AlphaFold 3 Access Opens for 200 EU Scientists via Google DeepMind Compute Grant

Google DeepMind and Google Cloud announced on October 14, 2026, that they will grant up to 200 top European scientific researchers direct access to five frontier models and a shared pool of $3 million in Google Cloud compute credits. The offer runs through the European Commission’s RAISE initiative and targets researchers already selected by the Commission for their work in life sciences, climate modeling, and genomics.

AlphaFold 3 and Additional Models Included in the Program

The models made available include AlphaFold 3 for protein and molecular structure prediction, WeatherNext for high-resolution weather forecasting, AlphaGenome for DNA sequence analysis, AlphaEvolve for algorithm design, and AlphaEarth Foundations for satellite and geospatial mapping. Researchers will use these tools on Google Cloud infrastructure, with the $3 million credit pool allocated across the cohort rather than awarded individually.

Access arrives through the RAISE virtual institute, which the European Commission launched in November 2025 to close the compute gap facing EU laboratories. The Commission handles participant selection; Google DeepMind supplies the models and credits. No per-researcher credit amounts or project duration limits have been published yet.

The announcement was made at an AI summit in Dublin. Google Cloud vice president Maureen Costello stated that the arrangement supports the EU’s AI for Science agenda by supplying both the models and the compute needed to run them at scale.

Context from Prior Model Releases and EU Funding

AlphaFold 3 first appeared in May 2024 through a collaboration between Google DeepMind and Isomorphic Labs. Non-commercial academic access to the model weights and code opened in November 2024. The current program expands that access by adding production-scale compute and four additional models that had previously been limited to internal or commercial partners.

RAISE forms part of the larger Horizon Europe framework, which earmarked roughly €107 million for an initial pilot and an additional €600 million for AI infrastructure and gigafactory access. The Google DeepMind contribution supplies the first large-scale private-sector compute commitment tied directly to the RAISE framework.

Earlier efforts by the Commission focused on building shared GPU clusters and negotiating discounted cloud rates. The new partnership adds frontier-model access that individual European labs could not otherwise obtain at this scale.

Practical Steps for Selected Researchers and Expected Outcomes

Researchers chosen by the Commission will receive account credentials and credit allocations in the coming weeks. They can then run inference and limited fine-tuning jobs on the listed models without incurring personal cloud charges up to the shared cap. Project results will be tracked through standard academic reporting channels rather than through a new centralized portal.

Potential applications include faster protein-ligand docking studies, improved regional extreme-weather forecasts, and large-scale genomic variant interpretation. These use cases align with existing EU research priorities in health, climate resilience, and biodiversity monitoring.

Future expansions will depend on measured scientific output and on whether the current credit pool proves sufficient for the selected projects. The Commission has indicated it will evaluate results before negotiating additional rounds with Google or other providers.