What this covers:
- Which parts of sustainability an operator controls, and where accurate asset data makes a difference
- What tenants, auditors and regulators expect, and why better records can give a provider an advantage
- Why high-density GPU racks and liquid cooling make PUE harder to read on its own
The variables, and which ones are yours
When people ask me what sustainability means for a data centre or telco operator, I give them a short list:
- How efficiently you run your sites: the power, the cooling, and the mechanical and electrical plant behind them.
- How often you send an engineer to a site, and how often they have to go back.
- What happens to equipment when you have finished with it.
- What the grid is made of at the time you draw your power.
- The building itself, and in some cases whether you needed to build it at all.
Each of these affects an operator’s environmental footprint. Some are covered by mandatory data centre reporting; others matter in corporate carbon accounting, customer sustainability assessments and day-to-day operations.
Some of that list is not ours to solve, and I would rather say so plainly. The carbon intensity of electricity depends on the grid, how energy is purchased and the accounting method used, and operators rely on meters, suppliers and specialist emissions data for those figures. How efficiently a server runs its software is largely a question for IT. Between the two sits the physical estate: what is installed, where it is, what it is connected to, what it draws and whether it is still needed. The asset register is the record of that estate, and it gives operators a way to understand and improve much of what happens within it.
Reporting is now a reputation and growth risk
The EU is leading on data centre sustainability reporting. Sites with 500 kW or more of installed IT power demand already report energy and water consumption, renewable energy use and other performance indicators each year. In September 2026, the European Commission advanced plans for a common sustainability rating scheme built on that information. The direction is towards transparency and comparability, which means operators need reliable measurements and supporting records to show how their figures were produced. Nameplate ratings can help establish installed capacity, but they are not a substitute for measured consumption where metering is required.
For colocation operators, pressure also comes from customers. Large organisations increasingly need reliable information on the energy and emissions of their outsourced infrastructure, and some expect their providers to help account for it. A provider that can explain its figures, show the quality of its data and demonstrate what it is doing to improve efficiency may have an advantage at renewal. Penalties for missing reporting obligations are possible under national rules, but the commercial risk matters too: a competitor with better evidence may be better placed to win the next tenant.
The same data has an operational benefit. Decisions about capacity, cooling and refresh are only as good as the information behind them. Metering shows how much energy is being used. An accurate physical asset record, updated in real time, helps explain where that energy goes and what is consuming it.
Running the site efficiently
Across data centres and telco networks, equipment can stay powered long after the service it supported has ended: a server nobody uses, a switch left behind after a customer moves out, kit installed for a project that never went live. Equipment doing useful work earns its power and cooling. Equipment doing nothing is pure cost, to the business and to the carbon figure. Research by Jonathan Koomey and Jon Taylor highlighted the problem of servers drawing energy while doing no useful work. The scale varies between organisations, and in colocation facilities the operator may not control, or even know, what is running on a customer’s equipment. The physical question remains: what is still installed, connected and consuming resources? You cannot investigate what you cannot find.
Airflow tells the same story at rack level. When equipment is removed and the gap is not blanked off, conditioned air can bypass the equipment it was meant to cool. Hot and cold air can mix, reducing cooling effectiveness and increasing the fan or cooling effort needed to maintain safe conditions. The operator pays for energy it could have avoided.
Overcooling can follow. Without a reliable view of rack loads, equipment requirements and environmental conditions, raising a cooling setpoint is a risk nobody wants to own, so sites may run colder than necessary rather than risk a hot spot. An accurate record of what is installed in each rack, combined with temperature and airflow monitoring, gives operations teams better evidence to make those adjustments safely.
Sending engineers only when they are needed
For telcos, this is one of the largest variables. A national network can run thousands of sites, many of them small and unstaffed, spread across the country to serve every community. Every visit costs fuel, time and money. When an engineer arrives to find the part is not there, the fault is elsewhere, or the connection they planned to move cannot be released, one visit becomes two. The usual cause is a record that did not match the site.
Colocation operators feel this less, because their engineers are already in the building. The principle still applies at a smaller scale: every wasted trip to a hall or rack is time that could have gone elsewhere.
Retiring equipment responsibly
At end of life, equipment can be resold, refurbished, recycled or sent to landfill, with very different environmental and financial consequences. The UN’s Global E-waste Monitor found that less than a quarter of the world’s electronic waste was documented as formally collected and recycled. To substantiate claims about responsible reuse and disposal, an operator needs the asset’s history: where it came from, where it was installed and what happened when it was removed, supported by transfer records, refurbishment documentation and recycling certificates.
The most sustainable data centre is the one you never built
This point can appeal to both the sustainability lead and the finance director. Capacity recovered from equipment nobody is using may not need to be replaced or built elsewhere. Recovering it can delay new infrastructure, reduce capital expenditure and avoid some of the embodied carbon in manufacturing equipment and constructing buildings. It will not remove every need for expansion, but making better use of what is already installed is often one of the most practical places to start, and it can improve the margin on the sites you already have.
PUE needs context
PUE has been the industry’s headline efficiency figure for years. It compares the energy consumed by the whole data centre with the energy consumed by its IT equipment, so everything outside the IT load, including much of the cooling and electrical infrastructure, counts as overhead. It is a useful measure, but it does not show how efficiently the IT equipment performs useful work.
High-density GPU racks and liquid cooling make that distinction more important. Cooling equipment can now sit inside servers, racks or separate distribution systems, with pumps, fans and controls powered and metered in different ways, and where that energy is counted affects how PUE should be read. Uptime Institute’s analysts have also noted that cutting server fan energy can improve overall efficiency while making PUE look worse, because the saving falls on the IT side of the calculation.
None of this makes PUE worthless. It does mean that comparing sites on PUE alone can hide real differences in cooling architecture, power density and energy consumption. Understanding what sits behind the number depends on accurate metering and on knowing what equipment is installed, how it is connected, and how it is cooled.
Small gains, repeated across an estate
No single grand project is needed. Blanking a rack, raising a setpoint, retiring a forgotten server and avoiding a second visit are each small improvements. Repeated across hundreds of racks or thousands of sites, they can add up to meaningful savings in energy, cost and carbon. Measuring those savings takes the right operational data, but finding the opportunities often starts with a physical asset register that matches what is actually installed.
Where Assetspire fits
We built Spire™ to provide that foundation: a current, accurate record of the physical estate, captured by the engineers doing the work at the time they do it. The grid and carbon specialists can build on it. Without it, they are building on estimates.
Sustainability will be judged on evidence
Your sustainability reputation will be judged on evidence, and much of that evidence depends on knowing what is physically installed. If your largest tenant asked you to show, asset by asset, how the carbon figure for their space was derived, could you?
Steve Beber is the CEO at Assetspire. If you would like to continue the conversation, he is available on LinkedIn.