AI's Power Surge Is Straining the Grid — What It Means for Your Carbon Reduction Plan

In the same week Carbon Brief revealed that CO2 emissions from UK data centres could be running "hundreds of times" higher than official estimates suggest, Beijing launched a three-year campaign forcing its steel, cement and aluminium plants to hit strict efficiency benchmarks or shut down. Two very different stories from opposite ends of the world, but the same lesson for any UK business building a carbon reduction plan: the emissions hiding in your energy supply and your supply chain are bigger — and under far more scrutiny — than most companies assume.

The AI boom nobody priced into their carbon accounting

Britain's current data centre estate is small compared with what's in the pipeline: around 1.8 gigawatts (GW) of capacity, consuming just over 2% of national electricity, according to Carbon Brief's own analysis. But 71 data centres have already reached "financial commitment", representing roughly 20GW of demand — and the full National Energy System Operator (NESO) grid connection queue lists around 140 proposed schemes seeking a combined 50GW, comparable to the UK's entire peak electricity demand. Most of that wider queue won't get built; Ofgem is now consulting on reforms to filter out speculative projects without firm plans. NESO forecasts consumption rising from around 5 terawatt-hours in 2025 to 22 terawatt-hours by 2030 — roughly 8–9% of Britain's total electricity demand.

Carbon Brief's own modelling, based on the smaller, already-committed 20GW pipeline rather than the full speculative queue, found that even modest reliance on gas-fired power could push emissions far beyond official projections. If gas supplied just 5% of electricity for an 11.2GW build-out, Carbon Brief calculated annual emissions of around 2 million tonnes of CO2 — more than ten times the government's own top estimate for 2035. Push gas reliance higher, and it estimated annual emissions could exceed 30 million tonnes of CO2, roughly equivalent to Denmark's total annual output. Globally, the picture is similarly stark: data centre electricity demand rose 17% in 2025 alone, nearly six times faster than overall global electricity demand growth, and the International Energy Agency expects gas and coal to still meet over 40% of the additional power needed for that growth by 2030.

China's three-year squeeze on heavy industry

While the UK grapples with AI-driven demand growth, China has taken the opposite approach with its industrial base. In June, the National Development and Reform Commission launched an "intensive campaign" running from 2026 to 2028 targeting nine of the country's most carbon-intensive sectors: steel, electrolytic aluminium, cement, flat glass, oil refining, ethylene, synthetic ammonia, methanol and coal-fired power. An NDRC official said in a Q&A published by the regulator itself that more than 70% of China's steel, aluminium, cement and flat glass capacity currently fails to meet energy efficiency benchmarks — the low-hanging fruit the new plan targets first, with non-compliant capacity phased out if it isn't upgraded by 2028.

This sits inside China's newly published 15th Five-Year Plan, which sets a target to cut carbon intensity (emissions per unit of GDP) by 17% between 2026 and 2030 — a slightly softer goal than the 18% target the country set, and missed, for the previous five years. Analysts have called it a cautious target that could allow absolute emissions to keep rising until 2030. But paired with mandatory efficiency upgrades and a plan to replace 30 million tonnes of coal a year with renewables, it's a reminder that decarbonisation is increasingly driven by top-down industrial policy as much as by voluntary corporate pledges.

What this means for SECR and ESG reporting

For UK organisations in scope of Streamlined Energy and Carbon Reporting (SECR), both stories matter more than they might first appear. As grid carbon intensity shifts with data-centre-driven gas reliance, location-based Scope 2 figures for any business — not just tech firms — could move in ways that aren't obvious from last year's SECR reporting baseline. And as cloud computing and AI tools become embedded in everyday operations, the emissions tied to that usage increasingly belong in Scope 3, under purchased goods and services.

The problem is that most cloud and AI vendors still don't provide activity-based emissions data granular enough for robust carbon accounting — many disclosures lag well behind the actual environmental footprint Carbon Brief and others are now describing. That's a live tension for ESG reporting more broadly: frameworks built around the GHG Protocol expect disclosure of material Scope 3 categories, but the data needed to do so accurately for cloud and AI usage is still catching up.

Building a carbon reduction plan that accounts for digital emissions

None of this changes the fundamentals of a good carbon reduction plan — it just adds a category most SMEs haven't accounted for yet. A few practical steps:

1. Audit your digital footprint. Map which cloud platforms, SaaS tools and AI services your business relies on, and request energy-mix and PUE (power usage effectiveness) data from each provider as part of your Scope 3 carbon accounting.
2. Favour renewable-region hosting. Many major cloud providers let you choose data centre regions — prioritise ones matched to grids with higher renewable penetration where functionality allows.
3. Set a target and timeline. Build digital emissions into your carbon reduction plan with a clear baseline year, using GHG Protocol or SBTi-aligned methodology so the target holds up to scrutiny.
4. Automate the tracking. Carbon management software that pulls utility, supplier and activity data together beats an annual spreadsheet exercise — especially as data centre disclosures keep changing.
5. Revisit annually. Grid carbon intensity, vendor transparency and regulation are all moving fast this year, so treat how to reduce carbon footprint as a live question, not a one-off audit.

Where carbon consulting and carbon management software fit in

This is exactly the kind of complexity that specialist carbon consulting exists to untangle. Turning fragmented, inconsistent hyperscaler disclosures into audit-ready SECR and ESG reporting isn't a job for guesswork — it requires carbon accounting software that can model activity-based and spend-based data side by side, and consultants who can sense-check the assumptions behind them.

Whether you're aiming for a genuine carbon neutral claim or simply trying to keep pace with tightening ESG reporting expectations, the direction of travel is the same: emissions that used to sit outside the boundary of carbon management — in your cloud stack, in your suppliers' industrial upgrades — are moving inside it. Building that into your carbon reduction plan now, with the right carbon management software and support, is considerably easier than retrofitting it once the reporting requirement catches up.

Curious what your own numbers would show? Book a free demo with Compare Your Footprint and see your Scope 1–3 emissions mapped out.