Batteries now beat gas peakers on cost in 43 markets. A battery still runs out after four hours.
Wood Mackenzie says a four-hour battery is now cheaper than an open-cycle gas turbine in all 43 markets where it modelled both. That is good news. The comparison is with a peaker that runs a few hours at a time, and a battery only delivers what something else has put into it.
Forty-three out of 43. In every market where Wood Mackenzie modelled both, a four-hour grid battery now costs less per unit of power than an open-cycle gas turbine, the consultancy said on 8 October 2026.
TechCrunch picked it up the next day under the headline that batteries are now cheaper than the gas turbines used at many data centres. AI data centre builders have been buying any turbine they can get, TechCrunch reports. Open-cycle units now take two to four years to arrive, and waitlists for the bigger closed-cycle units run into the early 2030s.
First, the two machines. An open-cycle gas turbine is a jet engine bolted to a generator. It starts fast and burns a lot of gas per unit of power, so utilities mostly run it as a "peaker" in short bursts on hot evenings. A four-hour battery stores power and gives it back over about four hours at full output.
Wood Mackenzie's numbers, from its release:
- Four-hour storage in the Middle East and Africa comes in at US$120 per MWh in 2026 and is forecast to fall to US$80 by 2035.
- The Asia Pacific average for grid storage is US$134 per MWh. China is more than 55% below that.
- Japan, Australia and the Philippines stay at a premium because of import duties, higher installation costs and local manufacturing policies.
- Single-axis tracking solar is the cheapest new power in 43 of 48 modelled markets.
Here is the part the headline leaves out. A gas turbine is a complete supply. Keep the gas coming and it keeps running, day or night, cloud or sun. A battery is a bucket. It holds four hours, and then it sits flat until solar, wind or the grid fills it again. Wood Mackenzie's public release does not say what price it assumes for the power that charges the battery.
So the fair reading is narrow. A battery beats a peaker at the peaker's own job, a few hours of top-up when demand spikes. It does not replace a plant that runs through a week of cloud.
Analogy: a turbine is a tap on the town water main. A battery is a rainwater tank. The tank is now cheaper than the plumbing for a new tap. It still holds only what the roof sends it.
Good: Cheaper storage is real, and it keeps getting cheaper. Turbines are scarce and getting dearer, so a battery is the faster thing to buy. For a business with solar on the roof, the battery is what moves daytime power into the evening.
Bad: The win is against peaking plants, not all-day supply. A four-hour battery sized for an evening peak is a different design from a site that wants to ride through several grey days. That second job takes more solar to fill the battery, more battery, or a generator in the mix.
Ugly: The release does not publish the charging cost assumption behind the storage figure, so we will not guess one. Australian buyers sit in Wood Mackenzie's premium group, so the global headline overstates how cheap storage is here. None of these sources gives a cost for any single site.
Soft next step: if you are weighing a battery for your site, start with two questions. How many hours after dark do you run, and what fills the battery on a cloudy week? Then a short consult at Sovryn Energy /consult if you want those answers read against your meter.
Source: Wood Mackenzie press release, 8 Oct 2026 - Four-hour battery storage now beats gas peaking on cost globally. Primary source for 43 of 43 markets, US$120 to US$80 per MWh (MEA), US$134 per MWh APAC average, China more than 55% lower, Japan/Australia/Philippines premium, solar cheapest in 43 of 48 markets, Ahmed Jameel Abdullah quote. Does not state the charging cost assumption. TechCrunch, 9 Oct 2026 - Batteries are now cheaper than natural gas turbines used at many data centers. Source for data centre turbine buying, two-to-four-year open-cycle lead times and early-2030s closed-cycle waitlists.