Can data centers in Singapore keep up with overwhelming demand for AI?
Mingcheng Lim, Singapore’s country head at STT GDC talks about the ambition to close enterprise AI infrastructure gaps, while handling capacity constraints, rising energy costs, and sustainability issues.
While many enterprises in Singapore have shown their ability to progress past early-stage AI adoption, compared with peers across Asia, most still face an infrastructure gap and struggle to support demanding, scalable AI workloads.
This is according to a regional study by ST Telemedia Global Data Centres (STT GDC), which found that enterprises across Asia have ambitious AI goals but remain stuck in the building stage (71%).
However, even with many enterprises in Singapore (43%) meeting clear goals with robust, scalable infrastructure, a mere 10% report having sufficient headroom for demanding, scalable AI workloads. And only 10% say their network infrastructure is optimized for advanced AI workloads — suggesting a wide infrastructure gap.
Yet, Singapore’s ambition to lead in terms of AI readiness is only set to grow, driven partly by national goals. In a country where the sector faces mounting challenges — from capacity and space constraints, to sustainability limitations — are data centers in Singapore ready to meet burgeoning demand for AI?
“When we started 10 years ago, this data center was considered big [with] a 13 MW facility,” Mingcheng Lim, country head, Singapore at STT GDC told CRN Asia. Lim was sharing about the firm’s centrally-located STT Singapore 5 facility in Tai Seng.
“Now, 10MW, in terms of AI, is only phase one of [a project]. You need 50 to 100 MW or 200 MW. But that puts constraints on [data centers] with our ability to expand and [lack of] immediate land and power.”
Lim believes expansion constraints remains a top challenge in Singapore, but is optimistic that JTC Corporation’s plans to host the country’s largest green data center park on Jurong Island can remedy that for the sector.
JTC, together with the Economic Development Board (EDB), have set aside roughly 20 hectares of land that can provide up to 700MW of power capacity for the facilities. STT GDC have submitted their bid to build out a data center in the park.
“Beyond that, it is now very controlled to grow in Singapore,” he said. “That's why I think Malaysia, specifically Johor, has grown so tremendously over the past three or four years in terms of data center expansion.”
Helping customers manage rising costs
The next challenge is managing both energy consumption and sustainability issues, particular with rising costs due to global disruptions and the ongoing war in the middle east.
“How we’re helping to address that is to be more energy efficient,” said Lim. “I think it’s always important to be mindful of finding ways, even in this data center, to optimize the cooling.
“We did the core containment for particular racks, [and] are finding ways to also leverage our AI to manage our building management system. So we can tweak the temperature setting, the fans and stuff to reduce the energy cost.”
As for data center sustainability, Singapore has grand ambitions to adhere to the Paris Agreement but there are currently limitations in terms of availability of green technologies here. This is where STT GDC’s presence across Southeast Asia, namely Malaysia, Indonesia, Thailand, Vietnam, and the Philippines balances out their operations.
“Our places in Malaysia and Indonesia, there’s a bit more abundance in terms of solar, hydro energy and others,” he said. “We’re trying to leverage off [those facilities] and do our part to help make our usage sustainable from this perspective.”
With AI demand set to grow, these issues will continue to prevail. In the first quarter of 2026 alone, spending on technology services in Asia Pacific grew by double digits, powered by AI-driven demand for cloud infrastructure services, reported Information Services Group (ISG).
ISG found that demand for infrastructure-as-a-service advanced 19% to a record US$5.6 billion. So how can providers help customers manage expectations?
“What we can do is talk to our customers, talk to our providers, and get a sense of the known unknowns,” said Lim. “Try and figure out where they are. Then we try and build flexibility and be a bit more staged in terms of deployment.
“Say, this is what I can do in the next 12 to 18 months, and this is what we agreed upon. There's opportunity to accelerate, but subject to suppliers and stuff like that, and we'll work towards that. So I think the main thing is to have a flexible mindset.”