At a quarry-lake site in Oregon, a 31 MW/62 MWh battery system developed by Calibrant Energy allowed a 72 MW Aligned Data Centers campus to connect to the grid without waiting for the utility to upgrade its network. With diesel alone, the campus would still be waiting for its connection. The project is one of the examples cited in a new S&P Global Energy study commissioned by Fluence, and it reflects a broader change in how data centre operators approach power. The report, “The Real Cost of Power,” concludes that battery energy storage systems (BESS) come out ahead of diesel, gas turbines and UPS-plus-supercapacitor configurations on a full lifecycle basis, in all the use cases it examined.
Three markets, one standard load
The analysis models a reference data centre with a 390 MW load supporting 300 MW of critical IT capacity over 25 years, in three contrasting markets: West Texas, Galway in Ireland and New South Wales in Australia. In each location, it tests three applications: backup power, flexible capacity and load smoothing. The comparison goes well beyond headline capital cost, taking in fixed and variable operating expenditure, augmentation, decommissioning, carbon exposure and non-financial factors such as speed to power and permitting risk. As the S&P Global authors put it, “the correct yardstick for comparison is not the cheapest equipment, but the best risk-adjusted lifecycle outcome.”
The smarter E Award 2026: Outstanding projects
The results are consistent across all three markets. In Texas, a standalone 390 MW/8-hour BESS for backup power has a lifecycle cost of $491 million in net present value terms, against $626 million for a diesel genset of the same rating, $794 million for a BESS-plus-diesel hybrid and $1.09 billion for a gas turbine. The pattern repeats in Ireland and Australia, where more reliable grids strengthen the case for storage further, since the batteries need to cover shorter outages. In Ireland, a standalone BESS comes in at $521 million NPV, against $672 million for diesel, $854 million for gas and $866 million for a hybrid. The Australian figures track closely, with BESS at $536 million and gas at $842 million.
Flexible capacity: where the gap widens
In the case of flexible capacity, the gap widens sharply. In Texas, a 240 MW/2-hour battery costs less than a third as much as a gas turbine and about a sixth as much as diesel. In Ireland, a 120 MW/2-hour BESS costs $136 million against $432 million for gas and $794 million for diesel. In Australia, the corresponding figures are $330 million for storage, $1.26 billion for gas and $2.65 billion for diesel. According to the report, the reason is that batteries separate the site’s instantaneous demand from its net draw on the grid, so operators can increase or reduce that draw to manage peaks, ramp rates and interconnection limits. “No other asset class can do this,” the authors note.
Spain – UNEF calls for more flexible rules on data centres
The results for load smoothing are much the same. Across all three markets, BESS costs roughly a third as much as a UPS-plus-supercapacitor configuration, with lifecycle NPVs of $83 million to $104 million against $295 million to $319 million. Emissions are the one exception here. Because batteries cycle more and draw from the grid, their footprint in this specific application is about 15 percent higher than the UPS alternative. In all the other cases, however, BESS is the lowest-emissions option, well ahead of diesel and gas. In the flexible capacity case in Texas, for instance, lifecycle CO₂ comes to about 2.9 million tonnes for storage against 5.1 million tonnes for gas and 5.3 million for diesel. The report adds that these figures use current grid intensities and are likely to fall further as power systems in Texas, Ireland and New South Wales decarbonise.
Policy tilts further toward storage
Policy is moving in the same direction. Ireland now requires new data centres to secure on-site generation or storage equivalent to their import capacity, and to source at least 80 percent of their annual electricity from additional renewables in Ireland. Texas has opened a Batch Zero interconnection process under Senate Bill 6, letting data centres use on-site resources, including batteries and load flexibility, to negotiate better connection terms. Australia is weighing comparable reforms covering ramping limits, fault ride-through and load smoothing, alongside mechanisms that would push data centres to contract firmed renewables and cover the transmission upgrades needed to serve them. At federal level in the US, the Investment Tax Credit for storage remains in place through 2033 under the One Big Beautiful Bill Act, with a phase-out running to 2035.
Europe has built the batteries, and now has to run them well
Some projects already combine several of these benefits. At xAI’s Colossus 1 data centre in Tennessee, a 150 MW/600 MWh battery provides backup during outages, sub-second voltage and frequency support, peak management and demand response participation from a single asset. In Ireland, batteries at Keppel’s DC 1 and DC 2 sites, managed by GridBeyond, are estimated to cut emissions by 240 tonnes of CO₂ a year. The S&P Global analysis treats each use case in isolation and therefore describes its own numbers as conservative. In practice, the report notes, “use case stacking capability can materially improve BESS project economics.”
A different test for developers
The authors are careful not to overclaim. Long-duration outage requirements, local interconnection rules and site-specific risk assumptions can still favour hybrid or thermal configurations, and batteries face their own exposure to import tariffs and global manufacturing constraints. But turbine lead times are now long, gas and diesel face tightening air quality rules, and winning local support is increasingly difficult on crowded urban sites. For developers trying to add AI capacity quickly despite growing grid and permitting constraints, speed to energisation, controllable demand and the flexibility to adapt later matter more than a decade ago. As the report itself frames it, the strategic question “is not whether BESS is the cheapest; it is whether BESS delivers the best full life value once timing, resilience, flexibility and risk are properly priced.” (hcn)
