Blackstone's AirTrunk Postpones $1.5 Billion REIT IPO: Yield Raised to 6% but No Takers, AI Infrastructure Financing Faces 5.2% US Treasury Yield

Within one week, three data center listing plans in the Asia-Pacific region were shelved one after another, exposing a deeper structural contradiction: the AI infrastructure expansion boom is built on assumptions of low interest rates in 2024, while current financing costs have returned to 2007 levels.

According to media reports, due to “rising interest rates and market volatility suppressing investor demand”, AirTrunk, a data center operator under Blackstone Group, has decided to postpone its Singapore Real Estate Investment Trust (REIT) listing to the first or second quarter of 2027, subject to market conditions. The original offering was expected to raise about $1.5 billion (approximately 2 billion SGD).

The impact of this delay is not limited to a single issuer. Last week, Australian data center company Firmus, backed by Nvidia, just withdrew its ASX listing plan, and Vitro, the data center business under Philippines telecom operator PLDT, also postponed its Philippine REIT IPO to 2027.

Goldman Sachs analysts forecast in their latest report that the Australia-New Zealand region will see a $50 billion to $65 billion AUD digital infrastructure ‘super cycle’ of capital expenditure, but at the same time, explicitly pointed out that financing is the real bottleneck restricting the sector’s development.

AirTrunk is the third data center listing plan in the Asia-Pacific region to be suspended or postponed within about a week.

Firmus withdrew its ASX listing application on October 8, having already been forced to cut its issue price by up to 25% from AUD 11 during the book-building process. Vitro attributed the delay of its Philippine REIT IPO to the “current market environment and rising interest rates.” Previously, SB Energy under SoftBank suspended its listing roadshow on September 22.

These three cases happened in rapid succession, reflecting a synchronized cooling of investor sentiment in the region, rather than isolated problems with individual projects.

Though the results are similar, the failure logic for the two deals is entirely different. Distinguishing between them helps to accurately understand the real root of the current market malaise.

The problem for Firmus is essentially a valuation issue. The company asked public market investors to subscribe at a valuation of about AUD 43.7 billion (about $30.4 billion), nearly triple its August private fundraising valuation of $10.5 billion, while citing unconventional metrics like “EV+1/EBIT+2” to bolster its pricing.

Australian local fund managers were generally cautious, with Jun Bei Liu of Ten Cap publicly stating, “My view is that (the issue valuation) should be below AUD 10 billion.”

AirTrunk’s case is entirely different. It is a large-scale operating platform spanning Australia, Japan, Singapore, Malaysia, Hong Kong, and India. In 2024 it was acquired by Blackstone Group and CPP Investments for AUD 24 billion, making it the largest recorded data center deal in the world at the time.

Its REIT was positioned as a “conservative path” to invest in AI infrastructure—assets already in a stable operation phase, tenants are large-scale cloud computing companies, and dividend yields are robust.

However, AirTrunk stumbled on interest rates. According to Bloomberg, AirTrunk had already considered raising its yield from the original 5.5% to about 6% prior to the issue to attract investors, but was still unsuccessful.

On Tuesday, the U.S. 10-year Treasury yield closed at 5.22%. At the “optimized” yield of 6%, AirTrunk’s REIT offered only about 78 basis points of excess return—for a single-industry, illiquid instrument with currency risk, tenant concentration risk, and construction risk, investors judged this compensation as far from sufficient.

AirTrunk had also planned a SGD 2 billion accompanying loan for asset acquisition and debt refinancing, which Bloomberg called a “crucial test for Asian data center financing, as local banks have neared their industry exposure caps.” This test failed as well.

Despite differing mechanisms, both deals sent the same message to the market: investors want more attractive pricing to compensate for the risks undertaken.

For Firmus, this meant a lower valuation; for AirTrunk, it meant a higher yield—though mathematically, a higher yield also means a lower asset valuation for Blackstone’s holdings.

The bond market is sending the same warning. Data center debt credit spreads continue to widen. Coreweave’s unsecured bond spread broke 1,000 basis points for the first time, while Oracle’s unsecured bonds are trading at deep junk status.

On October 13, Broadcom CEO Hock Tan warned at a conference hosted by DBS in Singapore that financing may now become the chief bottleneck for the industry, replacing component shortages. J.P. Morgan technology analyst Joshua Meyers called this comment the “most direct reason” for the tech sector’s decline that day.

Market setbacks don’t mean underlying demand has disappeared. According to Goldman Sachs’ sales team in their latest research on Australasian digital infrastructure, accelerated AI adoption will lead to a 0.7GW supply gap by 2030 in Australia and New Zealand, triggering a AUD 50-65 billion capex super cycle.

Yet Goldman Sachs’ neutral rating for listed company NEXTDC (NXT AU) comes with telling justification—financing itself is a constraint. Goldman’s analysts warn NEXTDC faces near-term margin pressure during large-scale cloud tenant migration, and a financing gap of over AUD 1 billion.

This contradiction points to the core: tens of billions of AUD in capex needs funding, yet the two most visible public funding channels—growth IPOs on ASX and yield-focused REITs on SGX—both clogged up within the same week. The alternatives point to private markets: private credit, insurance capital, and special purpose vehicles, but at the cost of higher financing expenses.

Goldman Sachs economists were blunt in their Global Economics report released this week: “Global interest rates have reached their highest level since 2007.” Yield products priced on 2024 rate assumptions will face inherent disadvantages in the 2026 rate environment.

AirTrunk’s postponement is a greater warning than Firmus. Firmus can be attributed to an issuer’s self-inflated valuation; but AirTrunk represents the most credible version of this transaction logic—real assets, real tenants, Blackstone as backer, and pro-actively increased yields.

If such a project cannot be issued successfully at 6% yield when 10-year Treasuries are at 5.2%, the problem is no longer an isolated misjudgment by an aggressive issuer but a structural mismatch in capital cost for the entire AI construction wave.

This boom began based on rate assumptions for 2024, yet is now seeking refinancing at 2007-level costs. Sooner or later, the difference will be borne by someone. For now, the public market’s stance is clear—that person won’t be them.

Internal discussions on AirTrunk’s listing size and timeline are still underway. The next notable milestone is Firmus’ partner DayOne in Batam, Indonesia, which reportedly plans to seek up to $5 billion in the U.S. market as early as November.