Google DeepMind and Google Cloud will give up to 200 European scientific researchers direct access to some of the company’s most advanced artificial-intelligence systems, together with a shared pool of $3 million in cloud-compute credits, the company announced on Wednesday.
The announcement came at an artificial-intelligence summit in Dublin, where Maureen Costello, Google Cloud’s vice president for the UK, Ireland and Sub-Saharan Africa, laid out the pledge. The model line-up is drawn from DeepMind’s scientific portfolio: AlphaFold 3 for protein and molecular structures, WeatherNext for weather forecasting, AlphaGenome for DNA research, AlphaEvolve, an algorithm-designing agent, and AlphaEarth Foundations, which maps the planet from satellite data. According to Reuters’ report on the announcement, the researchers will be chosen by the European Commission through the EU’s Resource for AI Science in Europe (RAISE) programme.
The stated aim is to help teams solve hard problems more quickly — from forecasting extreme weather to mapping molecular structures — and to narrow the computing gap that European research groups sometimes face. Split evenly, $3 million across 200 researchers works out at roughly $15,000 each: enough to test ideas that would otherwise stall, though modest by the standards of an industry that spends billions on data-centre capacity.
Europe’s own infrastructure push
The pledge lands in the middle of Europe’s drive to build scientific AI capacity at home. The EU’s Horizon Europe programme has committed 600 million euros to expand computational access for scientists and start-ups, annual AI research investment is set to double past 3 billion euros, and a separate 20 billion euro fund aims to build up to five AI gigafactories, according to WebProNews’ analysis of the programme. Google’s offer, modest beside those figures, buys it a visible role inside the EU’s AI-for-science agenda.
Access is not a product launch
One distinction matters for the research teams weighing the offer: this is an access commitment, not a new general-product release. Google has not published how the credits will be divided, when access begins, or how applicants will be evaluated — and participating teams will still need their own validation and reproducibility practices before relying on model outputs in consequential research. Governments are meanwhile tightening oversight of the leading AI developers, a reminder that frontier systems arrive with scrutiny attached, and personnel upheaval has also shaken leading AI labs this month.
