Global data center operator DayOne is preparing what could become one of the most closely watched infrastructure listings of the artificial intelligence cycle.

The company, affiliated with China’s GDS Holdings, is planning a dual listing in Singapore and the United States that could raise about $5 billion, Reuters reported Sunday, citing the Financial Times. The report said the transaction could value DayOne at roughly $20 billion, placing the company at the center of the market’s growing appetite for data-center assets tied to AI demand. Reuters said it could not immediately confirm the Financial Times report, and that DayOne did not immediately respond to a request for comment.

If the listing moves ahead, the U.S. portion would give American investors another way to buy into the physical backbone of the AI economy. While public attention often goes to model developers, chipmakers and software companies, the AI boom is also creating a separate race for buildings, power, cooling systems, fiber connections and long-term data-center capacity.

That is what makes the DayOne plan important. It is not just another IPO story. It is a test of whether investors are still willing to pay up for the infrastructure underneath artificial intelligence.

DayOne has already been linked to a major U.S. public-market push this year. In February, Reuters reported that the company was aiming to raise as much as $5 billion in a U.S. initial public offering and could seek a valuation of around $20 billion, citing a person familiar with the plan.

The latest report adds a second layer: Singapore. A dual listing would allow the company to tap U.S. capital markets while also giving Singapore a larger role in a sector that has become strategically important across Asia. Data centers are no longer just real estate assets. They are part of the supply chain for cloud computing, AI services, enterprise software and national digital infrastructure.

For U.S. investors, the timing is notable. Public markets have already shown renewed interest in data-center exposure. Blackstone Digital Infrastructure Trust raised $1.75 billion in a U.S. IPO this month, giving investors a new publicly traded vehicle focused on newly built data-center assets leased to hyperscale tenants. Reuters reported that the vehicle had reviewed about $25 billion in immediate acquisition opportunities across major U.S. markets including Northern Virginia, Ohio, Phoenix, Maryland and Austin. (Reuters)

That matters because the data-center trade has moved beyond a narrow technology story. It now touches real estate, energy, private equity, public markets and industrial policy. Large AI models require huge amounts of computing capacity. That capacity requires buildings. Those buildings require electricity, grid connections, land, cooling and long-term financing.

The result is a capital-intensive race. Companies that can secure sites, power and customers may become essential infrastructure providers for the next phase of AI deployment. Companies that cannot may be left behind, even if demand for AI services remains high.

But the market is also becoming harder to underwrite. Investors are being asked to value companies and vehicles built around future demand, long construction timelines and expensive energy needs. The question is no longer whether AI requires data centers. It does. The question is whether the economics of those data centers will justify the valuations now being discussed.

The U.S. power backdrop shows why the issue is bigger than listing enthusiasm. The U.S. Energy Information Administration expects power demand to rise from a record 4,195 billion kilowatt-hours in 2025 to 4,244 billion kilowatt-hours in 2026 and 4,381 billion kilowatt-hours in 2027, Reuters reported in April. Rising demand has been tied in part to data centers, AI and wider electrification.

That surge is turning electricity into one of the central constraints on AI growth. A company can raise capital, sign customers and announce expansion plans, but if power cannot be secured quickly and affordably, capacity can become delayed or more expensive than expected.

Reuters has also reported that incentives designed to accelerate grid upgrades are emerging as data-center demand raises concerns about power bills for households and businesses. That creates a political and regulatory risk for data-center operators: the same projects investors view as AI infrastructure can be viewed by communities as drivers of higher local electricity costs.

This is where the DayOne listing story becomes more than a capital-markets headline. It reflects the shift from AI software excitement to AI infrastructure economics. The companies winning in this next phase may not only be the ones with the best algorithms or the most advanced chips. They may be the companies that can finance, build and operate the facilities where AI actually runs.

A U.S. listing would also bring more scrutiny. Public-market investors will want to understand DayOne’s customer base, geographic exposure, power strategy, debt profile, expansion pipeline and relationship with GDS Holdings. They will also look at how the company handles geopolitical risk, especially given its connection to a China-linked data-center group and its potential attempt to raise capital in the United States.

For Singapore, the plan could support a broader ambition to become a stronger capital-markets hub for technology infrastructure. If a dual listing succeeds, it could encourage other Asia-based infrastructure or AI-linked companies to consider similar routes.

For the United States, the listing would add to a public-market theme already visible in 2026: AI infrastructure is becoming investable in more forms. Investors can already buy chipmakers, cloud platforms, utilities, industrial suppliers and data-center landlords. A company like DayOne would add another route into the same theme, focused on hyperscale data-center operations across international markets.

The opportunity is large, but it is not simple. The AI infrastructure boom is forcing investors to evaluate a long chain of dependencies. Data-center demand depends on cloud and AI usage. Cloud and AI usage depend on enterprise adoption, consumer demand and model performance. Data-center profitability depends on construction costs, power availability, customer contracts and financing conditions.

That means the best investors will not treat every AI infrastructure listing as automatically attractive. They will ask harder questions: Is demand contracted or speculative? Are power costs locked in or floating? Are customers investment-grade? Are facilities concentrated in high-demand markets? How exposed is the company to regulatory or geopolitical pressure? How much capital will be needed after the IPO?

Those questions will likely follow DayOne if it moves toward a public filing.

Still, the direction of the market is clear. AI is pulling capital into physical infrastructure. The early public-market winners may be the companies that convince investors they can turn AI demand into durable, contracted cash flows rather than one-time hype.

DayOne’s reported dual listing would be a major test of that thesis.

If investors embrace it, the message will be that the AI infrastructure trade still has room to run. If they hesitate, it may show that public markets are becoming more selective after months of heavy spending announcements and rising concerns about energy constraints.

For now, the planned listing gives the market another signal: the AI boom is no longer just about models and chips. It is increasingly about who can build the places where the models live.