The economics of powering the artificial-intelligence boom have tipped further toward batteries. A new Wood Mackenzie analysis finds that four-hour battery storage is now cheaper than open-cycle gas turbine "peaker" plants in all 43 markets the energy consultancy modeled, on every inhabited continent — a reversal the firm says is being driven largely by AI data centers' voracious appetite for electricity.
"This economic shift is decisive and widening," Ahmed Jameel Abdullah, a principal analyst at Wood Mackenzie, said of the findings, pointing to gas turbine shortages and fuel-price volatility pushing peaking costs up even as battery manufacturing costs keep falling.
The Numbers Behind the Shift
For projects reaching commercial operation in the United States this year, Wood Mackenzie told Utility Dive that four-hour storage runs 65% to 75% cheaper than new open-cycle gas peakers, depending on whether state carbon pricing applies. The firm did not publish exact dollar figures. In the Middle East and Africa, it expects four-hour storage costs to fall 33% to about $80 per megawatt-hour by 2035, effectively pushing gas peakers out of the market there on cost alone. Energy storage costs in China now run more than 55% below the rest of the Asia-Pacific region, the report found.
Gas is moving the opposite direction. Wood Mackenzie projects gas turbine prices will reach $600 per kilowatt by the end of 2027 — a 195% increase since 2019 — and the three dominant manufacturers, GE Vernova, Siemens Energy and Mitsubishi, are now sitting on combined order backlogs ranging from 35 to 116 gigawatts apiece. Open-cycle turbines that once shipped in a year or two now take two to four years to procure; waitlists for more efficient closed-cycle units stretch into the early 2030s.
Data Centers Are the Root Cause
The squeeze traces directly back to the AI buildout. Developers racing to bring new data center capacity online have been buying up gas turbines directly rather than waiting on utility procurement cycles, according to reporting from TechCrunch, driving up prices for everyone else who relies on turbines, including utilities that use them as peaking plants during periods of high demand. Wood Mackenzie says North American gas capacity investment is now entering a supply deficit that could last through the late 2030s, driven mainly by data center load growth that keeps thermal capital costs elevated. That strain is a contributor to the rising electricity prices utilities across the US have been passing on to ratepayers this year.
Solar remains the cheapest new power source overall — the report found single-axis tracker solar leads in 43 of 48 modeled markets, with onshore wind ahead in the other five — but U.S. tariffs and import restrictions on panels are pushing component prices up roughly 5% a year through 2030, with steeper increases forecast for residential and commercial installations. About 168 gigawatts of utility-scale solar capacity is shielded near-term by safe-harbor provisions tied to tax credits for projects that break ground or finish before the end of 2027.
What Comes Next
Wood Mackenzie expects the advantage to keep widening in the near term, helped by federal storage tax credits, before costs tick up after those credits begin phasing out starting in 2038. Even so, the firm projects storage costs will fall another 10% by 2060 as battery chemistries improve and domestic supply chains mature. For data center developers facing waits of up to four years just to get a grid interconnection, the report's authors say the near-term path increasingly runs through pairing gas generators with batteries as a bridge — using storage to absorb demand spikes and spare turbines from the kind of frequent ramping they weren't built for, rather than betting on new turbine capacity that may not arrive before the next generation of AI infrastructure is already built.
