The first three operational satellites in the Google-backed FireSat constellation reached orbit on July 7, 2026, launched aboard a SpaceX Falcon 9 from Vandenberg Space Force Base as wildfire smoke from more than 900 active blazes in Canada blanketed cities across North America. The microsatellites, built by California-based Muon Space, carry multispectral imagers capable of detecting fires as small as 5 by 5 meters — roughly 16 by 16 feet — through smoke and clouds. Google has put more than $15 million into the program, and the Bezos Earth Fund has committed $26 million.
After a three-month checkout, the satellites will begin feeding data to fire agencies in California, Colorado, Australia, and Portugal, covering every fire-prone region on Earth at least twice per day. FireSat is managed by the nonprofit Earth Fire Alliance and is the first satellite constellation purpose-built for wildfire detection rather than repurposed from other Earth-observation missions.
The full constellation is planned at more than 50 satellites, targeting hourly global revisits by 2029 and 20-minute revisits by the early 2030s. That cadence matters: existing weather and land-imaging satellites typically miss small ignition events, which grow exponentially in the first hours before crews arrive. A FireSat Protoflight satellite launched in March 2025 collected more than 1 million images and demonstrated it could pick out low-intensity fires that other orbital sensors missed entirely.
Key facts
- 01Three FireSat satellites launched aboard a SpaceX Falcon 9 from Vandenberg on July 7, 2026, marking initial operational capability.
- 02Google has committed over $15 million and the Bezos Earth Fund $26 million to the Muon Space-built constellation.
- 03Each satellite can detect fires as small as 5 by 5 meters through smoke and clouds using multispectral imaging.
- 04The full 50+ satellite constellation, targeted for the early 2030s, will deliver global imagery every 20 minutes.
- 05Earth Fire Alliance projects hourly revisits could save $1 billion in damages and prevent 22 million tons of carbon emissions.
Google Research plans to run the operational FireSat data against historical imagery using its own AI models, both to confirm very small fire detections and to feed predictive models of fire spread. Google framed the launch as another step in applying practical AI to climate resilience.
“another tangible step forward in putting practical AI to work for climate resilience.”— Google, Statement from Google
The Earth Fire Alliance has estimated that even at an hourly revisit rate, FireSat could save more than $1 billion in fire damage costs, prevent nearly 22 million tons of carbon emissions, and protect 3,500 homes and 1.3 million acres of land per year. Those figures assume detections translate into faster suppression, which in turn depends on ground and air resources being available — a constraint that has grown tighter each fire season.
The launch lands against a difficult backdrop for Google's climate story. The company's electricity usage grew 37% in 2025 as AI workloads scaled, and new US natural gas projects tied to data-center demand could collectively emit more than 129 million tons of greenhouse gases per year. Google has acknowledged the gap between its clean-energy procurement and its rising load, making FireSat one of several visible climate investments running alongside a much larger carbon footprint on the AI side of the ledger.
The immediate case for faster detection is unfolding in Canada's boreal forests. The Canadian Wildland Fire Information System counted nearly 900 active wildfires as of July 17, with more than 3,600 fires to date this year and 6.6 million acres burned. Smoke from those fires has pushed hazardous air pollution over more than 100 million people across Canada and the United States. Two of Canada's most destructive wildfire seasons occurred in 2023 and 2025, and the last three seasons rank among the 10 worst on record.
Detection is only one piece of the response. Fighting fires in remote boreal terrain requires fixed-wing air tankers and heavy-lift helicopters, and Canadian provinces have historically shouldered those costs alone. This year the federal government leased 10 additional aerial firefighting aircraft as surge assets, but dozens of blazes are currently classified as out-of-control and monitored rather than actively suppressed — a rationing decision fire agencies increasingly have to make. FireSat imagery of the Nipigon 6 fire in Ontario on June 15, 2025 showed active flames, fresh burn scars, and older scars in a single infrared pass, the kind of layered situational data ground crews rarely get in near real time.
The limits are worth naming. Detecting a fire earlier only reduces damage if agencies have the aircraft, crews, and clearance to reach it, and satellite cadence at initial capability is still twice-daily rather than the 20-minute goal. AI-assisted classification of small heat signatures also carries false-positive risk in landscapes with industrial heat sources, prescribed burns, and existing scars — a workload Google Research will have to prove out against operational data over the next several fire seasons.
For the AI industry, FireSat is a useful test of the argument that compute-heavy climate applications can plausibly offset some of the emissions from the data centers running them. The math is not close today: 22 million tons of prevented carbon is a real number, but it sits against a multi-hundred-million-ton annual footprint from AI-driven power buildout. What FireSat does prove is that the same satellite economics and model infrastructure that make consumer AI cheap can be redirected at a concrete public-safety problem — and that the frontier labs' willingness to fund those redirects, at meaningful nine-figure scale, is now part of how the sector defends its license to keep growing.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
