XPRIZE Wildfire names its $11 million awardees
Anduril's Lattice platform took the $1.2 million top prize as XPRIZE named the awardees of its $11 million Wildfire competition, paying out $4.55 million of the purse.
XPRIZE has named the awardees of its $11 million Wildfire competition, and the largest single award of the final round - $1.2 million - went to Anduril Industries for a software platform that carries a wildfire all the way from detection to suppression across cameras, sensors and drones.
Two tracks, two very different problems
The competition, launched in 2023, split wildfire in half. Track A asked teams to fuse space-based Earth observation with artificial intelligence to detect, characterise and report fires faster and more accurately across vast landscapes. Track B asked something much harder: could a team autonomously spot a fire and put it out completely within ten minutes of ignition, across 1,000 square kilometres of Alaskan wilderness, without disturbing decoy fires placed nearby?
Track B was settled in the field in June, and XPRIZE announced the results on 23 September at an awardee ceremony at the New York City Fire Museum. Anduril Industries took first place and $1.2 million for Lattice, its modular software platform, which provided an integrated detection-to-suppression pathway linking cameras, sensors and drones. Dryad Networks came second with $800,000 for Silvanet, a persistent field network of 170 sensors working with its Silvaguard drones, which the judges credited with end-to-end autonomy. AURA Foresight took third and $500,000 for autonomous, persistent detection that cleared the critical ten-minute gate.
Anduril also collected the Lockheed Martin Accurate Detection & Intelligence Bonus Prize, worth $1,000,000, for rapid and precise multi-sensor detection performance.
Detection is getting better. Suppression is still the wall
Track A awarded $1.05 million across five teams for advances in multi-sensor integration, detection methodology, commercial readiness, operational capability and algorithmic innovation. SIRIUS Wildfire Alliance won the $500,000 System Integration award for an end-to-end methodology that integrates synthetic aperture radar with optical data to support fire characterisation and spread modelling. Team Snuffed took $250,000 for Methodological Innovation, with a probabilistic data-fusion framework built to be transparent, auditable and honest about its own uncertainty. Three $100,000 special recognitions went to MyRadar for commercial viability, DeepFire for an operational fire-spread alert capability, and Mayday.AI for retargeting an algorithm built to find volcanic ash at wildfire smoke underneath cloud cover.
Add the final-round awards together and they account for $4.55 million of the $11 million purse. The judges' own language is careful, recognising distinct breakthroughs and meaningful advances rather than a finished solution. XPRIZE says the Track A results helped identify key technical gaps that must be addressed before reliable, real-time wildfire detection from space is possible, and that the judges picked winners on operational capability as much as on clever algorithms.
The competition now enters an Impact Phase aimed at turning both tracks into deployable tools through field trials, technology evaluation and knowledge sharing, with a focus on high-risk wildfire regions. PG&E, a co-title sponsor alongside the Gordon and Betty Moore Foundation, leads that phase. Minderoo Foundation is the presenting sponsor and co-designer of the space-based detection track, Lockheed Martin sponsors the bonus prize, and the Conrad N. Hilton Foundation, Costa Navarino, American Family Insurance, Fairfax Financial and the Roddenberry Foundation are supporting sponsors.
Our opinion
The number worth staring at is not the $1.2 million. It is the $6.45 million of the purse that the final round did not hand out. A competition that launched in 2023 with $11 million on the table and closed its last round paying $4.55 million has told you something the press release will not: the ten-minute suppression bar was the honest part of the brief, and nobody cleared it well enough to take everything. That is a feature, not an embarrassment. Prize purses that pay out on delivery are worth more than purses that pay out on effort.
Read the two tracks together and you can see where the difficulty actually sits. Track A handed $1.05 million to teams doing the work satellites and machine learning already do tolerably well, which is why five of them qualified for money. Track B needed hardware that finds a fire, decides what to do, and then physically ends it without human hands, and it rewarded three. Anduril won it with integration software rather than anything that puts water or retardant on a hillside, which is a fair reflection of where the industry is: the sensing layer is close to solved, the actuation layer is not, and the company that wins is the one that can wire the two together. Dryad's 170 sensors are the more interesting result for anyone who will never own a drone fleet, because a persistent mesh that survives a season in the bush is the unglamorous half of the problem.