively expanding its AI data center business in recent years and is currently building data centers in regions such as Norway and West Virginia, USA. With the rapid increase in demand for AI computing power from global technology companies, the company aims to raise more funds through the capital market to support its large-scale infrastructure expansion plans. Sources previously disclosed that Nscale has informed potential investors that its total contract revenue before the IPO amounts to approximately $51 billion, providing significant business support for its IPO. However, discussions regarding the IPO are still ongoing, and the specific financing scale and listing time may change, with a potential delay in the IPO also possible. Nscale declined to comment on this matter. Under the currently discussed plan, Nscale's US IPO could raise about $3 billion at most. If successfully completed, it would become one of the larger public financing projects in the AI infrastructure sector in recent times. The composition of Nscale's board of directors is also of considerable interest to the market, including former Meta Platforms (META.US) Chief Operating Officer Sheryl Sandberg and former Meta Global Affairs President and former UK Deputy Prime Minister Nick Clegg. Nscale's business model primarily revolves around building large AI data centers and procuring a significant amount of AI chips and other infrastructure equipment, then leasing the computing power to customers who need to train and run AI models. With the rapid development of generative AI, the demand for GPUs, data centers, and electrical resources from global technology companies has surged dramatically, prompting AI infrastructure companies like Nscale to invest billions of dollars to expand their computing capabilities. Sources previously revealed that the company plans to add around 10 gigawatts (GW) of power capacity to its AI computing centers. Prior to this, Nscale already had about 831 megawatts (MW) of operational or contracted power capacity. As a point of reference, 1 gigawatt of power can typically meet the electricity needs of about 750,000 American households at any given time. This means that if Nscale achieves its goal of adding 10 gigawatts, its AI infrastructure size will see significant expansion. The enormous power demand also reflects the capital-intensive nature of current AI infrastructure construction. Companies not only need to purchase a large number of AI chips but must also invest substantial funds in building data centers and securing sufficient power supply and related infrastructure resources. In addition to expanding the scale of its data centers, Nscale is also extending into the AI software and computing power management sectors. In July of this year, the company agreed to acquire the software startup Anyscale for $1.65 billion, hoping to leverage its technology to help customers use AI computing resources more efficiently. By combining data center infrastructure with software platforms, Nscale aims to provide not only computing power to its clients but also enhance the efficiency of AI computing resource utilization, thereby broadening its business coverage in the AI infrastructure industry chain. As Nscale prepares for its IPO, AI infrastructure companies are igniting a wave of financing frenzy in the capital markets. In May of this year, data center acquisition platform Blackstone Digital Infrastructure Trust raised $2 billion through its IPO; Brookfield-backed Csquare raised $1.21 billion in July after going public. At the same time, data center operator Switch has secretly submitted an application for an IPO, potentially going public as early as November this year. Previous reports indicated that the company was seeking a valuation close to $50 billion in one round of financing, including debt. Behind this series of financing activities is the rapid expansion of capital demand for global AI infrastructure construction. As large AI models require ever more computing resources, data center operators must invest huge amounts of money to purchase chips, build facilities, and secure sufficient power supply, prompting more and more companies to turn to equity and debt markets for financing.