Over the course of the 4-year, $11 million XPRIZE Wildfire competition to revolutionize the way we detect, manage, and stop wildfires before they become destructive, we’ve seen competing teams push the limits of innovation. Tasked with developing cutting-edge technology that will transform our approach to fighting increasingly destructive wildfires, teams are bringing breakthroughs from concept to scalable, real-world solutions.
Now, as we approach the final phase of the competition, 8 teams, representing 4 countries, have gathered at the New South Wales Rural Fire Service in Australia for the Space-Based Wildfire Detection & Intelligence Track Finals Testing. These teams are putting their solutions to the test and competing for the $3.5 million grand prize, presented by Minderoo Foundation.
As the final milestone before selecting winners, this represents the most competitive stage yet—showcasing top-performing teams under real-world conditions and serving as powerful proof that incentive competitions like XPRIZE can compress timelines, de-risk breakthrough technologies, and mobilize entire ecosystems to tackle urgent, real-world challenges.
Follow along as we share the latest developments from finals testing.
- Testing recap
- A shared vision for the future of wildfire management
- XPRIZE Teams Learn from Rural Fire Service Crews
- Testing in Real Conditions
May 19 | 3:41 pm PST
Testing recap
With XPRIZE Wildfire Space-Based Detection & Intelligence Finals Testing now complete, the competition has entered the judging phase for Track A. The judging panel is reviewing quantitative and qualitative measures, such as team outputs, reports, and data collected during the finals testing period to evaluate how each system performed in a real-world operational environment.
The Track A judging panel brought together experts across wildfire operations, remote sensing, geospatial technology, emergency response, space systems, AI, and environmental modeling. Their backgrounds span frontline fire management, satellite imagery, disaster response technology, bushfire science, national security, Geographic Information Systems (GIS), and advanced wildfire intelligence systems, giving the judging process both technical and operational depth.
Learn more about XPRIZE Wildfire’s Judges
The finalist teams brought a wide range of technical approaches to the challenge, including AI, machine learning, satellite imagery, physics-based modeling, fire spread forecasting, hyperspectral analysis, and low-Earth orbit satellite concepts. During finals testing, teams were encouraged to demonstrate as many of their capabilities as possible, from detection and characterization to reporting and data delivery.
Judges are evaluating teams’ solutions based on quantitative and qualitative measures, including speed, accuracy, quality, complexity/comprehensiveness, and actionability of observations. The assessment includes teams’ ability to detect, observe, characterize, and report on fires across New South Wales during the defined testing period.
- MyRadar uses patented miniaturized satellite technology, spectral imagers, onboard AI optimizations, and environmental alerting tools to support rapid detection and real-time situational awareness.
- DeepFire combines AI-based wildfire prediction and detection systems, using predictive risk maps to adjust confidence in detections and model possible fire behavior, including danger level, speed, and direction of spread.
- Fire Eye uses multi-source hyperspectral image analysis, ensemble models, cloud computation, ground control stations, and multiple communication channels to predict fires and share information effectively.
- Mayday.AI combines real-time global wildfire detection and monitoring with early-warning access and community-focused disaster response tools.
- Redback Fire Team uses a two-phase approach designed to detect wildfires quickly and characterize fire behavior through a hybrid model system that accounts for geographic, seasonal, and diurnal variation.
- Snuffed proposes a “string of pearls” constellation of micro-satellites, or CubeSats, in low-Earth orbit, designed to support rapid revisit times and early wildfire detection.
- SIRIUS Wildfire Alliance integrates Earth observations, AI, GIS, wildfire spread modeling, and ground and space-edge computing to support high-resolution wildfire nowcasting, forecasting, false positive detection, and resource allocation.
- Ember Guard fuses satellite data, atmospheric conditions, social media, deep learning, LLMs, and AI techniques to detect wildfires and forecast likely fire propagation and intensity.
Together, these teams represent the broader shift from wildfire detection to wildfire intelligence. The future of fire response will not depend on a single data source or model alone. It will depend on approaches that can combine observations, analytics, context, and confidence into information that fire agencies can use.
“Testing through the XPRIZE Wildfire Finals gave us a rare opportunity to validate Mayday.ai’s system in a true operational environment, under rigorous judging criteria and against the complex wildfire realities of New South Wales. While much of our prior operational experience has been in the Global North, testing in Australia exposed us to fuel profiles more representative of many Global South regions, where dead fuels, eucalyptus canopy, and low-radiative early ignitions can significantly suppress traditional thermal detection.
This environment helped validate one of our most important breakthroughs: combining cloud masking, smoke-first logic, and AI-based computer vision verification to detect vertical smoke columns from geostationary sensors positioned approximately 23,000 miles above Earth. This approach enables extremely early detection—often close to inception points—even when thermal signatures are weak or obscured, providing fire agencies with faster, more actionable intelligence under real-world conditions.”
– Kian Mirshahi (Mayday.ai)
As judging continues, Track A remains focused on one of the most important opportunities in wildfire response: helping agencies gain faster, clearer, and more reliable information that can be used to intercept fires before they become destructive.
May 01 | 1:46 pm PST
A Shared Vision for the Future of Wildfire Management
On April 15, XPRIZE Wildfire hosted a VIP event bringing together 50+ attendees across finalist teams, judges, sponsors, partners, academics, and leaders from the global wildfire community at the NSW Rural Fire Service (RFS) Headquarters for a VIP event in coordination with the Rural Fire Service and the Minderoo Foundation. The event provided insights into the Space-Based Detection & Intelligence Finals testing and showcased real-world solutions for wildfire detection and intelligence.
Teams held presentations which provided insights on how satellites can act as "sentinels" and artificial intelligence (AI) can provide "foresight," predicting how a fire will move before it gains strength. Teams were met by real-world validation from leading experts during a panel featuring Rural Fire Service experts and XPRIZE Wildfire judges.
Peter McKechnie, Deputy Commissioner of the NSW Rural Fire Service, spoke to how space-based detection will move the needle the same way that fire alarms in the home changed structural fire alerts, citing an urgent need for early detection. One attendee shared, “Satellites won’t replace firefighters; but, they can give them what they need most: time, precision and situational awareness. Day or night, regardless of weather, season or terrain.”
Overall, the event was an exciting moment that painted a clear image of how technology innovations will interact with traditional approaches and shape the future of wildfire management.
April 22 | 8:42 am PST
Where Innovation Meets Experience: XPRIZE Teams Learn from Rural Fire Service Crews
Last week, XPRIZE Wildfire teams, judges, and operations staff attended a planned hazard reduction burn in Wahroonga, New South Wales, Australia. Located 45 minutes outside of Sydney, Wahroona is a 16-hectare forest nestled within a residential area. Filled with Turpentine, Eucalyptus, and Ironbark trees, as well as native shrubs and small trees, controlled burns are conducted here to reduce the fuel supply in an area at high risk for wildfires and in close proximity to homes, businesses, and a hospital.
During the burn, teams learned about the methods and practices used by the Rural Fire Service for conducting prescribed burns and learned about active fire behavior, environmental conditions for prescribed fire, and fuel types. This was a unique opportunity to bring together technologists and firefighters to build a deeper understanding of how rapid, precise, and accurate satellite data can support wildfire management.
April 16 | 12:55 pm PST
From Space to Ground: Testing in Real Conditions
Since Finals started on April 9th, teams have been testing their solutions in real-world conditions, helping them refine and prove operational readiness. Based out of the NSW Rural Fire Service Headquarters in Homebush, Australia, teams are utilizing AI, machine learning and space-based sensing to detect and characterize fires. Weather conditions across New South Wales have offered ample opportunities for hazard reduction burns, and even bushfires in some areas.
Each morning begins with a briefing from experts at the NSW Rural Fire Service, who present teams with a real-world operational scenario to tackle based on current burn activity. Over the next 12 hours, teams operate their systems to detect and characterize all prescribed fires and brushfires across New South Wales, an area larger than 800,000 square kilometers, and must also determine false positives. Throughout testing, XPRIZE Wildfire judges, a panel of global leaders in wildfire innovation, observe teams on site, analyze data in real time, and evaluate the performance and reliability of each system."
For more information about XPRIZE Wildfire, visit xprize.org/wildfire.
XPRIZE Wildfire is supported by the Gordon and Betty Moore Foundation, Pacific Gas & Electric (PG&E), Minderoo Foundation, Lockheed Martin, Conrad N. Hilton Foundation, Costa Navarino, American Family Insurance, Fairfax Financial, the Roddenberry Foundation, and individual benefactors. The Minderoo Foundation is the presenting sponsor of the XPRIZE Wildfire Space-Based Detection and Intelligence Track. Testing is conducted in partnership with the New South Wales Rural Fire Service, and with software support from ESRI.