Wood Mackenzie published its 2026 levelized cost of electricity report on October 8 with a finding that should change how data center developers stage power. Four-hour battery storage now costs less than an open-cycle gas turbine in all 43 markets where both technologies were modeled. That is every continent, and the peaking plant is exactly what developers have leaned on to bridge the years between a signed interconnection agreement and an energized campus.
The turbine side of that trade is getting worse, not better. Gas turbine prices have been driven up by AI data center developers buying every model they can find, and an open-cycle turbine now takes two to four years to procure, according to TechCrunch. Waitlists for the more efficient closed-cycle machines run into the early 2030s. Rising fuel volatility does the rest. “This economic shift is decisive and widening,” Wood Mackenzie principal analyst Ahmed Jameel Abdullah said in the release.
The battery side is getting cheaper on a schedule. In the Middle East and Africa, four-hour storage sits at $120 per megawatt hour in 2026 and is forecast to fall 33 percent to $80 by 2035, displacing gas peaking across every gas market in the region. Utility-scale solar in the same region already leads globally at $37 per megawatt hour, headed below $20 in Saudi Arabia and the United Arab Emirates by 2033. China remains the cost benchmark, with storage more than 55 percent below the rest of Asia Pacific, which is what manufacturing scale looks like.
The waitlist is the constraint, and it moved to the batteries
For an operator, the useful reading is that a peaker was a scheduling decision as much as a cost decision, and the schedule just flipped. A gas peaker ordered today arrives in the early 2030s. A four-hour battery is a factory build with a shortening lead time and a falling price curve, and software decides when it discharges, so it can also arbitrage a grid that already pays for flexibility.
Two caveats worth carrying. Wood Mackenzie has solar under pressure in North America from tariffs and import restrictions, so the cheapest option in a US interconnection queue is not automatically the cheapest one to build this year. And four hours of storage covers a daily peak, not a multi-day wind drought or a stretch of high demand, so it displaces peaking plants rather than the firm capacity behind them.
Three questions for your own project. What duration is your grid operator actually paying for? If your peaker order sits in a queue until 2030, what does the campus do in 2028? And who owns the storage model now that the cheaper answer is a battery rather than a turbine?
Related reading. Texas now charges a large load for its place in the grid queue, and the 4,500 megawatt number Meta fought to keep off the record.
Get the next one before it is old news
Independent analysis of cloud-native infrastructure, Kubernetes and data center economics. No vendor spin.