Typeost

Tech news in plain type

Latest / Business

Business

Datacenter Emissions Set to Climb 70% by 2030, IEA Warns

The International Energy Agency projects datacenter emissions rising 70% by 2030, with AI driving most of the surge in electricity demand.

According to The Guardian, the International Energy Agency has warned that datacenters worldwide will generate 70% more carbon emissions within the next four years. The outlet reports that the energy watchdog attributes much of this rise to the United States and China turning to fossil fuels to power their artificial intelligence growth plans.

The same report, as summarized by The Guardian, projects that datacenters will consume twice as much electricity in 2030 as they do today. The agency expects the global expansion of AI to account for roughly three-quarters of that increase in demand.

The Guardian's account frames the forecast as a warning rather than a settled outcome. It describes the IEA as the world's energy watchdog, and the projection rests on the agency's own modeling of how power is sourced as demand grows.

The material available for this story is limited to The Guardian's summary, so details such as the specific countries' energy mix, the methodology behind the figures, and any policy responses from governments or companies are not covered here.

The forecast lands at a moment when AI companies are racing to build larger computing facilities, and when power grids in several regions are already under strain. Whether the projected emissions materialize will depend on how quickly those facilities are supplied with cleaner sources.

Background

Datacenters are the physical buildings that house servers, storage and networking equipment used for cloud computing, online services and, increasingly, AI model training and inference. They run continuously and require large amounts of electricity both to power computing hardware and to keep it cool. For most of the past two decades, efficiency gains in chips, cooling and facility design kept total electricity use relatively flat even as internet traffic grew sharply. That balance has shifted as AI workloads, which rely on dense clusters of specialized processors, have expanded rapidly.

Electricity demand from computing is unusually difficult to plan for. A new facility can take years to permit and build, while the grid connections and generation capacity it depends on often lag behind. Utilities in several regions have reported long queues of large-load requests, and some operators have turned to on-site generation or to long-term power contracts to secure supply. The mix of sources feeding a grid largely determines the carbon footprint of any given facility. Where coal or gas sets the marginal supply, added demand translates directly into added emissions. Where wind, solar, hydropower or nuclear dominate, the same demand carries a smaller footprint.

Governments have approached the issue differently. Some have introduced reporting requirements for energy use at large facilities, while others have focused on attracting investment with incentives and faster permitting. Technology companies have made public commitments to match their consumption with renewable purchases or to reach net-zero targets, though critics argue that such accounting can obscure the real-time mix of power a facility draws. Energy agencies such as the IEA typically publish scenario-based projections, which depend heavily on assumptions about policy, technology costs and the pace of construction. Forecasts of this kind are useful for identifying direction and scale, but they are not predictions of exact outcomes.

Why it matters

The projection connects two of the most consequential trends of the decade: the rapid buildout of AI and the effort to cut greenhouse gas emissions. If datacenter demand roughly doubles, the question is not only how much electricity is needed but where it comes from. A rise in emissions of this size would make it harder for countries to meet climate targets that assume power grids are decarbonizing. It would also put pressure on utilities and regulators who must decide which new generation to build and how to allocate limited transmission capacity.

For ordinary households, the implications are indirect but real. Large new loads can affect wholesale electricity prices, grid reliability and the timing of infrastructure upgrades. Communities hosting major facilities often weigh tax revenue and jobs against water use, noise, land changes and the strain on local power systems. For businesses that rely on AI services, the energy picture may influence pricing and the availability of compute over time.

The forecast also raises a question about accountability. Emissions figures depend on where and when power is consumed, and on how companies account for renewable purchases. A clear, consistent measure of datacenter emissions would help policymakers and the public judge whether efficiency improvements and clean power deals are keeping pace with growth. The IEA's warning, as reported, suggests that under current trajectories they may not be.

Finally, the story underscores a tension within the technology industry itself. Companies promoting AI as a tool for solving large problems, including climate modeling and grid optimization, must also confront the energy cost of the systems they are building. How that tension is resolved will shape investment decisions, public trust and the policies that govern where new capacity is sited.

What to watch

Watch for how governments respond to the projection, particularly in the United States and China, which the report singles out. Policy moves on grid planning, emissions reporting for large facilities and permitting for new power plants will indicate whether the forecast is being taken as a call to act.

Pay attention to corporate energy deals. Long-term contracts for nuclear, geothermal, wind and solar power, as well as plans for on-site generation, will show how operators intend to meet rising demand. Independent verification of those claims will matter as much as the announcements themselves.

Track the efficiency side of the equation. Advances in chip design, cooling and workload scheduling could reduce the electricity needed per unit of computing, and whether they keep pace with the expansion of AI will strongly influence the final emissions picture.

Viewpoints

Commentary from our columns: three angles on this story.

The Rulebook privacy, safety and regulation

The projection from the International Energy Agency is a reminder that the rules for digital infrastructure have mostly been written around data and privacy, not around the electricity and water it consumes. Emissions from datacenters have usually been reported in aggregate, or through corporate sustainability pages that are hard to compare. If the agency's warning is borne out, regulators will need clearer standards: consistent measurement of energy and carbon at the facility level, transparent disclosure of how power is sourced, and planning rules that account for the grid impact of large new loads before they are approved rather than after.

Such rules need not be hostile to AI development. Reporting requirements can be designed so that operators with cleaner supply face fewer hurdles, and permitting can be tied to commitments that new capacity will be matched by new clean generation or storage. The hard part is avoiding a patchwork in which each jurisdiction measures differently and companies shift load to wherever the rules are weakest. Coordination across borders, even if imperfect, would make the numbers meaningful.

The public also deserves a seat in these decisions. Communities that host facilities bear the local costs of power lines, water use and land change, and they should know what is being built and what it will draw. Clear rules now would be far easier than retrofitting accountability onto a buildout that has already gathered momentum.

The Ledger the business angle

From a business standpoint, the IEA's forecast is a story about who pays for power and who gets to plan around it. Datacenter operators are investing heavily, and their returns depend on reliable, affordable electricity. A projected doubling of consumption means utilities, grid operators and generators will need to make multibillion-dollar decisions about transmission and new plants, often years before demand is fully proven. Those costs do not disappear; they get recovered through rates, tax incentives or contracts, and how that allocation is structured will determine who carries the risk.

There is also a competitive dimension. Companies that secure long-term clean power at predictable prices may gain an edge, while those dependent on spot markets or fossil-heavy grids could face rising costs or regulatory pressure. Carbon exposure is becoming a financial variable, not just a reputational one, and investors are beginning to ask how much of a company's compute growth is tied to emissions-intensive supply.

The question for the sector is whether the economics of efficiency and clean energy can keep pace with the economics of scale. If they can, the buildout may look sustainable on paper and in cash flow. If not, the cost of delay could land on consumers and public budgets.

The Optimist the hopeful reading

There is a hopeful reading of this forecast, and it starts with the word "projection." The International Energy Agency is describing a path that depends on choices not yet made. Power demand this large has historically pushed governments and companies to invest in new generation, and the same wave of investment can be steered toward wind, solar, storage, geothermal and advanced nuclear. Each new contract for clean power is a sign that the buildout and the energy transition do not have to be at odds.

Efficiency is also moving. Chip designers, cooling engineers and software teams have repeatedly found ways to do more computing with less electricity, and the pressure of rising costs gives them strong reasons to keep going. A warning like this one can focus attention on exactly the improvements that matter most.

The most encouraging outcome would be a sector that treats its energy footprint as a design goal from the start. The IEA's numbers are a call to action, and people across the industry, in government and in the public have the chance to answer it.

More in Business