Artificial intelligence is moving to the frontlines of climate innovation, as 23 startups from around the world are selected to join the 2025 Compute for Climate Fellowship, a global programme led by the International Research Centre on Artificial Intelligence (IRCAI), under the auspices of UNESCO, and Amazon Web Services (AWS).
The Fellowship supports the development of AI-driven proof-of-concepts that address systemic environmental challenges, from reducing emissions in shipping to accelerating the discovery of sustainable materials and enabling precision agriculture. Each selected startup will receive full funding to develop their concept, along with access to advanced cloud compute, AI tools, and expert mentorship from IRCAI and AWS.
This year’s cohort brings together organisations using AI in unexpected and ambitious ways: from developing autonomous video systems to monitor the oceans, to designing protein foundation models for climate and health applications, to building next-generation fusion energy systems powered by inertial lasers. Many are also focused on embedding AI into under-resourced geographies or sectors traditionally overlooked in the global innovation landscape.
AI as an engine for system change
“The intersection of climate innovation and generative AI is no longer theoretical – it’s happening now, and it’s reshaping what’s possible,” said Jon Jones, Vice President and Global Head of Startups at AWS. “This fellowship is about fuelling this momentum by backing solutions with the potential to drive systemic change.”
Among the selected ventures are Cusp AI, a UK-based startup building an AI-enabled materials search engine, and Amini, based in Kenya, which is building AI-native infrastructure to convert fragmented satellite and ground data into usable intelligence across the Global South. Meanwhile, BeyondMath, also from the UK, is developing foundation models to simulate physical systems in real time, a leap toward AI-assisted engineering at planetary scale.
IRCAI’s Chief Operating Officer, Davor Orlic, emphasised the programme’s interdisciplinary approach. “With this year’s cohort, we’re deepening public-private partnerships that unite industry, government, and research to harness AI for climate action,” he said. “Anchored in cutting-edge compute, rigorous R&D, and world-class talent, we’re excited to support startups driving real, scalable solutions with the power to make a transformative impact for our planet.”
The scope of the programme reflects the complexity of the climate crisis. Participating startups are tackling issues as varied as urban heat mitigation (FortyGuard), methane conversion (Windfall Bio), AI-enhanced geothermal systems (Eavor Technologies), and AI-guided crop evolution (Avalo). Several, including Climate X and CounterCurrent, are using AI to model climate risks and optimise logistics for high-impact sectors such as finance and global shipping.
Infrastructure, mentorship and visibility
What makes the Fellowship distinctive is not only the provision of AWS’s computing resources, but also its emphasis on rapid AI development through integrated support. Previous participants describe the programme as a critical enabler in getting research into the field. Kenny Lee, CEO of Aigen, said the Fellowship helped move development timelines “from weeks to days,” allowing the company to scale its solar-powered robotic platform for use in active farm environments.
The Fellowship also aligns closely with the United Nations’ Sustainable Development Goals and IRCAI’s broader mission to advance responsible AI. As coordinator of the European Green AI project RAIDO and publisher of the annual Global Top 100 AI list, IRCAI acts as a global hub for AI solutions that meet both scientific and social benchmarks.
As AI continues to evolve as a foundational technology, programmes like Compute for Climate are helping to redefine what applied innovation looks like in an era of global crisis. With technical infrastructure no longer a limiting factor, the focus is shifting to how fast responsible, scalable models can move from lab to impact, and who will lead that transition. For the 23 startups in this year’s cohort, the answer is already in motion.
