DayOne Data Centers, a Singapore-based builder of hyperscale computing campuses across Asia and Europe, has filed to go public in New York, in what bankers expect to be one of the largest tests yet of investor appetite for the artificial-intelligence infrastructure buildout.
The company registered a Form F-1 with the U.S. Securities and Exchange Commission this month and applied to list its American depositary shares on the Nasdaq Global Select Market under the ticker "DODC." The filing itself leaves the number of shares and the price range blank, but people briefed on the plans have told Bloomberg and Reuters that DayOne is seeking to raise as much as $5 billion at a valuation near $20 billion, according to The Star. A placeholder in Renaissance Capital's IPO tracker pegs the deal closer to $3 billion, a reminder that the final size is still being negotiated.
Revenue Triples, Losses Widen
The filing's financials capture both sides of the data-center story. Revenue for the six months ended June 30 jumped to $512 million, from $151.5 million a year earlier, as hyperscale and technology customers rushed to lock up capacity for AI training and inference. Net losses widened over the same period to $77.2 million, after the company lost $367.1 million in all of 2025, a figure padded by $341.8 million in stock-based compensation tied to its separation from its former parent.
DayOne reports 4.6 gigawatts of secured or in-development capacity across ten markets, including Malaysia, Thailand, Japan, Finland and Spain, with 962 megawatts already in service and about 2.3 gigawatts booked by customers. Malaysia alone generated 87% of revenue in the first half of this year, a concentration the filing itself flags as a risk should power supply, regulation or demand in that one market shift. The company was spun out of Shanghai-based GDS Holdings, which has kept a 35.6% stake in the renamed entity; other early backers include SoftBank's Vision Fund, Hillhouse Investment, Boyu Capital and Coatue Management.
DayOne traces its roots to 2022, when Shanghai-based GDS Holdings established an international arm in Singapore to chase data-center demand outside mainland China. That unit was rebranded as DayOne in January 2025 and has since raised three rounds of outside capital — Series A, B and C rounds — from the institutional and strategic investors now listed on its cap table, ahead of this month's public filing.
A Fall IPO Market Under Strain
Morgan Stanley, J.P. Morgan, BofA Securities, Citigroup and BNP Paribas are underwriting the offering, but they are bringing it to market at an awkward moment. Bond yields have climbed since the Federal Reserve's September rate increase, and several large offerings have been delayed or repriced this autumn as investors grow choosier about which slice of the AI buildout they are willing to fund with fresh equity. The filing also lands just days after jitters over AI-related financing rattled technology shares more broadly, a reminder that public investors are no longer treating every data-center or AI balance sheet the same way.
DayOne is not alone in testing that appetite. Rival operators including Vantage Data Centers and CyrusOne have each explored IPOs of their own this year, according to trade press, as the surge in AI spending collides with how much debt and equity markets are willing to absorb to build the physical infrastructure behind it. Much of that buildout has so far been financed with private credit and project-level debt rather than public equity, which is part of why DayOne's filing is being watched closely: a strong reception would signal that public shareholders, not just private lenders, are willing to underwrite the next phase of AI infrastructure spending.
No pricing date has been set. DayOne's roadshow is expected to unfold over the coming weeks, with the Federal Reserve's next policy meeting on Oct. 27-28 likely to influence the rate backdrop against which bankers price the deal. For an industry built on long-term contracts and short-term capital needs, the reception DayOne gets could set the terms for the next wave of data-center listings.
