Equinix announced the opening of SG6 in Singapore on November 6. Its English release states that the first phase provides 8 MW of capacity and approximately 1,550 cabinets, with up to 20 MW of IT capacity at full build-out.
For technical buyers, another detail deserves attention. The company's Japanese announcement says customer IT servers deployed at SG6 must have recognized energy-efficiency certification or demonstrate equivalent efficiency, in connection with Singapore standard SS 715:2025.
My operational reading is that server eligibility should be resolved before equipment ships. The opening creates a new placement option. A successful deployment still needs agreement on the particular hardware, its facility interfaces and the evidence required for acceptance.
Resolve the hardware evidence before ordering
Ask the operator what documentation it accepts for the exact server configuration being proposed.
A useful submission should identify the server model, installed components and relevant certification or equivalent evidence. Have the hardware supplier explain which configurations the documentation covers. Ask the operator how it handles changes after approval, including a substituted component or a revised server build.
This is a proposed procurement process, not an interpretation of every requirement in the standard. Obtain the applicable admission criteria directly from Equinix and assign responsibility for resolving any uncertainty.
The practical objective is to avoid discovering a documentation gap after the hardware has arrived.
Match cooling capabilities to the actual rack
Equinix describes SG6 as supporting high-density AI infrastructure with liquid-cooling capabilities. That facility-level description is a starting point for the interface discussion.
For the proposed deployment, request the supported temperature, flow and pressure ranges, coolant requirements and physical connection details. Confirm where responsibility passes between the facility, coolant distribution equipment and the servers.
Include the residual air-cooled load in that discussion. Ask the server supplier to identify heat that remains outside its liquid-cooling path and have the operator confirm how that heat will be handled in the intended placement.
Record those answers against the equipment list. A change in the rack design should trigger an interface review before installation.
Keep power reservations and operating assumptions together
The announced phase capacity does not specify the allocation available to an individual customer.
My recommendation is to review the contracted power alongside the rack's expected behavior. Agree on the planning basis for normal operation, peak demand and any permitted load changes. Confirm how the installation is expected to behave during maintenance or a supply-path event.
Have the operator and hardware supplier review the same assumptions. Differences between their documents should be resolved while the design can still change.
This is also a useful point to identify the telemetry each party will receive. A shared record of power and cooling conditions can help distinguish an equipment issue from an interface problem during commissioning.
Treat the PUE figure as a design target
Equinix says SG6 is designed for a Power Usage Effectiveness of 1.20 and is targeting Green Mark Platinum certification. Those are design and certification targets in the opening announcement.
For an operational comparison, ask what measured information will become available and under which reporting boundary. Record the period, facility loading and relevant operating conditions alongside the result.
Keep that facility metric separate from the application's useful work. A workload acceptance exercise should also measure the service outcomes the customer needs, such as latency, throughput and completion time, under an agreed configuration.
That provides a clearer basis for deciding whether the deployment meets its objectives than treating a design figure as a measured result for a particular workload.
Agree on acceptance before production traffic
A practical next step is a joint readiness review covering hardware eligibility, electrical service, cooling interfaces and installation responsibilities.
Turn unresolved items into explicit prerequisites for delivery or commissioning. Agree on the test workload, the evidence to retain and the people authorized to accept the result. Include the escalation route for an issue that crosses the facility and server support boundaries.
The SG6 opening is the new development. For an AI infrastructure team considering the site, the immediate work is to connect that opportunity to an approved hardware configuration and a documented acceptance process. That makes the eventual move from delivery to production easier to assess and manage.
