Data Center Sustainability Metrics: Hidden Emissions

By 2024, Google claimed its data centers were 1.5 times more energy efficient than the industry average. In 2025, Microsoft has committed billions to nuclear power for AI workloads. The data center industry tracks energy usage efficiency down to three decimal places and optimizes water usage intensity through machine precision. We report direct emissions and energy emissions with religious fervor.

These are laudable advances, but these measures represent only 30% of total IT sector emissions. The majority of emissions come not directly from data centers or the energy they use, but from the end-user devices that actually access the data centers, emissions from hardware manufacturing, and software inefficiencies. We are frantically optimizing less than a third of the IT sector’s environmental impact, while the bulk of the problem goes unmeasured.

Incomplete regulatory frameworks are part of the problem. In Europe, the Corporate Sustainability Reporting Directive (CSRD) now requires 11,700 companies to report their emissions using these incomplete frameworks. The next phase of the directive, covering over 40,000 additional businesses, was initially planned for 2026 (but will likely be delayed until 2028). In the United States, the standards body responsible for IT sustainability measures (ISO/IEC JTC 1/SC 39) is conducting an active review of its standards through 2026, with a key plenary meeting in May 2026.

Now is the time to act. If we don’t fix the measurement frameworks, we risk ending up with incomplete data collection and optimizing a fraction of what matters for the next 5-10 years, before the next major standards revision.

Limited metrics

Walk into any modern data center and you will see sustainability instruments everywhere. Power efficiency (PUE) monitors track every watt. Water use efficiency (WUE) systems measure water usage to the nearest gallon. Sophisticated monitoring captures everything from server utilization to cooling efficiency to renewable energy percentages.

But here’s what these measures miss: End-user devices overall emit 1.5 to 2 times more carbon than all data centers combined, according to McKinsey’s 2022 report. The biggest problem is the smartphones, laptops, and tablets we use to access these ultra-efficient data centers.

Data center operations, measured by energy consumption efficiency, account for only 24% of total emissions.

In the conservative range of the McKinsey report, devices emit 1.5 times more than data centers. This means that data centers account for 40% of total IT emissions, while devices account for 60%.

Additionally, around 75% of device emissions occur not during use, but during manufacturing: this is called embodied carbon. For data centers, only 40% is embodied carbon and 60% comes from operations (as measured by PUE).

In total, data center operations, as measured by PUE, account for only 24% of total emissions. The embodied carbon of data centers is 16%, that of devices is 45%, and that of device operations is 15%.

Under the current EU CSRD, companies must report their emissions in three categories: direct emissions from owned sources, indirect emissions from purchased energy and a third category for everything else.

This “everything else” category includes appliance emissions and embodied carbon. However, these emissions are reported as overall totals broken down by accounting category (capital goods, goods and services purchased, use of products sold), but not by product type. The proportion of end-user devices versus data center infrastructure, or between employee laptops and network equipment, remains unclear and, therefore, unoptimized.

Reuse of embodied carbon and materials

Manufacturing a single smartphone generates around 50 kg of CO₂ equivalent (CO₂e). For a laptop, that’s 200 kg of CO₂e. With 1 billion smartphones replaced each year, this represents 50 million tonnes of CO₂e per year solely thanks to the manufacturing of smartphones, before anyone even turns them on. On average, smartphones are replaced every 2 years, laptops every 3 to 4 years and printers every 5 years. Data center servers are replaced approximately every 5 years.

Extending the life cycle of smartphones to 3 years instead of 2 would reduce annual emissions from the manufacturing sector by 33%. At scale, this dwarfs data center optimization gains.

There are programs aimed at reusing old components that are still functional and integrating them into new servers. GreenSKUs and similar initiatives show that an 8% reduction in embodied carbon is achievable. But these are pilot programs and not systematic approaches. And importantly, they are measured only in the context of the data center, not across the entire IT stack.

Imagine applying the same circular economy principles to devices. With over 2 billion laptops in existence worldwide and replacement cycles of 2-3 years, even modest lifespan extensions result in massive emissions reductions. Extending the life cycle of smartphones to 3 years instead of 2 would reduce annual emissions from the manufacturing sector by 33%. At scale, this dwarfs data center optimization gains.

Yet data center reuse is measured, reported and optimized. This is not the case for device reuse, as the frameworks do not require it.

The invisible role of software

As a leader in load balancing infrastructure on IBM Cloud, I see how software architecture decisions impact power consumption. Inefficient code doesn’t just slow things down: it increases both data center power consumption and device battery drain.

For example, researchers at the University of Waterloo showed that they could reduce energy consumption in data centers by 30% by changing just 30 lines of code. From my point of view, this result is not an anomaly, it is typical. Poor software architecture leads to unnecessary data transfers, redundant calculations, and excessive resource usage. But unlike data center efficiency, there is no commonly accepted measure of software efficiency.

This matters more than ever. As AI workloads drive a massive expansion of data centers — expected to consume between 6.7 and 12 percent of total U.S. electricity by 2028, according to the Lawrence Berkeley National Laboratory — software efficiency becomes essential.

What needs to change

The solution is not to stop measuring data center efficiency. It’s about measuring the durability of devices with the same rigor. Specifically, standards bodies (particularly ISO/IEC JTC 1/SC 39 WG4: Holistic Sustainability Metrics) should expand frameworks to include device lifecycle tracking, software efficiency metrics, and hardware reuse standards.

To track device lifecycles, we need standardized reporting on embodied carbon in devices, broken down separately by device. An overall figure in an “everything else” category is insufficient. We need specific device categories with visible manufacturing emissions and replacement cycles.

To include software efficiency, I advocate developing a PUE equivalent for software, such as energy per transaction, per API call, or per user session. This should be a reportable metric in sustainability frameworks so companies can demonstrate software optimization gains.

To encourage hardware reuse, we need to systematize reuse measures across the entire IT stack: servers and devices. This includes tracking repair rates, developing large-scale refurbishment programs, and tracking component reuse with the same detail currently applied to data center hardware.

To put it all together, we need a unified IT emissions tracking dashboard. CSRD reports should show device embodied carbon as well as data center operational emissions, making the full picture of IT sustainability visible at a glance.

These are not radical changes: they are extensions of measurement principles already proven in the data center context. The first step is to recognize what we are not measuring. The second is to build the frameworks to measure it. And the third is to require companies to present the complete picture: data centers and devices, servers and smartphones, infrastructure and software.

Because you can’t fix what you can’t see. And right now, we don’t see 70% of the problem.