The International Energy Agency (IEA) has warned that global data centers will increase carbon emissions by 70% by 2030. The report's central finding is that the US and China are turning back to fossil fuels for power generation to meet their AI growth targets. This directly contradicts the "clean AI" narrative that major tech companies have repeated for years.

What happened?

According to the IEA's assessment, data centers will consume twice as much electricity in 2030 as they do today, with three-quarters of that increase driven by the growth of AI infrastructure. The agency projects that this demand will largely be met by natural gas and coal plants in the US and China, pushing data center-related carbon emissions up 70% within four years.

Why it matters

Over the past three years, Microsoft, Google, Amazon, and Meta have announced their AI investments alongside nuclear energy deals, solar and wind power purchase agreements, and "net-zero carbon by 2030" pledges. Microsoft's deal to restart a shuttered nuclear reactor and Google's investments in small modular reactors (SMRs) are the best-known examples of this showcase. But bringing these projects online takes years, while data center demand multiplies within months. That gap is being filled by gas turbine plants in several US states and by China delaying its target to reduce coal's share of power generation. The result is a widening gap between corporate sustainability reports and the grid's actual energy mix. This isn't a new contradiction — it's a decade-old issue from the growth of cloud computing that has sharpened under AI.

This picture also reveals the priorities of the AI race. To avoid falling behind competitors in model training and inference capacity, companies are choosing energy supply based on speed, with carbon footprint taking a back seat. That's why the IEA's warning matters: the body that shapes energy policy is now formally confirming a reality that doesn't match the story the industry has been telling about itself.

How does this affect energy-importing countries like Turkey?

Data center demand in the US and China directly affects global energy markets. If both countries increase natural gas and LNG imports, global LNG prices — which Turkey also depends on — come under upward pressure, a chain that feeds into electricity generation costs and ultimately consumer bills. In addition, demand for AI hardware continues to create shortages in GPU and advanced-node chip production (such as TSMC capacity), which drives up hardware costs for cloud, gaming, and enterprise IT investments in Turkey.

PartyCommitmentActual practice
US Big TechNet-zero via nuclear/renewablesRapid capacity growth via gas turbine plants
ChinaPhased coal exitDelaying coal-share reduction amid AI demand
Energy-importing countries—LNG price pressure, rising chip costs

What's next?

Whether these figures are updated in the IEA's next energy reports will be a gauge of whether AI demand is truly growing at the pace forecast. The rollout timeline for nuclear/SMR projects from companies like Microsoft and Google is also critical — if these projects are delayed, fossil fuel dependence becomes structural rather than temporary. Tracking how corporate "carbon neutral" pledges are revised in annual sustainability reports will show whether the gap between rhetoric and reality keeps widening.

Ultimately, AI companies' green energy showcase doesn't mask the grid's day-to-day reality: demand is fast, clean capacity is slow, and the gap is being closed with fossil fuels. For energy-importing countries, this isn't a distant industry debate — it's a cost item that flows directly into electricity and hardware bills.

For more information, see IEA resources.