Singapore, 30 September 2026 — Joe Gooi, CEO of SC Zeus Data Centers, delivered a keynote at Data Centre World Asia 2026 on how data centre operators can respond to growing AI and cloud demand across Asia Pacific.

Speaking at the Critical Infrastructure Theatre, Joe presented “From Megawatts to AI Compute” addressing the conference theme, “Adapting to Rapid Change: How Data Centres Can Keep Up with the Surge in AI and Cloud Demand.” His central message was that securing power is only the first step. The measure of a data centres value is whether it can turn that power into productive compute for customers.

From secured power to productive compute

AI is changing the scale and technical requirements of data centre demand at the same time. Customers are seeking larger capacity blocks, higher rack densities and cooling systems capable of supporting evolving compute platforms. Meanwhile, the power connections, equipment and buildings needed to serve them can take years to deliver.

Joe urged the industry to be more precise about what a capacity figure represents. Power may be contracted, but it must still become energizable, commissionable and deployable before a customer can install equipment and run workloads. Each stage has its own dependencies and delivery risks.

For customers, the essential question is when their compute can begin operating. Answering it requires operators to coordinate the full chain from grid connection through electrical distribution, cooling, commissioning and customer deployment.

Designing infrastructure that can adapt

A data centre may operate for decades, while the technology it houses can change within a few years. Joe described the need to distinguish long-life infrastructure, such as the utility interface and primary electrical systems, from the rack configurations and customer interfaces that may evolve over time.

Cooling is a clear example. AI deployments can require different combinations of air, hybrid and direct liquid cooling. Decisions about thermal architecture affect which equipment a facility can support, the density it can accommodate and how quickly a customer can deploy.

Joe also addressed the investment decisions behind flexibility. Operators need to secure power and build essential infrastructure early, while phasing customer-specific fit-out in line with committed demand. The aim is to preserve useful options where requirements are likely to change, without adding flexibility that has no clear value.

Applying the approach at ZEUS OSA 1

Joe highlighted ZEUS OSA 1, SC Zeus’s Osaka AI-ready hyperscale campus, as an example of this approach in practice. The project has a secured 100 MW campus power allocation and 70 MW of planned IT capacity across two phases.

The campus has been designed to accommodate high-density AI infrastructure, including a rack power design of 230 kW per rack for the specified NVIDIA Vera Rubin configuration. Its planned architecture can support air, hybrid and direct liquid cooling configurations on the same floor, allowing customer fit-out to respond to different deployment requirements.

Joe distinguished between the project’s current status and its intended capabilities: the campus power allocation is secured, the IT capacity is planned, and the rack-scale and cooling capabilities are designed for. Making those distinctions clear helps customers and partners understand the path from a development plan to usable capacity. These milestones represent different stages of development: campus power is secured, IT capacity is planned, and the high-density and cooling capabilities are incorporated into the completed design. The next step is to translate these foundations into commissioned capacity ready for customer deployment.

Coordinating the path to deployment

Joe described an expanding role for data centre operators. Utility schedules, technology choices, customer requirements and capital plans all affect one another. Bringing those decisions together early can expose conflicts before they delay a deployment.

He closed with five priorities for the industry: secure power early, design for a range of requirements, make cooling a strategic part of the customer conversation, align partners early and phase customer-specific investment against committed demand.

As AI demand grows across Asia Pacific, SC Zeus is focused on a practical outcome: developing infrastructure that can convert scarce power into productive compute, while remaining adaptable as customer needs and technology change.