The first FireSat satellites rode a SpaceX Transporter-14 rideshare into orbit this week, marking the start of a constellation that promises to shrink wildfire detection from hours to minutes. Backed by Google.org, the Moore Foundation, and the Environmental Defense Fund, the program plans to deploy more than fifty small satellites equipped with multispectral sensors and onboard machine learning processors. Each spacecraft can spot a fire as small as five meters by five meters — roughly the size of a classroom — and downlink the alert before the blaze spreads beyond a single acre.
How the Constellation Works
Traditional geostationary satellites such as GOES scan the Western Hemisphere every five to fifteen minutes, but their spatial resolution tops out at five hundred meters to two kilometers. Polar-orbiting sensors like VIIRS improve resolution to three hundred seventy-five meters but revisit any given spot only twice per day. FireSat changes the math by flying a flock of smallsats in low Earth orbit, each carrying a custom multispectral imager tuned to the thermal signatures of early ignition. The onboard processor runs a TensorFlow Lite model trained on labeled fire perimeters from years of agency data. When the model scores a pixel above the ignition threshold, the satellite compresses the detection into a fifty-kilobyte alert packet and beams it to a ground station within the same orbit pass. Latency from ignition to alert averages twelve minutes in early testing.
Open Data Changes the Economics of Fire Response
Perhaps the most consequential design choice is the data policy. Every FireSat detection will be published openly and free of charge to fire agencies worldwide. No subscription tiers, no licensing fees, no proprietary dashboards. The Moore Foundation and Environmental Defense Fund structured the grant agreements to require open distribution as a condition of funding. Google Cloud provides the machine learning pipeline that models fire progression once a detection lands, but the raw alerts and the forecast layers both flow through open APIs. For cash-strapped municipal fire departments and national agencies in the Global South, this removes the budget barrier that has kept satellite intelligence out of reach. Early partners include CAL FIRE, the Canadian Interagency Forest Fire Centre, and Australia Rural Fire Service.
What This Means for Founders Building Climate AI
FireSat represents a rare proof point that planetary-scale AI for public good can attract sustained philanthropic and commercial capital. The project combines three trends that founders should track: constellation-scale sensor deployment is now affordable for non-government actors, inference in orbit eliminates the bandwidth bottleneck that has plagued Earth observation for decades, and open data mandates can align with sustainable business models. Google Cloud gains a showcase workload for its Vertex AI edge platform. The Moore Foundation advances its wildfire resilience thesis. The satellite bus provider, Muon Space, secures a reference mission for its modular platform. Each stakeholder captures value without locking the data behind a paywall. For startups building in climate adaptation, the lesson is clear: the most defensible moats may come from operating the infrastructure that makes open data possible, not from hoarding the data itself.
Open Questions as the Constellation Scales
Several challenges remain before FireSat reaches its full fifty-plus satellite complement. The current launch cadence depends on rideshare availability, which can slip by months. Onboard model updates require uplink windows and validation cycles that are slower than cloud retraining. False positive rates in early testing hover around eight percent, driven mainly by industrial heat sources and solar reflection off water bodies. The team plans to reduce that below two percent through active learning loops where agency feedback retrains the model quarterly. Finally, the open data commitment will be tested when commercial Earth observation firms lobby for restrictions, arguing that free government-grade data undercuts their market. How the partnership navigates that pressure will set a precedent for every open science constellation that follows.