GHGSat CEO: AI Doesn’t Have to Be a Climate Villain
The CEO of GHGSat believes AI can translate satellite emissions data into actionable solutions for climate policy, though it won’t solve the problem on its own.

The column’s author notes that the rapid growth of data centers is increasing strain on energy systems and creating environmental risks. However, in his view, climate technologies should not abandon AI: it is already being used to analyze protein structures, detect cancer early, and forecast weather.
At GHGSat, satellites monitor millions of sites annually. AI assesses the likelihood of methane emissions in images and prioritizes the most promising observations for processing. Specialists then confirm the emissions, and the client receives not a mass of images but an alert about a specific source at their facility.
The next level involves integrating diverse data sets. To assess emission trends in the oil and gas sector in Europe or Central Asia, it is necessary to compare company reports, regulator records, and satellite measurements, while also accounting for geopolitical and macroeconomic conditions. Such work can take months, whereas AI can accelerate it and present results in a format suitable for officials or journalists.
A similar approach is used to monitor deforestation. AI classifies signals about tree loss by possible causes—such as road construction or wildfires—to guide the appropriate response. Geospatial foundation models, trained on large archives of satellite imagery, are only beginning to develop.
However, available data alone does not guarantee change. Decisions must be made by infrastructure operators, regulators, and financial institutions. Ahead of the upcoming COP31, the author expects methane to once again be a key agenda item: while measuring and explaining climate risks is now possible for a broader audience, action will still depend on them.




