Models, datasets, and capital for restoration, biodiversity, and decarbonization. The work of redirecting AI's compute and capital toward the largest collective-action problem humans have ever faced — and the species we share the planet with.
The compute and energy footprint of frontier AI is, on its own, a climate question — one that the industry's growth trajectory keeps making larger. At the same time, AI tooling is among the highest-leverage instruments we have for the ecological work itself: high-resolution climate modeling, biodiversity monitoring at planetary scale, materials discovery for storage and decarbonization, grid optimization that makes a renewables-first grid actually operable. Both things are true. The catalog's job in this domain is to surface the projects that treat AI as instrument-of-restoration rather than driver-of-extraction — and to be honest about the difference.
Open climate ML, downscaled regional models, foundation weather models, and the decision-support tooling that translates raw projections into something policymakers, planners, and frontline communities can actually use.
A volunteer organization of researchers and practitioners working on ML applied to climate. Runs workshops at NeurIPS and ICML, publishes the canonical "Tackling Climate Change with Machine Learning" survey, mentors early-career researchers, and coordinates open-compute requests for climate-relevant projects. The connective tissue of the field.
ClimSim
An open ML benchmark designed for hybrid physical-and-learned climate emulators. Tens of millions of high-resolution simulation samples released so researchers outside the major modeling centers can train and evaluate emulators against a common reference. NeurIPS 2023 dataset track.
Climate scenarios and visualization built for non-specialists — planners, communities, businesses, journalists. Translates the underlying climate-model output into mappable, interpretable views without flattening the uncertainty. Independent and grant-funded.
Mila Climate Change AI Lab
Yoshua Bengio's group at Mila runs sustained climate-AI work alongside its core ML research, including downscaling, extreme-event prediction, and the ethics-of-deployment questions most labs skip. A reliable bellwether for what serious academic climate ML looks like.
ClimaX, Pangu-Weather, and other open weather foundation models
A new class of high-resolution AI-driven weather and climate models — from Microsoft Research, Huawei, and academic groups — with weights or architectures released openly. Significant for the field because they push frontier capability outside the closed national modeling centers.
Public satellite data, planetary-scale compute, and the pipelines that take raw imagery and turn it into something a community group, a forest carbon project, or a fisheries regulator can act on. The substrate most environmental AI is built on top of.
Petabyte-scale catalog of open environmental data (Sentinel, Landsat, MODIS, and more) with a colocated compute environment researchers can run in for free. The closest thing to a public-utility EO platform from a hyperscaler, and a default starting point for serious environmental ML.
Long-running planetary-scale geospatial analysis platform. Free for research, education, and most non-profit use; paid commercial tiers exist. The dominant academic EO compute environment for the last decade, with multi-decade satellite archives indexed and ready to query.
NASA's open public archive of Earth-observing satellite data. The original public-domain EO commons. Increasingly accessible through cloud-native APIs and Zarr-formatted datasets, which makes it usable for ML pipelines without legacy HDF-wrangling.
Copernicus & Sentinel Hub
The European Space Agency's Sentinel constellation and the Copernicus Open Access Hub provide free, full-resolution radar and optical imagery on a regular global revisit cadence. The most consequential public EO program of the last decade, and the data backbone of most EU-side climate and biodiversity monitoring.
Remote-sensing and ML for forest carbon project verification, with a focus on improving the credibility of nature-based projects rather than wallpapering over their problems. Operates under the assumption that the offsets market deserves more skepticism, not less.
Natural Capital Exchange. Combines high-resolution forest inventory ML with a marketplace for landowner-side participation in forest carbon. Listed here for its model tooling and willingness to publish methodology, alongside the catalog's general skepticism of offset programs (see //08).
Citizen science platforms, AI-assisted species identification, individual-animal recognition for conservation research, and the institutions building the data layer for whatever post-extinction biodiversity work the next century actually requires.
A joint initiative of the California Academy of Sciences and the National Geographic Society. AI-assisted species identification trained on tens of millions of community-submitted observations, with research-grade outputs flowing into GBIF and academic biodiversity work. The largest civilian biodiversity monitoring effort in existence.
Cornell Lab's bird observation database (eBird) and AI-driven mobile ID app (Merlin), built on top of decades of citizen-science data. The audio-ID model in Merlin is one of the more impressive deployed wildlife-AI systems on the consumer side, and the data underwrites real population trend research.
Open-source platform for AI-assisted individual-animal identification — whales, sharks, giraffes, big cats — from photo evidence. Used by researchers and conservation NGOs to track population dynamics over time without invasive tagging. Now operated under the Conservation X Labs umbrella.
A research nonprofit applying foundation-model methodology to non-human animal communication. Long-horizon, scientifically careful, foundation-funded. Listed here because it's one of the few well-resourced groups taking the question seriously without overclaiming results.
Grants and prize-driven funding for AI and innovation projects targeting biodiversity loss. Runs the Con X Tech Prize and similar programs that move small-but-meaningful capital into early-stage conservation tooling, including individual-animal-ID and acoustic monitoring work.
Open-source grid AI, materials discovery for storage and clean generation, carbon-aware computing standards, and the policy-and-tooling work that turns a decarbonization curve from a slogan into a buildout pipeline.
A nonprofit product lab building open-source ML for grid and solar forecasting, deployed at the UK National Grid ESO and elsewhere. Codebase is public, models retrain transparently, and improvement targets are framed in megatons of avoided emissions rather than business KPIs.
A nonprofit providing real-time and forecast marginal-emissions data for electricity grids worldwide. The API of choice for any tool that wants to actually time-shift load in response to grid carbon intensity rather than just claim to. Underwrites a lot of the carbon-aware computing ecosystem.
A Linux Foundation effort developing open standards (Software Carbon Intensity spec, Carbon-Aware SDK, Impact Framework) for measuring and reducing software's emissions. Member-supported, pre-competitive, and the closest the industry has to a shared technical floor on what "green software" means.
Nonprofit focused on household and community electrification — heat pumps, induction, EVs, rooftop solar. Combines policy advocacy with practical consumer-facing tooling (incentive calculators, contractor pipelines). One of the few organizations operationalizing decarbonization at the household scale.
Open materials discovery initiatives (e.g. Materials Project, Open Catalyst)
A loose constellation of openly-licensed datasets and ML-driven discovery efforts for energy materials — battery chemistries, catalysts for clean fuels, photovoltaic compounds. The Lawrence Berkeley-led Materials Project and Meta-FAIR's Open Catalyst datasets are the most cited reference points; the field is wider.
Capital, infrastructure, and journalism flowing to where climate harm lands first — and to the communities organizing the response. The field test for whether climate AI is a tool for restoration or just another extractive market.
A national coalition of frontline organizations operating under the Just Transition framework. The political vehicle for a lot of US-side climate-justice organizing, and a useful reference for distinguishing the work that includes affected communities from the work that talks about them.
A US-wide network of community organizations on the front line of climate-driven flooding, wildfire, and extreme weather. Connects affected communities to scientific, legal, and technical support. The mailing list and coalition base for a lot of the actual climate-adaptation work happening at neighborhood scale.
Independent nonprofit newsroom covering climate, justice, and solutions. Grist's investigative work on heat-island disparities, frontline-community displacement, and offset-market accountability is a reference point for what climate journalism with a justice frame actually looks like.
A research-and-strategy organization producing technical analysis grounded in environmental-justice principles — intended to be used by frontline organizations in policy fights, not by funders looking for cover. Distinct from the broader academic equity-research economy.
Community early-warning systems for vulnerable regions
A growing set of community-led and humanitarian-funded early-warning projects — flood, heat, cyclone, wildfire — that bring AI-assisted nowcasting to populations the national-meteorological-agency pipelines under-serve. Often built on top of WMO and Copernicus data, with last-mile delivery via SMS, radio, or community responders.
Capital aligned with restoration, biodiversity, decarbonization, and the just-transition work that climate philanthropy is finally taking seriously. The funder ecosystem most realistically able to support multi-year work in this domain.
The largest single private climate-and-nature commitment in history. Active program areas include nature-based solutions, climate justice, monitoring and accountability, and AI-for-nature grants. Operates at a scale that makes it a structural actor in the field rather than a participant.
A regranting and strategy organization that has shaped a significant share of US and international climate philanthropy for over a decade. Works through a network of funder collaboratives and regional partners on transport, buildings, industry, food, and finance decarbonization.
Hewlett's Environment program has been one of the most consistent multi-decade funders of climate and clean-energy policy in the United States. Supports the institutional infrastructure (think tanks, advocacy, research) that the rest of the field operates inside.
A climate-only foundation funded by an independent endowment — legally and operationally distinct from Sequoia Capital. Active in fossil-fuel phase-out, methane reduction, and climate-resilient agriculture. Has moved meaningful capital quickly and quietly.
Long-running funder of global development work. Recent strategy centers a Climate & Power program targeting energy access in the global south alongside decarbonization in the global north — one of the more honest framings of the just-transition problem in mainstream philanthropy.
Energy Foundation & aligned regional/topical funds
A constellation of regranting and topical funds — Energy Foundation, the European Climate Foundation, regional just-transition funds — coordinated through ClimateWorks and similar collaboratives. Where most climate movement organizations actually receive their grant capital, even if the brand on the check is different.
This page is a sample of the Climate & Biosphere domain. Inclusion is editorial — not a commercial endorsement, not a vetting of every claim a project makes, and not a recommendation that any single tool fits every situation. Foundations listed here are not partners; they're documented as the funder ecosystem most aligned with this work, public information, no relationship implied. If you build in this space and we missed you, submit your project below.
Beyond the catalog-wide exclusions, Climate & Biosphere has its own line. Climate-flavored language has become marketing camouflage for a great deal of work that increases harm. Naming the line is part of the work.
If your work fits the Climate & Biosphere domain — and is honest about which line it sits on — we want it in the catalog. Submissions are reviewed editorially before inclusion. We may reach out for clarifying details.
Pulled live from the AI for Planet catalog — reviewed editorially, with maturity and evidence flagged. Inclusion is not endorsement; see the curatorial note below. Submissions for the Climate & Biosphere domain open Q3 2026.