The AI data-centre boom is creating enormous demand for infrastructure — but there is another question investors need to ask: who will finance all that infrastructure, and at what valuation?
Blackstone-backed AirTrunk is reportedly preparing a potential Singapore-listed REIT, alongside plans to raise about S$2 billion of debt financing. If the transaction proceeds, it could become much more than another data-centre REIT listing.
It could provide a new public-market benchmark for what institutional-quality AI infrastructure is actually worth.
In this video, we examine why the potential AirTrunk listing matters to Singapore investors and why it could be particularly relevant to existing data-centre REITs such as Keppel DC REIT.
The key issue is the relationship between AI demand and the cost of capital.
AI may drive stronger data-centre demand, higher utilisation, rental growth and development opportunities. But data centres are also highly capital-intensive assets that require large amounts of debt and equity. Higher borrowing costs, refinancing requirements and rising required REIT yields can therefore reduce how much of that operational growth ultimately reaches shareholders.
That makes AirTrunk’s eventual IPO valuation and initial yield particularly important.
A low yield could indicate strong public-market appetite for high-quality hyperscale assets. A higher required yield could signal that investors are demanding greater compensation for leverage, tenant concentration, interest-rate and development risks.
For Keppel DC REIT investors, AirTrunk could therefore serve as a benchmark rather than a simple buy-or-sell signal. The two portfolios may differ in geography, lease structures, tenant concentration, gearing, development exposure and funding costs, so the comparison needs to be based on the underlying numbers rather than simply the “data-centre REIT” label.
The video also examines the broader capital-recycling thesis. If private data-centre owners can transfer mature assets into listed REITs at attractive valuations, they could potentially recycle capital into new data centres, power capacity and other AI infrastructure. But if public markets demand much higher yields, that model becomes less attractive.
When the AirTrunk transaction documents eventually become available, investors should focus on metrics including IPO valuation, initial yield, gearing, WALE, lease expiries, rental escalations, rental reversion, tenant concentration, cost of debt and development requirements.
The deeper lesson is that AI demand does not automatically equal AI investment returns.
The infrastructure may be needed. The rents may rise. The assets may become more valuable. But shareholders ultimately own the cash flow left after financing costs and capital requirements.
Watch the full video below for the detailed analysis of AirTrunk’s potential REIT listing, what its valuation could tell us about AI infrastructure, and what Singapore investors should watch for in Keppel DC REIT and the wider data-centre REIT sector.