The Battery Is Becoming the Fire Escape

Extreme heat and wildfire risk turn electricity from a monthly bill into a survival system. When the grid fails, the consequences are not abstract: refrigeration, cooling, communications, medical devices and water systems all become contingent on whichever piece of infrastructure did not fail first. The useful question is therefore less whether resilience is fashionable and more whether it arrives at the homes that need it before the next outage does.

Vermont’s Green Mountain Power offers one concrete answer: aggregate household batteries, solar, electric vehicles, commercial demand response and utility-scale storage into a virtual power plant. The utility says its programme now supplies about 110 megawatts—comparable to a small-to-average gas plant—and has effectively displaced its oil, diesel and gas peakers, though four peaking plants still operate occasionally at the regional grid operator’s request. More than 5,500 households have battery systems, and participants can keep backup power while the utility can call on stored energy for the wider grid. The BBC’s account is here.

That arrangement matters because it makes a resilience asset do double duty. A household battery can be private insurance during an outage and public capacity during a grid peak. One customer chose two leased batteries over a planned $12,000 gas generator, explaining: “I'd like to say I did it for the environment, but I did it for myself,” — Pollyanna Bladyka. That is not a failure of civic spirit; it is how durable public systems often get built in America: give people a direct reason to participate, then make the aggregate useful.

GMP’s model is not a magic trick. Batteries do not create electricity; software coordinates stored energy and reduces demand when the grid is strained. About half of the utility’s home-battery participants also have solar, according to GMP. The attraction is that capacity can be added in parcels rather than waiting years for a conventional plant, transmission line, permitting battle and ceremonial shovel with seven logos on it.

The economics are unusually legible. GMP offers two batteries for $55 a month on a 10-year lease, says it saved customers $11 million last year, and says seven of its 10 most damaging storms occurred in the past decade, causing more than $225 million in damage. Those are utility-provided claims, not independent verdicts, but they describe the pressure pushing this experiment along. The utility serves more than 75% of Vermont, wants to eliminate outages by 2030, and has seen more than 2,500 enrolments since regulators lifted a cap in 2023. Read the underlying report.

For hotter, fire-prone parts of the United States, the lesson is not that every resident should become a miniature power station with an app and a warranty dispute. It is that backup power needs to be treated as part of public-health infrastructure, with particular attention to renters, lower-income households and people whose health depends on uninterrupted cooling or equipment. Otherwise the grid becomes exquisitely flexible for those who can finance flexibility, while everyone else receives the traditional resilience plan: locate an extension cord and hope.

Scaling remains difficult. Customer enrolment costs money, wholesale markets can be hard to navigate, and utility incentives often favour building large assets over harnessing flexible demand. As former Energy Department official Jen Downing put it, “Utilities don't have much incentive to harness demand flexibility because they make more money if they just build stuff,” — Jen Downing. The federal upside is considerable but still prospective: a 2025 Energy Department report estimated 160 gigawatts of virtual-power-plant capacity could be unlocked by 2030. The hard part is not imagining the distributed grid. It is changing the rules that currently pay handsomely for the opposite.

The same grid that needs more distributed capacity is being asked to support an expanding digital industrial estate. A recent investigation found that most large Dutch data centres are not publicly disclosing their electricity and drinking-water use despite an EU reporting obligation for facilities with at least 500 kilowatts of installed capacity. Public figures covered electricity use at only 44 of 186 qualifying commercial Dutch centres and water use at 47. The reporting is here.

The Netherlands is not the United States, and its figures should not be smuggled across the Atlantic in a trench coat. But the conflict is recognisable. Dutch data centres used 5.1 billion kilowatt-hours in 2024, 4.6% of national electricity consumption, according to Statistics Netherlands; grid operator TenneT projects their share could reach 10% to 15% by 2030. Water consumption is estimated at only 0.1% of national use, yet that national average says little about whether a particular facility is drawing heavily during a local drought. The report links both electricity and water demand to cloud services and AI.

This is where climate resilience stops being a pleasant noun and becomes an allocation problem. Every community facing summer peaks, smoke-driven indoor cooling needs, drought or fire-weather outages should be able to see the major new loads competing for local power and water. Reporting rules without public reporting are merely transparency rendered in invisible ink.

Vermont’s virtual plant and the Dutch disclosure gap describe opposite versions of the same civic choice. One makes household capacity visible and usable when the system is stressed. The other leaves a growing institutional load difficult to measure in public. Resilience is not a bunker, nor is it an inspirational rendering of solar panels at dusk. It is the decidedly unromantic work of deciding who gets reliable power, who pays for it, and whether the public can inspect the answer before the thermometer delivers its audit.