Predicting the flame: how AI is reshaping our response to wildfires

As summer approaches and temperatures rise and vegetation dries in the northern hemisphere, communities, governments, and businesses are once again bracing themselves for wildfire season. Though a natural feature of many ecosystems, the frequency, scale, and impact of wildfires are intensifying. And while summer has yet to fully tighten its grip on the north, 2025 has already shown us just how bad things could get.

Though a natural feature of many ecosystems, the frequency, scale, and impact of wildfires are intensifying

In January, a series of devastating wildfires engulfed Los Angeles and San Diego County. Fanned by hurricane-strength winds and intensified by drought, the fires claimed at least 30 lives¹, displaced hundreds of thousands², and destroyed over 16,000 structures. The Eaton and Palisades fires alone were among the most destructive in California’s history³—and serve as stark reminders of a grim global trend.

Watch our video on tackling wildfires using AI in collaboration with the Financial Times

Between 2001 and 2023, annual global tree cover loss due to wildfires increased fourfold, from less than 2.5 million hectares to more than 10 million hectares.⁴ Wildfires degrade soil, pollute water, reduce biodiversity, and release vast quantities of carbon into the atmosphere. On average, they cost the global economy over USD 50 billion a year.⁵ Some projections suggest wildfire incidence could rise another 50% by the end of the century.⁶

These trends are both consequences and drivers of two of the critical pain points driving today’s global systems change: climate change and nature degradation. As droughts intensify and vegetation dries out, fires become more likely. When they burn, wildfires release carbon, destroy natural carbon sinks, and erode the resilience of ecosystems. Mitigating wildfire risk is, therefore, essential to accelerating the systems change we need around decarbonisation and nature regeneration.

Mitigating wildfire risk is, therefore, essential to accelerating the systems change we need around decarbonisation and nature regeneration

However, wildfires are fast-moving and complex. They often erupt in remote terrain and spread rapidly, forcing responders to contend with limited visibility, fragmented information, and severe time constraints. To manage wildfire risk effectively, then, we must embrace the ongoing system transformation driven by the digital economy. To that end, researchers, emergency services, and private companies are increasingly turning to artificial intelligence (AI) and machine learning to help detect, predict, and manage wildfires.

Read more: Economic systems and wildfires | Lombard Odier

Artificial wildfire intelligence

Any fire needs three basic ingredients to ignite and sustain itself: heat, oxygen, and fuel, known collectively as the fire triangle. When environmental conditions align—such as heat from a lightning strike, fuel from dry vegetation, and oxygen from strong winds—the tiniest spark can result in catastrophe.

Understanding how such conditions cause and influence wildfires is, therefore, crucial. But that understanding must be timely, precise, and localised, and here traditional monitoring techniques often fall short. This is where AI and machine learning can help.

An AI is a computer system capable of performing tasks that typically require human intelligence, such as recognising images, processing language, or making decisions. Machine learning is an approach to enabling AI that involves training algorithms to detect patterns in data. For wildfires, such data include weather information, satellite imagery, vegetation maps, and historical fire data, alongside human factors that make wildfires more probable, like road access, land use, and population density. Machine learning increases with each new data point, improving AI performance with each wildfire season and incident.

Together, these technologies offer transformative potential in three key areas of wildfire management: forecasting, detection, and modelling.

Technology offers transformative potential in three key areas of wildfire management: forecasting, detection, and modelling

Fire vs. data

AIs trained on historical and real-time weather data can forecast wildfires by monitoring environmental conditions that influence the fire triangle. Satellites play a central role in such efforts. Using imagery from sensors like NASA’s Visible Infrared Imaging Radiometer Suite (VIIRS) and Moderate Resolution Imaging Spectroradiometer (MODIS), AIs can monitor factors like vegetation dryness, atmospheric aerosols, and thermal anomalies, identify high-risk zones, and trigger alerts before fires begin.⁷ In Turkey, the FireAid initiative’s AI-powered wildfire risk map can predict wildfires 24 hours in advance with 80% accuracy, helping emergency services prepare and respond more effectively.⁸

Once a fire breaks out, detection becomes a race against time. AI-enabled camera systems mounted on towers or mountaintops—like the several hundred Pano has deployed in the US—can scan vast areas, identifying tell-tale signs of fire such as smoke or heat shimmer. In some cases, these systems have alerted authorities before any human detected the fire.⁹

related content.

rethink sustainabilityNatural Capital

rethink sustainabilityNatural Capitalnet zerosustainable business

From policy to ‘pain points’ – Transition Investment Summit explores today’s sustainable investment landscape

May 20, 2025

system changes.

learn more

Other detection methods use chemical sensors to identify gases emitted by fires—such as carbon monoxide or hydrogen—even in poor visibility, offering an advantage over visual systems. In California’s Jackson Demonstration State Forest, for instance, the California Department of Forestry and Fire Protection (CALFIRE) partnered with tech company Dryad on the installation of such sensors, making it possible to spot fires while they’re still small enough to extinguish quickly.¹⁰

Satellites also contribute at the detection stage. For example, the FireSat initiative, a collaboration between public and private partners, is building a global network of AI-enabled satellites capable of spotting fires as small as 5 m² with imaging updates every 20 minutes.¹¹ These systems are especially useful in remote or inaccessible terrain with limited ground-based detection.

AIs can also help simulate a fire’s likely trajectory and speed with generative models trained on satellite imagery, weather patterns, topography, and vegetation types. Utilities and fire agencies are already using platforms like TechnoSylva’s to help decide where to deploy crews, when to issue evacuation orders, and how to protect infrastructure.¹²

Together, such tools are transforming wildfire management. By enabling earlier intervention and more strategic decision-making, AI has the potential to reduce risk, limit destruction, and save lives.

Igniting innovation

Despite the growing promise of AI and machine learning in wildfire management, several challenges stand in the way of broader adoption and impact.

One of the most immediate is cost. AI-enabled camera systems and sensor networks can require significant upfront investment and ongoing maintenance—Pano’s camera stations, for instance, each cost around USD 50,000 per year to operate. Meanwhile, public wildfire budgets remain heavily skewed towards response rather than prevention, making it difficult to secure funding for preventative systems.

By enabling earlier intervention and more strategic decision-making, AI has the potential to reduce risk, limit destruction, and save lives

Another barrier is data. Variations in terrain, climate, and land use limit the utility of one-size-fits-all models, and the vast volumes of high-quality, localised data machine learning systems require are often sparse or inconsistent. For such AIs to work effectively, we need more local monitoring, standardised data collection, and open collaboration between public, private, and academic stakeholders. Initiatives like FireAid, which has made its source code and training data publicly available,¹³ represent a step in the right direction.

Integration is another key concern, with many emergency services and government bodies relying on older digital infrastructure that may be incompatible with AI platforms. The result is a fragmented ecosystem that can hinder coordination. Improving interoperability through shared standards, interfaces, and training will be essential to fully realising the benefits of AI in fighting wildfires.

There are also issues of trust and transparency. Many AIs function as ‘black boxes’ that offer users scant insight into the basis of their outputs, which can limit confidence and slow adoption. Developers are responding with more transparent models and human-in-the-loop systems that allow AI to augment, rather than replace, expert judgement. Moreover, clear governance frameworks with safeguards around data privacy, ethical oversight, and legal liability are needed to ensure AI surveillance and tracking tools are deployed responsibly.

Many of these challenges are surmountable and, indeed, are already being addressed by researchers, startups, and public bodies. But progress will depend on collaboration, a willingness to rethink outdated risk management models and, crucially, sustained investment.

A hot investment

As climate change intensifies and wildfires become more frequent, the financial case for prevention is becoming harder to ignore. For investors, this represents a growing opportunity to support the solutions we need.

AI-enabled wildfire technologies are no longer experimental. Commercial clients with strong incentives to reduce fire-related losses—such as utilities, landowners, and forestry operators—are already using them to protect their infrastructure, assess risk, and allocate resources more efficiently. This business-to-business demand supports stable, recurring revenue that does not depend solely on public funding.

The scalability of AI-driven solutions is also compelling. Once developed, AIs can be adapted and deployed across diverse geographies at relatively low incremental cost, particularly where satellite infrastructure or open-source platforms are already in place. As adoption grows, so too does the quality of training data, creating a positive feedback loop that improves accuracy and widens applicability.

Investment opportunities also abound beyond the core technology. Adjacent sectors like data infrastructure, edge computing, remote sensing hardware, and climate risk analytics comprise a growing ecosystem that supports AI-based wildfire resilience. For far-sighted investors, this ecosystem represents a frontier for systems change investment where financial returns align with environmental and social impact.

Read more: Building a Circular Bioeconomy to address the growing risk of wildfires | Lombard Odier

Adjacent sectors like data infrastructure, edge computing, remote sensing hardware, and climate risk analytics comprise a growing ecosystem that supports AI-based wildfire resilience

Kindling a more resilient future

Wildfires are no longer isolated disasters. They are symptoms of escalating global pain points driven by climate change, environmental degradation, and underinvestment in prevention. But they are also a proving ground for the technologies and investment strategies that will shape a more resilient future. AIs are already helping us better understand, predict, and respond to wildfires, underscoring how the global systems change unfolding around data productivity can support climate adaptation by improving how we respond to crises and enabling faster, smarter, more targeted interventions.

By supporting decarbonisation, enhancing nature resilience, and protecting lives and livelihoods, AI-enabled wildfire solutions also directly contribute to the global systems change driving the transition to a net-zero, nature-positive, socially fair, digitally enabled economy. For investors, they offer a compelling combination of risk mitigation, technological progress, and long-term impact. And for communities on the front lines of climate change, they offer something more: the possibility of a world where wildfires no longer have the upper hand.

Addressing the root causes of increasingly severe wildfires requires a profound transformation of our economic and ecological systems

While AI offers powerful capabilities for forecasting, detection, and response, it represents just one part of a much larger solution. Addressing the root causes of increasingly severe wildfires requires a profound transformation of our economic and ecological systems. This means rethinking how we design and manage landscapes, and reassessing the value we place on forests, land use, and rural economies. Such a transformation must be systemic, grounded in principles of resilience, regeneration, and long-term stewardship.

AI can play a vital role in this shift by enhancing how we collect, interpret, and act on environmental data, enabling more informed and proactive landscape management. But we can only realise the full potential of AI in fighting wildfires in tandem with coordinated policy reform, sustained investment, and a renewed, more respectful relationship with nature.