Ai / Business | DayOne is considering expanding its Series C funding round to more than $4 billion, according to Bloomberg. The move shows how AI investment is shifting from model labs toward data centers, power access and the physical infrastructure needed to support compute demand.
DayOne is weighing a Series C expansion to more than $4 billion, a sign that AI money is moving fast toward the physical infrastructure behind the models.
DayOne Data Centers is no longer just raising growth capital. It is becoming a marker for where the AI boom is sending serious money next: into land, power, cooling, utility relationships and the project finance needed to turn demand for compute into real capacity.
As Bloomberg reported on May 14, the Singapore headquartered hyperscale data center platform is considering upsizing its Series C funding round to more than $4 billion, roughly twice the size of the closing it announced in January. The potential increase is said to be led mostly by existing backers, which matters because follow-on capital of this size usually says more than a new logo on a cap table. It suggests investors who already know the business want more exposure before the next phase is priced in.
DayOne’s own January announcement gives the scale of the move. The company said it had signed definitive agreements for more than $2.0 billion in Series C equity financing, led by Coatue and supported by institutions including the Indonesia Investment Authority. That financing came after DayOne raised an aggregate $1.9 billion across its Series A and Series B rounds in 2024, then secured a mezzanine facility of €500 million that can expand to €1 billion from Brookfield and a global sovereign investor in December 2025.
For the last two years, most of the attention in AI capital markets has gone to model labs, chipmakers and cloud software companies. That made sense. OpenAI, Anthropic, Nvidia and their peers were the most visible part of the story. But the next constraint is less glamorous and more difficult to solve. You need grid access, permitted land, specialized cooling, construction capacity and customers willing to commit years ahead of delivery.
That is why DayOne is interesting beyond the headline number. The company says the January financing supports expansion across Asia-Pacific and Europe, including Finland campuses in Lahti and Kouvola, the SIJORI region covering Singapore, Johor and the Riau Islands, and markets such as Thailand, Japan and Hong Kong. It also says the capital will help deliver against secured customer commitments of about 1GW. In data center terms, that is not a branding claim. It is a large operating obligation.
AI-ready capacity is different from ordinary server space. High density racks create heat that older facilities were not designed to handle. Liquid cooling and low carbon power pathways are becoming commercial requirements, not nice additions. DayOne has leaned into that language, describing its portfolio around high-density, liquid-cooling enabled designs, renewable and low carbon power options and faster prefabricated deployment models.
The practical point for founders and investors is simple. AI is turning infrastructure timing into a competitive advantage. A model company that cannot secure enough compute may lose speed. A cloud provider that cannot lock in power may lose customers. A data center operator that can deliver capacity in the right market at the right time suddenly looks less like real estate and more like a strategic bottleneck.
Private capital may be delaying the public market moment
Bloomberg also reported that DayOne had been close to filing confidentially for a US listing in March, but is now focused on completing the upsized Series C before deciding on an IPO. That is a useful signal. When private investors are willing to double a round, the public market becomes less urgent.
There is a tradeoff. Staying private gives DayOne more room to fund construction without exposing every quarter of spending to public market judgment. It also lets existing investors defend their positions before a possible listing. But it can delay price discovery for the broader market, especially when infrastructure assets are being valued partly on future AI demand that is still hard to forecast cleanly.
This is where venture starts to look more like infrastructure finance. The old startup rhythm was product, users, revenue, scale. Data centers need a different rhythm: site control, grid connection, anchor customers, debt capacity, construction execution and long-term utilization. The returns can be meaningful, but the risks sit in the physical world. A delayed power connection is not a software bug. A permitting problem cannot be patched over a weekend.
DayOne is also operating in regions where demand is strong but local constraints matter. Southeast Asia has become a major target for cloud and AI infrastructure, helped by connectivity, enterprise growth and government interest in digital investment. Europe offers large customer demand, but also stricter energy, sustainability and community expectations. Finland’s appeal comes partly from climate, power availability and the ability to reuse heat, but even there projects still need approvals, grid planning and local acceptance.
The bigger implication is that AI infrastructure may become one of venture’s new asset classes, sitting between growth equity, private credit and traditional real assets. That will change who wins deals. The best investors will not only ask whether demand for AI compute is rising. They will ask who can actually build, finance and operate the capacity that demand requires.
For now, DayOne’s possible $4 billion Series C is still under consideration, not a completed transaction. But the direction is clear. The AI market is moving from promises about intelligence to commitments around capacity, and the next winners may be the companies that can turn megawatts into usable compute faster than everyone else.
Also read: HMRC is putting British AI at the center of tax enforcement • India is becoming the first real test of AI job disruption • AI is speeding up Linux flaw discovery as Fragnesia hits servers