AI Data Centres Emitted 136x More Than Thought — UK Confirms Hidden Climate Cost in 2026

AI Data Centres Emitted 136x More Than Thought — UK Confirms Hidden Climate Cost in 2026

AI data centre emissions have been underestimated by up to 136 times, according to newly disclosed UK government projections in 2026. The revelation, confirmed by internal climate modeling, exposes a critical gap in how emissions from digital infrastructure are measured and reported. While direct energy use by servers and cooling systems has been tracked, the full lifecycle impact—including manufacturing, supply chains, and indirect power consumption—has been systematically overlooked.

Why Scope 3 Emissions Are Ignored in AI Carbon Accounting

The underestimation stems primarily from the exclusion of Scope 3 emissions, which encompass all indirect emissions across a company’s value chain. According to RPS Group, Scope 3 emissions often constitute the largest portion of an organization’s carbon footprint, yet they remain the least regulated and least reported. In the case of AI data centres, this includes emissions from semiconductor production, global hardware logistics, and the electricity grid’s carbon intensity where power is sourced.

The Hidden Energy Demand of AI Training Runs

Previous climate assessments relied heavily on direct energy consumption metrics, assuming that cloud providers’ renewable energy purchases offset their impact. But new analysis shows that even with renewable procurement, the sheer scale of AI training runs—requiring thousands of high-performance GPUs running for weeks—creates an emissions burden far exceeding prior estimates. The UK’s revised figures suggest that a single large AI model can emit more CO₂ than 300 average cars over a year, not just during operation but across its entire development lifecycle.

UK’s Reporting Gap vs. Global Standards

Experts point to a broader pattern of underestimated climate risks. A recent study cited by Forbes highlights how policymakers globally have consistently downplayed the cumulative impact of emerging technologies, favoring short-term economic growth over long-term environmental accountability. While the EU and US are drafting mandatory Scope 3 disclosure rules, the UK’s admission has accelerated calls for standardized, enforceable reporting frameworks tailored to digital infrastructure.

Green AI or Greenwashing? The Industry’s Dilemma

Industry leaders have responded with mixed reactions. While some major cloud providers have pledged to adopt full lifecycle accounting, others argue that the methodologies for measuring Scope 3 emissions in digital infrastructure are still immature. Critics counter that the technology exists—what’s lacking is political will. Without transparent, auditable carbon accounting, net-zero pledges risk becoming greenwashing exercises.

What’s at Stake: AI’s 2030 Energy Forecast

If current trajectories hold, AI-driven data centres could consume up to 8% of global electricity by 2030—equivalent to the entire current energy use of Japan. This surge in digital infrastructure emissions threatens to undermine decades of climate progress. Accurate measurement is the first step toward meaningful action, and without urgent, standardized reporting and enforceable limits, the digital revolution may become one of the largest, fastest-growing contributors to global warming.