Comparing data centre sustainability requires more than selecting a provider that uses the words “green” or “renewable”. This guide gives technology buyers a practical framework for reviewing energy efficiency, renewable-power claims, carbon reporting, water use, waste, and equipment lifecycle data across colocation, cloud, and dedicated-server facilities. It also provides a repeatable tracking process for procurement reviews, ESG reporting, and hosting contract renewals.
Overview
A sustainability comparison should examine both the facility and the service being purchased. A data centre may operate an efficient building, for example, while a specific hosting product uses older hardware, has a low server utilisation rate, or depends on a power mix that changes by region and time. Conversely, a provider may offer useful carbon accounting but disclose little about water, waste, or hardware replacement.
Start by defining the boundary of the comparison. Are you assessing a single colocation site, a cloud region, a dedicated-server product, or the provider’s entire estate? Record the facility location, service type, reporting period, and whether the figures apply to the whole site or only to IT equipment. Without these details, apparently similar claims may not be comparable.
Sustainability should also be assessed alongside operational requirements. A lower-impact facility is not automatically the right choice if it cannot meet your latency, resilience, data residency, security, or capacity needs. For context on location and infrastructure trade-offs, see the data centre location comparison guide and the data centre due diligence checklist.
What to track
Power usage effectiveness
Power usage effectiveness, or PUE, compares total facility energy with the energy used by IT equipment. A lower PUE generally indicates that less additional energy is being used for cooling, power conversion, lighting, and other facility functions. However, PUE is not a complete sustainability score. It does not show the carbon intensity of the electricity, the workload performed, the age of equipment, or the amount of renewable power procured.
Ask for the reporting period, measurement boundary, calculation method, and whether the figure is an annual average or a design target. A design PUE is not the same as measured operational performance. Where possible, request separate figures for the facility you would actually use rather than a company-wide average.
Renewable-energy sourcing
“Renewable powered” can describe several different arrangements. A facility may generate some power on site, buy renewable electricity through a contract, match consumption with certificates, or report a grid-adjusted emissions figure. These approaches should not be treated as identical.
Ask the provider to explain what is being matched, for which period, and whether the claim applies to the specific data centre or to the wider organisation. Request information about the underlying instruments or contracts where relevant, while recognising that the accounting treatment may depend on the reporting framework used. Also record the facility’s residual grid electricity and location, since regional grid conditions can affect operational emissions.
Carbon reporting
Track the emissions boundary, not just a single carbon number. Ask whether the provider reports direct facility emissions, purchased-energy emissions, and relevant value-chain emissions. For a hosting buyer, embodied emissions from servers, networking equipment, batteries, construction, and replacement cycles may be important even when operational electricity is reported separately.
Check whether figures are independently assured, internally reviewed, estimated, or based on provider-specific assumptions. A transparent methodology with stated limitations is more useful for comparison than an impressive figure with no boundary or period.
Water, cooling, and local conditions
Water usage effectiveness, or WUE, can help buyers understand water used for data-centre cooling relative to IT energy or computing activity, depending on the provider’s methodology. Ask whether the figure includes potable water, reclaimed water, rainwater, or water used indirectly for power generation. Also ask how seasonal conditions affect cooling and whether the site faces local water-stress concerns.
Cooling technology matters, but labels such as “free cooling” or “liquid cooling” need context. Request measured results and the operating conditions under which they were achieved. A design feature may have limited benefit if it is rarely available in the facility’s climate or workload profile.
Waste and equipment lifecycle
Track how servers, storage devices, batteries, packaging, and cooling components are handled at end of life. Useful questions include whether equipment is repaired, reused, refurbished, recycled, or destroyed; how data-bearing devices are sanitised; and whether the provider reports diversion rates or hazardous-waste controls.
Hardware efficiency is part of the picture. A newer server may deliver more work per unit of energy, but frequent replacement can increase embodied emissions and electronic waste. Ask about typical refresh cycles, component reuse, spare-parts policies, and the ability to select efficient hardware without compromising performance.
Evidence and comparison table
Keep a record of the source and date for every figure. Provider sustainability pages are useful starting points, but procurement teams should request facility-level documentation, methodology notes, audit statements, and contract-relevant commitments where available.
| Hosting model | Key sustainability evidence to request | Common comparison limitation |
|---|---|---|
| Colocation | Facility PUE, power procurement, cooling and water data, waste processes, metering method, and customer energy-allocation rules. | Your equipment’s age, utilisation, and refresh cycle may sit outside the facility’s headline metrics. |
| Cloud | Region-level energy and carbon methodology, workload reporting tools, renewable-power claims, hardware lifecycle information, and service-region boundaries. | Shared infrastructure makes it difficult to attribute facility impacts to one account without provider-specific allocation data. |
| Dedicated server | Server model and utilisation assumptions, facility metrics, power allocation, hardware replacement policy, and disposal controls. | A server may be reserved for one customer but still be underused, reducing efficiency per unit of useful work. |
Cadence and checkpoints
Use a three-stage review rather than treating sustainability as a one-time buying criterion.
- At shortlist stage: collect the latest facility-level PUE, renewable-energy explanation, carbon boundary, water information, waste policy, and evidence of assurance. Mark missing data as “not disclosed” rather than assuming it is zero.
- Before contract signature: confirm which claims apply to the selected region, product, and term. Clarify whether reporting is included, whether methodology changes will be notified, and whether sustainability requirements can be included in service reviews.
- Monthly or quarterly: monitor service-region changes, energy or carbon dashboards, renewable-power updates, incident reports, hardware changes, and any change to the provider’s reporting method. Most buyers do not need to recalculate every metric each month, but they should log newly published figures and material changes.
- Annually: refresh the comparison using the latest reporting period and review whether your own workload, utilisation, traffic pattern, or server footprint has changed. Update assumptions used in ESG or carbon reporting.
Maintain a simple register with these columns: provider, facility or region, hosting model, reporting period, PUE, renewable-power basis, carbon boundary, water metric, waste evidence, hardware information, assurance status, source URL, and review date. This makes changes visible and prevents old marketing claims from being carried into a new procurement decision.
How to interpret changes
A changing metric is not automatically an improvement or deterioration. A provider may report a higher PUE after moving from a design estimate to measured operational data. A lower carbon figure may result from a revised emissions factor, a new renewable-energy contract, or a change in reporting boundaries rather than a physical reduction in energy use.
When a number changes, ask four questions: what changed operationally, what changed in the calculation, what period is being compared, and does the change apply to the facility or only to the corporate average? Preserve both the old and new methodology in your records.
Compare like with like wherever possible. Do not rank a colocation facility’s PUE directly against a cloud provider’s workload carbon estimate. Instead, assess whether each provider gives enough information to understand the impact of your chosen service. For example, a dedicated server comparison should consider utilisation and hardware generation, while a cloud comparison should consider region selection, workload efficiency, and the provider’s allocation method.
Sustainability should remain part of a wider risk assessment. A facility with useful environmental reporting may still be unsuitable if it lacks required connectivity, resilience, backup arrangements, or compliance controls. Review those factors alongside sustainability using the relevant guidance on hosted application disaster recovery and low-latency data centre locations.
When to revisit
Revisit your data centre sustainability comparison at least annually, and use a monthly or quarterly watchlist for material developments. Start an earlier review when a provider changes facility, cloud region, energy-sourcing policy, reporting boundary, cooling design, hardware platform, or waste contractor. Contract renewal, an office or workload migration, a new ESG reporting requirement, and a major capacity expansion are also useful checkpoints.
At each review, archive the previous evidence, request clarification for unexplained changes, and update the shortlist rather than relying on the original procurement notes. If the assessment identifies a better-fit location or hosting model, test performance and migration risk before making a change. The server migration guide can help structure that evaluation.
The practical goal is not to find a universal “greenest” provider. It is to build a defensible, repeatable comparison that connects facility data with your actual workload, location, resilience needs, and reporting obligations. That approach makes sustainability claims easier to verify and keeps the decision useful as both infrastructure and provider disclosures evolve.
