NextFin News - DayOne Data Centers is seeking a loan to fund its Hong Kong data center operations ahead of a planned US initial public offering, a move that underscores how the AI-driven data center boom is forcing even the best-capitalized private infrastructure platforms to return to lenders before facing public-market scrutiny. The Singapore-headquartered company - once the international arm of China's GDS Holdings - has confidentially filed for a US listing and is considering raising roughly $5 billion at a valuation of about $20 billion, people familiar with the matter have said. The fresh borrowing, focused on Hong Kong, arrives only weeks after DayOne closed a $4.5 billion equity round and follows its pursuit earlier this year of what would be the largest loan for the sector by any firm in Asia.
The company declined to comment on the financing.
The Sequence: Equity, Then Debt, Then an IPO
The financing push has come in rapid succession. In January, DayOne raised more than $2 billion in a Series C tranche led by Coatue Management at a valuation of about $10 billion. On June 5, it announced the final closing of the round at $4.5 billion in total gross proceeds, saying the capital would accelerate expansion across Singapore, Malaysia, Indonesia, Thailand, Japan, Hong Kong, Finland and Spain. Since its founding in 2022, the company said it has secured more than 1.5 gigawatts of total bookings for capacity across Asia Pacific and Europe.
Then, in March, DayOne sought to double an existing loan facility to as much as $7 billion, up from a $3.4 billion-equivalent original, in what would be the largest borrowing for the data center sector by any company in Asia. That facility was tied to expanding operations in Malaysia, structured as an amend-and-extend with tranches in US dollars and Malaysian ringgit maturing in 2030. The new Hong Kong-focused loan reported this month is a separate borrowing - and its size has not been disclosed.
On August 11, the company was reported to have confidentially filed for its US initial public offering, with a listing possible as soon as the next quarter. A dual listing on Nasdaq and the Singapore Exchange has been under consideration, though the Singapore component has been described as less concrete than the US listing. Bank of America, Citigroup, JPMorgan Chase and Morgan Stanley have been named in connection with the IPO process.
The pattern is the story: DayOne is layering equity, debt, and an IPO in quick succession - a capital-raising sprint that reflects both the opportunity and the pressure of the AI infrastructure moment. The $4.5 billion equity raise and a potential $7 billion loan together imply more than $11 billion of capital being marshaled around a company that, as recently as December 2024, was valued at roughly $1.3 billion after a $1.2 billion Series B round. That is a more than 15-fold increase in stated value in less than two years - a pace that public-market investors will examine line by line.
Why Debt Before an IPO: The Capital-Intensity Trap
The question public-market investors will ask is simple: why borrow at all after raising $4.5 billion? The answer lies in the economics of hyperscale data center development. A data center platform is not a software company - it is a capital-intensive industrial builder whose growth is gated by power procurement, construction timelines, and the ability to finance assets that will not generate cash flow for years.
Equity is expensive and dilutive, particularly when a company is trying to preserve a valuation narrative ahead of an IPO. Debt, by contrast, lets a developer accelerate capacity buildout without resetting its equity price - and, crucially, it lets the company demonstrate to public investors that it can service leverage with contracted revenue. Loans of this type are typically repaid from long-term leases to investment-grade tenants such as Microsoft, Alphabet's Google, or Amazon, providing steady cash flow to service the debt.
This is the transmission mechanism behind the headline: the loan is not a sign of distress, but a deliberate pre-IPO balance-sheet maneuver. By locking in financing now, DayOne can convert bookings into energized capacity faster, which in turn supports the valuation it hopes to sell to public investors. The risk is that leverage amplifies both sides of the cycle - if construction slips or tenants delay, debt service does not.
There is also a signaling dimension. A company that funds growth entirely with equity tells the market it cannot access debt markets on acceptable terms. A company that blends the two tells a different story: that lenders have underwritten its contracted cash flows and that the balance sheet can carry a load. For a pre-IPO developer, that signal matters as much as the capital itself.
In Hong Kong, the assets are already taking shape. DayOne's Kwai Chung cluster delivers 66 megawatts of carrier-neutral capacity across roughly 70,000 square meters of gross floor area, the company says. Its first self-built Hong Kong facility, HK1, provides 18.8MW across about 22,931 square meters and rises 22 stories - one of Asia's tallest purpose-built data centers. The company also operates or will soon operate 18 data centers within the Guangdong-Hong Kong-Macao Greater Bay Area.
The Hong Kong Constraint: Why This Market Needs Its Own Financing
Hong Kong is not just another line item in DayOne's expansion plan - it is one of the most capital-intensive and supply-constrained data center markets in Asia Pacific, which is why it warrants a dedicated financing facility rather than being folded into the Malaysia loan.
The city is land-scarce by definition. Tseung Kwan O alone accounts for nearly 41% of Hong Kong's existing white-floor data center space and approximately 25.7% of upcoming supply, according to market analysis. The Hong Kong data center market is projected to grow from $3.62 billion in 2025 to $5.81 billion by 2031, an 8.2% compound annual growth rate - but that growth is bottlenecked by two inputs that cannot be imported: land and power.
DayOne's Kwai Chung cluster is built on brownfield sites, converting former industrial and logistics facilities rather than competing for greenfield land. That is a strategic necessity as much as an environmental one: in a market where entitled, powered sites are the scarce asset, brownfield conversion is often the only viable path to scale. The trade-off is cost and complexity - retrofitting a cold storage warehouse into a high-density, AI-ready facility is more expensive per megawatt than building on a greenfield campus, but it buys access to a location that would otherwise be unavailable.
Power is the second constraint. High-density AI racks draw far more electricity per square meter than traditional colocation, and Hong Kong's grid, while reliable, has limited headroom in its densest urban districts. A developer that has already secured power capacity for a 66MW cluster holds an asset that new entrants cannot simply replicate - permitting and grid connection take years, not quarters. This is the structural scarcity argument in its purest form.
Cyclical or Structural: Two Forces at Work
The data center financing boom is being driven by two distinct forces that must be separated rather than blended. The demand side is structural: AI workloads require purpose-built, high-density facilities that cannot be repurposed from existing stock, and the power and land constraints in markets like Hong Kong create durable scarcity for sites that are already entitled and energized. That is a regime shift, not a cycle - the capacity gap will not self-correct because permitting and grid connection take years, not quarters.
The financing side, however, is cyclical. Lenders are currently eager to deploy capital into AI-adjacent infrastructure, and loan terms reflect that appetite. But the first cracks are visible: in July, Credit Agricole CIB sought to sell down about HK$150 million of a HK$1.6 billion loan it and other banks extended to ESR Group for a data center conversion in Kwai Chung - a move that highlighted banks reshuffling lending to keep exposure to the booming sector within limits. If credit conditions tighten before DayOne's assets are fully leased, the company would face the worst of both worlds - high leverage and expensive refinancing.
The evidence floor for the structural call: Hong Kong's data center market is projected to grow from $3.62 billion in 2025 to $5.81 billion by 2031, an 8.2% compound annual growth rate, with Tseung Kwan O alone accounting for nearly 41% of existing white-floor space. The evidence floor for the cyclical call: the sector's biggest loans - DayOne's $7 billion facility, SoftBank's $40 billion loan - were all struck within a narrow window of peak lender enthusiasm in 2025 and 2026.
Short term, the financing is cyclical and will revert when credit tightens. Long term, the scarcity of entitled, powered sites in Hong Kong is structural and will not revert on its own. DayOne is betting the structural leg outlasts the cyclical one.
The Peer Comparison: What Public Markets Already Know
DayOne's IPO will not be priced in a vacuum. Public data center operators already trade on a set of metrics that will become the benchmark for DayOne's valuation: funds from operations per share, same-store revenue growth, occupancy, and the ratio of enterprise value to earnings before interest, taxes, depreciation and amortization.
Equinix and Digital Realty, the two largest listed data center REITs, trade at premiums that reflect stabilized, contracted, geographically diversified cash flows - the profile of a mature owner-operator, not a development platform. GDS Holdings, DayOne's former parent and China's largest data center operator, has traded at a steep discount to those peers, reflecting China-specific regulatory and macro risk. DayOne's challenge is to convince investors it deserves an Equinix-like multiple on assets that are still being built, while carrying the execution risk that GDS's discount already prices in.
The valuation gap between those reference points is wide. A $20 billion valuation on $4.5 billion of equity raised in June implies that new money is being asked to accept a 100% premium to the January round's $10 billion mark in a matter of months - and then to fund an IPO at that level. Public investors will not simply accept the last private round as a floor; they will underwrite the assets, the leases, and the leverage.
The Counter-Thesis: Valuation Already Prices Perfection
The strongest argument against DayOne's strategy is that a $20 billion valuation already assumes flawless execution. At that price, the company is not being underwritten as a capital-intensive developer with execution risk - it is being priced as a stabilized infrastructure platform with contracted, predictable cash flows. The gap between those two identities is where public-market investors will find their margin of safety, or their loss.
DayOne is not a stabilized data-center REIT. It is a high-growth development platform exposed to power procurement, construction execution, customer concentration, financing costs, and cross-border operating complexity. Public investors will discount the valuation for that risk - which is precisely why the company is racing to convert bookings into operating assets before the IPO. Every megawatt that becomes revenue before listing narrows the discount.
The falsifying signal is quantifiable: if DayOne's IPO prices below $15 billion - a 25% discount to the reported $20 billion target - the market will have rejected the "flawless execution" narrative and re-rated the company as a developer rather than an infrastructure owner. Watch the final IPO pricing and the proportion of energized versus contracted capacity disclosed in the prospectus. That threshold is an analytical benchmark, not a reported figure; the point is that a discount of that magnitude would signal the market is pricing execution risk, not scarcity value.
What's Next: Three Time Horizons
Short term (next quarter): The Hong Kong loan closes or is restructured, and DayOne progresses its confidential SEC filing toward a public launch. Expect continued silence from the company - it declined to comment on the financing talks. The base case is a successful closing; the downside case is that banks, mindful of sector exposure limits shown by the ESR loan sale, demand tighter terms.
Medium term (6-18 months): The IPO lands, likely on Nasdaq, with a Singapore component still under consideration. Base case: a roughly $5 billion raise at a valuation between $15 billion and $20 billion. Upside case: AI capex momentum carries the print to the full $20 billion. Downside case: a wider market risk-off event or a widening in credit spreads forces a smaller raise at a lower valuation.
Long term (structural): The winners in Asian data center infrastructure will be those that secured power and land early - a scarce asset that DayOne's Kwai Chung cluster represents. If the structural demand thesis holds, today's leverage will look cheap in five years. If AI capex proves cyclical, the companies with the most debt will face the deepest repricing.
"HK1 is DayOne's first self-built and self-operated data center in Hong Kong. It marks a significant milestone in our plans to establish a high-performance data center cluster in the city."
William Huang, DayOne's Chairman and CEO, said in December 2023, when the company completed customer move-in at its first Hong Kong facility. The board has since been strengthened for the public-market transition: Nicolas Aguzin was appointed to the board in May 2026, with the company saying his experience leading complex international financial institutions would support DayOne as it deepens relationships with global stakeholders.
The bottom line: DayOne's loan is a bet that debt-funded speed today will buy a higher valuation tomorrow - and that public markets will reward converted capacity more generously than contracted promises. It is a wager on execution, and the margin for error is thinner than the balance sheet suggests.
Explore more exclusive insights at nextfin.ai.
