Google Funds Ai2 Climate Model Project

Google.org and Google Cloud are backing Seattle-based Ai2 with funding and cloud credits to build an open-source climate projection model.

Updated on Nov. 4, 2026 in Artificial Intelligence

Live Poll

Do you trust long-range climate projections to guide local water planning decisions?

Google.org is providing about $5 million in funding and $1.8 million in cloud credits to Ai2 for a three-year effort to build NeuralGCM+, an open-source climate projection model. The model builds on Google's NeuralGCM, which mixes traditional physics equations with AI.

Why it matters

Google's NeuralGCM work focuses on weather forecasts looking ahead a year at most, while decades-long climate projections require different expertise. The project aims to give researchers a reliable tool for longer-term climate planning, including water decisions in Seattle's region.

The funding is a commitment of about $5 million plus $1.8 million in cloud credits from Google.org and Google Cloud, with four new staff joining an eight-person team. Reliability of the model's warmer-world predictions has not yet been demonstrated, and a 2024 stress test showed physically implausible results at 7 degrees Fahrenheit above current ocean temperatures.

The players

Ai2

The Allen Institute for AI is a Seattle-based nonprofit research institute founded in 2014 by Paul Allen, focused on artificial intelligence research.

Google.org

Google's philanthropic arm, which provides funding for social and scientific initiatives around the world.

The details

Ai2 will connect NeuralGCM+ to an ocean and sea ice model to run stable multi-decade simulations and will add more established physics, checking its warmer-world predictions against traditional climate models. The original physics-plus-AI plan was set aside because AI could not be trained well inside older Fortran-based climate models.

2014: Paul Allen founded the Ai2 nonprofit AI institute.
Around 2016: Paul Allen's climate modeling effort began at Vulcan.
2021: The climate modeling group moved to Ai2.
2024: Google published NeuralGCM research in Nature.
November 4, 2026: Google's funding for Ai2's climate model was announced.

Roadmap

The project extends Google's NeuralGCM, published in Nature in 2024, from year-ahead weather forecasting into multi-decade climate projection, a horizon Google has shown less interest in. It reflects a broader pattern of hybrid physics-and-AI models attempting to replace or supplement traditional climate simulation tools.

Once released, NeuralGCM training code is expected within the first year of the project, giving researchers access to a climate model they can run and check themselves. Reliability in a warmer world may take longer than three years, so projections from the model will not be dependable for planning decisions in the near term.

The takeaway

Open, physics-grounded AI climate tools are still early, and their predictions should be verified against established models before they guide long-range decisions about water and rainfall.