Green Data Centre Checklist: 7 Infrastructure Indicators of True Energy Efficiency

A PUE of 1.2 to 1.3 is regarded as optimal for modern data centres with smart cooling and power systems. Older facilities often have PUES between 1.5 and 1.6, while values exceeding 1.8 suggest considerable untapped efficiency potential that remains unaddressed. However, the key consideration is how the PUE is measured; relying on a single number without context does not provide a meaningful assessment.

To ensure accuracy, request annualised PUE data obtained through continuous facility-level metering over an entire calendar year, rather than a snapshot taken under ideal conditions. The Green Grid PUE Category 3 represents the highest verification standard, requiring ongoing measurement at the facility boundary. Third-party audit reports aligned with ASHRAE TC 9.9 add further credibility. Operators unwilling to provide annual, continuously measured PUE data should be approached with caution, regardless of the figures they present in sales discussions.

The savings are substantial and straightforward to calculate. For an organisation with a 500kW IT load, operating in a PUE 1.6 facility instead of a PUE 1.2 facility results in an unnecessary 200kW energy consumption every hour. At ₹8/kWh when running around the clock, this difference amounts to roughly ₹1–1.4 crore annually, purely due to efficiency disparities, without any changes to the IT setup.

Over a typical five-year colocation contract, this leads to avoidable electricity costs of ₹5–7 crore. Larger deployments see proportionally greater savings. Besides direct energy costs, an energy-efficient data centre also cuts cooling maintenance expenses, lowers the risk of equipment failures caused by heat, and reduces carbon credit liabilities for organisations committed to net-zero or BRSR reporting. Generally, investing in efficiency during the selection process offers a stronger business case than negotiating discounts on colocation prices.

PUE and carbon footprint are interconnected but distinct metrics and over-relying on PUE alone when selecting a data centre is a common mistake. PUE indicates how efficiently a facility uses energy, but it does not reflect the energy source or its carbon impact.

For example, a facility with a PUE of 1.2 on a coal-heavy grid will have a larger carbon footprint than one with a PUE of 1.4 powered by renewable energy through a Power Purchase Agreement. To fully assess carbon impact, consider three factors together: the annualised PUE, the proportion and verification of renewable energy used, and whether the operator can provide specific Scope 1 and Scope 2 emissions data for your deployment. In India, tenants with BRSR reporting duties or net-zero commitments need Scope 2 emissions data at the individual tenant level, not just facility-wide averages, as this appears in sustainability disclosures.

Your answer depends on your current and future rack density. For deployments under 15–20 kW per rack, well-maintained precision InRow cooling with hot- and cold-aisle containment is sufficient and efficient. Once you exceed this threshold, the cooling setup must adapt to your workload, and many colocation agreements pose a hidden risk by not specifying how the facility will handle higher density demands.

For AI, GPU, or HPC workloads above 30–50 kW per rack, direct-to-chip liquid cooling with Coolant Distribution Units (CDUs) and cold plates is now standard, cutting cooling energy use by 30–60% compared to indoor air cooling. Rear-door heat exchangers serve as a practical transition from air to liquid cooling without extensive infrastructure changes.

When choosing a facility, ask for their cooling roadmap for racks above 20 kW and verify if liquid-cooling systems are available in your designated zone, not just elsewhere in the building. A facility that cannot provide this information in 2026 may not be capable of supporting efficient workload growth.

Choosing a colocation provider requires assessing infrastructure that isn't directly visible or controllable. Hence, asking the right questions during the RFP and site visit is crucial. The seven-indicator framework discussed here is directly applicable, along with some additional points specific to colocation relationships.

First, verify if the contract specifies energy transparency, ensuring you receive tenant-level energy consumption data instead of just facility-wide metrics. Second, inquire whether the renewable energy commitments apply to your contracted capacity or just the facility's overall energy mix, as these can differ. Third, review the facility's PUE history over the past three to five years and ask about infrastructure investments that led to improvements; these details reveal operational discipline beyond just the current PUE figure.

Finally, determine if the operator adheres to recognised sustainability standards such as the Climate Neutral Data Centre Pact, LEED, or ISO 50001. Operators with third-party verification demonstrate greater transparency, which safeguards your long-term operational costs and carbon reduction commitments.