AI Data Centres Fuel a US Gas Revival: What It Means for India

The News

Gas is reclaiming its place in American power planning, and the appetite of AI data centres is a big reason why. A report released last week by the nonprofit Environmental Defense Fund and research firm Atlas Public Policy finds the US is now building more fossil fuel infrastructure than at any point in a decade, even as renewable growth cools.

Drawing on US Energy Information Administration data, the authors show fossil fuel plants rising from just 5 percent of all planned new capacity in 2024 to 30 percent by the second quarter of 2026. Planned fossil capacity would reach 68 gigawatts by 2030, a 45 percent jump on construction plans from only six months earlier. If the trend holds, fossil fuels could account for 70 percent of the development pipeline by the end of the decade.

Renewable projects are also being cancelled faster: 4.9 percent of them, against 2.9 percent of fossil projects. That is happening even though new utility-scale solar and wind still costs $68 to $69 per megawatt-hour on average, compared with $90 per megawatt-hour for new combined-cycle gas plants.

Why It Matters

The shift is a policy choice layered on top of a demand shock. US electricity demand is growing at its fastest pace in more than a decade as data centres, factories and electric vehicles connect to the grid. The Trump administration has favoured fossil fuels to meet that load, using stop-work orders, permit blocks and the removal of tax incentives against wind and solar. Only two years ago, in 2024, solar made up 81 percent of added US capacity and wind and solar together out-generated coal for the first time.

For the AI industry, the lesson is that compute is now an energy story. Hyperscalers that once marketed renewable-backed campuses are signing up to whatever power arrives fastest, and gas turbines can be sited quickly. The report warns that if the pipeline is fully built, it could lock in higher prices and pollution for decades, and that many plans may still collapse if AI load forecasts prove inflated.

Indian Angle

India faces the same collision between AI ambition and power supply, with a different fuel mix. Coal still generates the bulk of Indian electricity, and the country has committed to 500 GW of non-fossil capacity by 2030. Operators such as Yotta, AdaniConneX, CtrlS and NTT are expanding data centre capacity in Mumbai, Chennai, Hyderabad and Noida, while the IndiaAI Mission is subsidising GPU access for domestic model builders such as Sarvam.

The US cost figures are a useful benchmark. If renewables remain cheaper than new gas even in America, Indian data centre developers have a strong commercial case for round-the-clock renewable contracts through open access and green energy procurement rather than captive gas or diesel backup. India imports much of its LNG, so the gas price volatility that the report highlights would hit rupee-denominated operating costs harder here than in the US.

For investors, the takeaway is to scrutinise power-purchase terms when evaluating listed data centre plays, power utilities and AI infrastructure startups. Energy cost and availability, not just GPU supply, will decide which Indian AI ventures can scale profitably.

FAQ

Who published the report?

The analysis comes from the Environmental Defense Fund, a US nonprofit, and research firm Atlas Public Policy. It draws on US Energy Information Administration data on planned capacity, cancellations and costs.

How much fossil capacity is planned in the US?

Planned fossil fuel capacity would reach 68 gigawatts by 2030, a 45 percent increase on plans from six months earlier. Fossil projects rose from 5 percent of planned new capacity in 2024 to 30 percent by mid-2026, and could reach 70 percent by the end of the decade.

Is gas power cheaper than renewables?

No, according to the report. New utility-scale solar and wind average $68 to $69 per megawatt-hour, while new combined-cycle gas plants average $90 per megawatt-hour. Gas also carries fuel price volatility and pollution costs.

What does this mean for Indian data centre investors?

It underlines that power is becoming the binding constraint on AI infrastructure. Indian investors should examine how data centre operators source electricity, how exposed they are to imported fuel prices, and whether renewable contracts are locked in for the long term before backing capacity expansions.

This story was reported by The Verge. Read the full original coverage at The Verge.