STT GDC Is Raising S$3.1 Billion Against Its Data Centres: the 3 Things I Changed in Our Infrastructure Hiring Plan

Two years ago I added three reliability engineers to a plan because a large operator announced a capacity expansion and I assumed the local market was about to tighten. It did not tighten. The capacity went to Johor, the hiring went with it, and I had three requisitions priced for a competition that never arrived in Singapore. That mistake is the reason I read Monday’s financing story differently from how I would have read it in 2024.

What Was Reported on 2 November

Bloomberg reported on 2 November 2026 that STT GDC is in talks with banks for a syndicated loan of roughly S$3.1 billion, about US$2.4 billion. The reported structure matters more than the headline figure: a three-year facility at an interest margin of 165 basis points over the Singapore Overnight Rate Average, with an extension option of up to two additional years.

The lender group is the four names you would expect in a Singapore syndication of this size. Standard Chartered, United Overseas Bank, Oversea-Chinese Banking Corporation and DBS Bank are described as global coordinators, mandated lead arrangers and bookrunners, with proceeds going towards financing the company’s assets. Two caveats are load-bearing: the reporting rests on unnamed sources, and the facility is not described as finalised.

Context explains the timing. The borrowing follows the September completion of the acquisition of STT GDC by a consortium led by KKR and Singapore Telecommunications, in a deal valued at S$6.6 billion, which DBS, OCBC and UOB had themselves helped finance through a separate package. Coverage of the fundraising places it inside a much larger pattern: Bloomberg-compiled data put AI-related debt issuance above US$650 billion since the start of 2025, and Moody’s has projected data centre investment of at least US$3 trillion over the next five years.

Why This Is Not the Hiring Signal It Looks Like

The instinct is to treat a multi-billion financing by a Singapore-headquartered operator as evidence that Singapore infrastructure hiring is about to get harder. That was my instinct in 2024 and it cost me three mispriced requisitions.

Two things break the inference. The first is what the money is for. A syndicated facility raised against existing assets finances capital expenditure and refinancing, not payroll. Data centre capex is overwhelmingly land, power, shell, cooling and equipment; the engineering headcount attached to a given megawatt is small and it arrives late, during commissioning and operations, not at the point of financing.

The second is geography, and it is the one that caught me. STT GDC operates across roughly a dozen countries, including India, South Korea, Japan, Malaysia, the United Kingdom, Germany and Italy. Where new capacity physically lands is decided by power availability, land and grid connection timelines rather than by where the treasury function sits. Singapore has been a constrained market for new capacity for years, which is exactly why so much regional growth has gone to Johor and beyond. We wrote about that dynamic when the same operator committed to a S$1.37 billion Johor data centre in August, and the lesson held: a Singapore borrower is not a Singapore employer.

So the correct question is not whether the loan is large. It is whether any of the capacity it funds lands somewhere that competes for engineers you are trying to hire. The reporting does not answer that, and nobody should pretend otherwise.

The 3 Things I Changed on Monday

1. Reliability and platform roles moved out of the product benchmark basket. We had been pricing site reliability and platform engineers against product companies, because that is who we thought we were competing with. The realistic competing bidder for those profiles is an infrastructure operator or one of its contractors, and they pay on a different logic: steadier, less equity, more shift and on-call structure. I kept our numbers and added a column for what the operator offers that we do not. Our notes on structuring developer compensation in Singapore cover rebuilding a basket without importing somebody else’s assumptions.

2. I checked which of our own vendor contracts float. This is the second-order effect nobody puts in a hiring plan and it is the one most likely to hit a budget. If financing costs across the sector are rising, some of that gets passed through in colocation, bandwidth and managed service pricing at renewal. I asked our finance lead for the renewal dates and pricing mechanics on our three largest infrastructure contracts. Two are fixed through 2027, one reprices in April. That single answer is worth more to next year’s engineering budget than any view I hold about AI debt.

3. I wrote down what would actually constitute evidence. Rather than acting on the report, I listed the three observable things that would tell me the local market is tightening: local postings appearing at the operator or its named integrators, contract rates for Singapore-based reliability and data centre engineers moving, and our own counteroffer rate on platform roles rising. None had moved as of Monday. If they do, I have a prepared response. If they do not, I have spent nothing, which is the outcome I failed to achieve in 2024.

The Figure I Will Not Plan Against

The US$650 billion, and its larger cousin, the US$3 trillion. Not because either is implausible as an aggregate, but because neither tells a Singapore employer anything actionable, and both are extraordinarily good at making a hiring manager feel informed.

Consider what the aggregate conceals. A global debt total spans jurisdictions with completely different power markets, labour markets and construction timelines. A projection of trillions in data centre investment over five years includes land, power infrastructure, cooling and chips, with the engineering labour component a small fraction of the whole and distributed wherever the capacity physically goes. Reasoning from a global total to your local shortlist requires at least three assumptions that nobody states: how much is Asia-Pacific, how much of that is Singapore rather than Johor or Jakarta, and what share of the spend is engineering payroll rather than concrete and transformers.

There is a sharper reason for restraint here. The facility itself is reported as still in discussion, sourced to unnamed people. I have watched syndications of this size get restructured, resized or shelved between first report and signing, and a hiring plan that moves on a pre-signing report is a hiring plan that moves twice. The cost of waiting two weeks for confirmation is zero. The cost of opening three requisitions against a deal that resizes is what I paid in 2024.

So I act on structure and direction rather than magnitude. The structure says efficiency rather than expansion. The direction says the asset class still attracts lenders at a workable margin, which is genuinely useful information about sector health. What neither says is that your Singapore shortlist got harder this week. Our notes on hiring a site reliability engineer in Singapore cover the role itself, and our notes on building an on-call rotation cover the structural piece operators compete on hardest, which is whether a person’s nights are predictable.

Back to the Three Requisitions I Mispriced

What went wrong in 2024 was not the forecast, it was the unit of analysis. I reasoned from an operator’s balance sheet to my own shortlist and skipped every step in between, because the announcement was large and specific and my plan needed a justification. The capacity was built. The jobs were real. They were two hundred kilometres away in a different labour market with different wages, and my requisitions sat open at a premium nobody local was paying.

The habit I took from it is narrow and has served me since: when infrastructure news breaks, I write down what I would expect to observe locally within sixty days if the story were going to affect me, then I check. Most of the time nothing moves and I have lost an hour. Occasionally something moves early and I am two months ahead of the market. Either way the plan does not change on the strength of a press report, which in a sector funding itself with this much debt is the discipline that matters most.

If you take one thing from Monday’s story, take the structure of the facility rather than its size. A three-year term at a floating margin tells you the borrower is keeping options open in an expensive money environment. That is a sensible thing for them to do, and for you it means the engineering demand it eventually produces is far more likely to be about running capacity efficiently than about building teams quickly.

FAQ: STT GDC Financing and Singapore Infrastructure Hiring