ESG Meets IT: Smarter Networks, Cleaner Footprints
From server rooms to boardrooms, IT carbon is under audit.
Technology chiefs spent the better part of a decade chasing digital transformation without once being asked to account for its environmental cost. That free ride is finished. As disclosure mandates tighten and investors sharpen their scrutiny of Scope 3 reporting, the collision of ESG meets IT has moved from a fringe sustainability conference talking point to a live item on the audit committee’s agenda. The question confronting every CIO is no longer whether their infrastructure has a carbon problem, but how large that problem actually is.
The headline figures are sobering. Data centres and transmission networks account for approximately 1% of global energy-related greenhouse gas emissions, according to the International Energy Agency. Under its central scenario, data centre electricity consumption will more than double by 2030 to 945 terawatt-hours, roughly equivalent to Japan’s entire current demand. AI is the accelerant: workloads accounted for between 15% and 20% of total data centre power demand in 2024 and could consume half of it by the end of the decade. Morgan Stanley projects data centre emissions will triple from around 200 million metric tonnes today to 600 million tonnes annually by 2030. For companies that have staked reputational capital on net zero pledges, these are deeply inconvenient numbers. The question of how ESG meets IT in practice has never been more urgent.
Data Centre Efficiency: Necessary but Not Sufficient
The instinct has been to optimise what is most easily measured. Power usage effectiveness has become the benchmark, and the hyperscalers compete on it with almost athletic intensity. AWS posted a global PUE of 1.15 in 2024, comfortably below the public cloud average of 1.25. Google cut data centre emissions by 12% in the same year, even as energy use surged, by bringing 2.5 gigawatts of contracted clean energy online.
Impressive. Also incomplete. Research published in IEEE Spectrum argues that the metrics the industry obsesses over capture only around 30% of total IT sector emissions. End-user devices collectively emit 1.5 to two times more carbon than all data centres combined, and roughly 75% of those device emissions are embodied carbon, generated during manufacturing. “We are frantically optimising less than a third of the IT sector’s environmental impact, while the bulk of the problem goes unmeasured,” the authors note. Any CIO who treats a healthy PUE score as a credible sustainability strategy is measuring the wrong thing.
ESG Meets IT in the Code Itself
If hardware efficiency has been over-indexed, software efficiency has been almost entirely overlooked. This is starting to change, and it may prove to be the most powerful lever IT teams possess.
Poorly architected software is not merely technical debt. It is an energy liability. Every redundant database query, every bloated JavaScript bundle, every unoptimised algorithm translates into additional compute cycles and additional kilowatt-hours. Asim Hussain, executive director of the Green Software Foundation and director of green software at Intel, has been blunt about the gap. The foundation’s Software Carbon Intensity specification, now approved by the International Organization for Standardization, gives engineering teams a carbon-per-unit-of-work metric to sit alongside latency and uptime. As Hussain has argued, measurement is the prerequisite: once engineers can see the problem, they will solve it.
The practical gains are material. Choosing inherently efficient languages such as Rust or C over resource-hungry alternatives like Python cuts execution times and, by extension, electricity consumption. At QCon London, one team reported that migrating from sprawling microservices to a modular monolith delivered measurable reductions in resource consumption and carbon output without significant additional cost. Frontend optimisation, including lazy loading, JavaScript trimming and static generation, reduces the energy burden not only on servers but on the end-user devices accessing them, tackling both sides of the emissions equation at once.
Carbon-aware scheduling adds a further dimension. By shifting batch workloads to periods of low grid carbon intensity, or routing them to cleaner regions, IT teams can cut emissions from a given task without changing the task itself. Several cloud providers now expose regional carbon intensity data through their APIs, making this operationally feasible rather than theoretical.
ESG Meets IT: The Hardware Lifecycle Reckoning
The European Union’s Corporate Sustainability Reporting Directive is steadily widening the scope of mandatory disclosure. For IT departments, the hardest scrutiny will fall on Scope 3: the embodied carbon embedded in every server, switch, laptop and handset they procure.
The challenge is stark. Microsoft’s total emissions have climbed roughly 29% since its 2020 baseline, driven overwhelmingly by Scope 3. The company’s own sustainability report attributed the increase to construction materials and hardware components, particularly semiconductors, servers and racks. Google’s Scope 3 emissions rose more than 20% in 2024 alone, even as its operational footprint improved. These are not outliers. They are structural features of an industry building at unprecedented pace.
Extending device lifespans by even one to two years can deliver outsized reductions in embodied emissions. Dell has adopted a circular economy approach, recovering old hardware and recycling materials into new production. Yet the industry simultaneously fuels software-driven obsolescence, where functioning devices are rendered unusable by bloated updates demanding ever greater processing power. Global e-waste is projected to exceed 75 million metric tonnes by 2030. Writing efficient code that runs on older hardware is not just good engineering. It is good governance.
Cloud Migration: A Carbon Strategy That Demands Scrutiny
For many organisations, the single highest-impact decision remains where and how they host workloads. AWS claims its infrastructure runs up to 4.1 times more efficiently than typical on-premises alternatives. Google has signed power purchase agreements covering more than 22 gigawatts of clean energy since 2010, contracting over eight gigawatts in 2024 alone.
These claims warrant careful examination. Geography matters enormously: the carbon intensity of electricity grids varies by orders of magnitude across regions. Relocating an application from a coal-dependent grid to one dominated by renewables can transform its emissions profile overnight. But location-based accounting tells a different story from market-based accounting. Research from Stand.earth found that Microsoft’s data centre emissions under location-based methods were nearly 22 times higher than the company’s reported figure. Renewable energy certificates, widely used by hyperscalers to claim green credentials, do not always translate into genuinely decarbonised supply. IT leaders migrating to the cloud need to interrogate not just the provider’s headline commitments but the grid reality of the specific region hosting their workloads.
IT Sustainability: From Pledge to Mandate
The year 2025 was when ESG meets IT shifted from voluntary ambition to enforceable expectation. The W3C published its Web Sustainability Guidelines, offering the first structured international framework for reducing the environmental footprint of digital products. Industry leaders began integrating carbon expenditure into total cost of ownership models for AI, creating what some analysts call a “return on compute” framework that prices environmental cost alongside financial return.
The organisations that navigate this best will be those that treat ESG meets IT not as a compliance checkbox but as an operating principle woven through procurement, architecture, software engineering and cloud strategy. Leaner code runs faster. Efficient hardware lasts longer. Cleaner cloud regions are frequently cheaper. The incentives are, for once, aligned. The companies slow to recognise this will not merely fall behind on regulation. They will fall behind on performance.
