Battery Storage Safety: What Moss Landing Proves and Doesn't

On January 16, 2025, a fire broke out inside one of the largest battery installations in the world, at Moss Landing on California's Monterey Bay. Almost two years later, that single fire still drives much of the case against battery projects that look nothing like it.

The pattern shows up in local news. Voice of OC recently reported that a battery storage application in south Orange County was withdrawn after sustained local opposition. In upstate New York, a Lansing project is seeking a zoning variance. A Washington Post opinion piece called the wider backlash "a costly case of mistaken identity." That phrase holds up, and it also leaves out a few things. This piece separates what is documented from what is argued, and labels the guesses as guesses.

What happened at Moss Landing

Fact: The fire began in Vistra's Moss Landing Phase I facility, a 300-megawatt, 1,200-megawatt-hour system installed inside a converted turbine hall at a former gas power plant. Monterey County officials ordered roughly 1,200 nearby residents to evacuate, and a stretch of Highway 1 was closed. The fire burned for days, and a large share of the batteries in the building were destroyed.

Fact: This was not the site's first incident. Vistra's facility had overheating events in September 2021 and February 2022. PG&E's separate Elkhorn battery at the same complex had a fire in a Tesla Megapack unit in September 2022.

Fact: Researchers at San José State University's Moss Landing Marine Laboratories reported elevated levels of nickel, manganese, and cobalt in soil near the Elkhorn Slough after the fire. Residents' worries about fallout were based on real measurements.

Three things that made it unusual

Fact, then interpretation. Moss Landing Phase I differed from most current projects in three ways.

It was indoors. The batteries filled one large building. Newer projects mostly use separate outdoor containers with space between them. That spacing is designed to keep a fire in one enclosure from spreading to the next, while a shared room with shared air lets heat and gases travel. This is the main reason the Post's "mistaken identity" argument has merit.

It used NMC chemistry. Vistra's system used nickel manganese cobalt cells from LG Energy Solution. Most grid storage now uses lithium iron phosphate (LFP), which goes into thermal runaway at higher temperatures and releases less oxygen as it fails. LFP cells can still burn. They are harder to set off and tend to burn less intensely.

It was enormous and concentrated. The building held 1,200 megawatt-hours of energy at one site. A typical community-scale project facing a planning board is a fraction of that size. For a sense of how small distributed systems can be, see how Australia's community batteries actually work.

Interpretation: Using Moss Landing to oppose a 20-megawatt outdoor LFP project is like using a refinery explosion to oppose a gas station. Both involve fuel. Very little else carries over.

What the failure data says

Fact: The Electric Power Research Institute (EPRI) maintains a public database of battery storage failure incidents. Its analysis found that the global failure rate for grid-scale batteries, measured per gigawatt deployed, fell by about 97 percent between 2018 and 2023. Deployment grew rapidly over that period while the number of incidents stayed roughly flat.

Fact: Two standards account for much of that improvement. NFPA 855 sets installation rules, including spacing and fire protection requirements. UL 9540A is a test method that measures whether a fire in one cell or module spreads to its neighbors. Many US jurisdictions now require compliance with both before they issue permits.

Interpretation: A 97 percent drop in the failure rate does not mean a 97 percent drop in local risk. Residents do not live next to a global average. They live next to one specific site. The numbers still point clearly one way, though. Batteries built to current standards fail much less often than those from the 2018 generation, and Moss Landing Phase I came online in 2020, right at the edge of that shift.

Where the critics have a point

The "mistaken identity" framing is useful, but it can lead people to ignore complaints that are legitimate.

Emergency planning often trails construction. Volunteer fire departments in rural areas, which are often where these projects end up, may not have trained on lithium fires. The usual response is to let the unit burn out and cool nearby equipment. To a resident watching, that can look a lot like doing nothing. When developers explain that strategy only after a fire has started, they lose public trust quickly.

Air and soil monitoring is not standard. The Moss Landing metals findings came from academic researchers, not from a required monitoring plan. Neighbors who ask for baseline sampling before construction are asking for the same thing a careful engineer would.

The existing fleet keeps its original design. New standards apply to new projects. Older indoor and NMC installations remain in service, and regulators have been slower to address that equipment than to approve new sites.

Interpretation: Opponents tend to win their arguments about process and lose their arguments about physics. A county that requires a written emergency response plan, funds training for local firefighters, and mandates baseline environmental sampling answers most of the reasonable objections. Banning the technology outright answers the unreasonable ones and raises costs for everyone else on the grid.

Why delay costs money

Fact: The US Energy Information Administration expected developers to add about 18 gigawatts of utility-scale battery capacity in 2025, close to double what was added in 2024. Batteries are now the default way to move cheap midday solar power into the evening peak.

Interpretation: When a project is withdrawn, the need for that capacity remains. The utility fills the gap with a more expensive peaker plant, an import contract, or a project somewhere else that goes back into the interconnection queue. As we covered in why four-hour batteries now beat gas peakers on cost, the cheaper option is increasingly the battery. Ratepayers cover the difference when it gets blocked, even though the cost never shows up as a line item on their bills.

What to watch next

Prediction: Over the next two years, more local ordinances will move from outright moratoriums toward conditional approval. Expect required emergency plans, setbacks tied to UL 9540A results, and funding for fire departments to become common terms in permits. The south Orange County withdrawal looks more like the end of the moratorium approach than the start of a trend.

Prediction: The next serious incident, when it happens, will most likely involve an older indoor or NMC installation. If so, the useful policy question will be what to do with the legacy fleet, not whether to approve new LFP containers.

Prediction, with less confidence: Insurers may move faster than legislators. If underwriters start pricing legacy designs as a separate risk class, the market could retire some of that equipment before any regulator requires it.

The Moss Landing fire was real, and so was its fallout. Taking it seriously means applying its lessons to the projects that share its design, not to every battery project that comes before a planning board. For the bigger picture on where these systems fit, our complete guide to energy storage and the grid covers the rest.