In November, more than 50 student innovators at UC Santa Cruz spent two days developing artificial intelligence (AI) powered solutions for wildfire mitigation and climate resilience at a “Reboot the Earth” hackathon co-hosted by the United Nations and the Baskin School of Engineering. The student teams addressed one of two challenges developed by the Food and Agriculture Organization (FAO) and CAL FIRE, including smart water and irrigation planning under fire stress and AI-accelerated environmental review for wildfire mitigation.
One of the winners, Fire Oracle, used machine learning to accelerate prescribed burn planning. The AI was trained on NASA wildfire data, vegetation indices, soil moisture levels and geospatial information to predict high-risk wildfire locations.
Wildfires expand beyond the West
In 2025, the US had 51,000 wildfires, 5,000 more than the ten-year national average.
Wildfire risk has historically been viewed as a West Coast issue, but shifting climate and weather patterns are pushing that risk into new regions. In 2024, New York and Massachusetts experienced more than 2,500 wildfires. In Massachusetts in 2024, 4,262 acres burned in fires, a more than 3x increase from 2023.
As wildfire risk grows in the Northeast, utilities that haven’t traditionally faced large-scale fire threats are starting to adopt modeling and planning tools used in high-risk Western states.
National Grid has recently partnered with Rhizome, a climate resilience planning platform, to deploy AI technology that identifies and helps prevent wildfire risks across its electric transmission and distribution networks in Massachusetts, New York and the United Kingdom.
Casey Kirkpatrick, Director of Group Strategic Engineering at National Grid, said that utilities face a range of climate and weather-related risks—from wind and heat to icing. “We’re committed to understanding these risks holistically, and when we invest in resilience, we prioritize multi-value solutions that can mitigate several risks at once,” he said.
Sim City for the grid
Rhizome’s platform, gridFIRM was launched in 2024 and uses simulation to show how extreme weather events, including fire, affect utility assets. It also quantifies long-term risks of utility asset-related wildfires. Acting as a “Sim City for the grid,” the software allows utilities to simulate how fires and storms will impact their assets. It provides a picture of how to prioritize investments to make the grid resilient as climate change intensifies storms.
National Grid will use the platform to identify high-risk areas where utility assets could potentially ignite wildfires, quantify and prioritize wildfire risks across transmission and distribution networks, and develop cost-effective prevention and response strategies. The platform also expands the company’s broader grid-resilience planning capabilities.
Mishal Thadani, CEO and Co-Founder of Rhizome, noted that while National Grid’s service territory faces lower risk for widespread wildfires than the West Coast, “low risk” isn’t “no risk.”
“National Grid is looking to protect the communities they serve by ensuring that they invest in the safety of the grid in areas that are most prone to fire risk,” added Thadani.
Ingesting environmetal data
gridFIRM uses physics-based approaches and machine learning algorithms to compute the likelihood that a utility asset may spark and ignite a fire, and how that likelihood may change over time. It incorporates historical geospatial, weather, climate and vegetation datasets, models of climate impacts and data on utility assets, failures and ignitions across the grid.
Thadani said the platform improves as more environmental data and system records are ingested. “Importantly, gridFIRM also uses machine learning on historic investments to understand which types of investments will be the most effective in each vulnerable location,” he added.
National Grid Partners, the company’s strategic investment arm, has committed $100 million to deploying AI across operations as part of a broader effort to strengthen safety, reliability and long-term planning. “Investments and modeling tools that identify risk earlier and help us plan ahead, enhance network resilience and manage long-term costs,” said Kirkpatrick.
