Google DeepMind and Google Cloud will give up to 200 European researchers direct access to the company’s frontier AI models, plus $3 million in compute credits, the company announced Wednesday, October 14.

Reuters first reported the move, which runs through RAISE, the European Union’s Resource for AI Science in Europe. Google framed it as support for the EU’s AI for Science agenda: hand working scientists the models and the computing power to run them, and let them tackle problems that would otherwise stall for lack of resources.

It’s a modest amount of money by AI industry standards. It is also a revealing one. The scarcest resource in modern science isn’t always talent or ideas. Often, it’s compute.

What European researchers actually get

The program covers five of Google’s scientific tools, each aimed at a different kind of hard problem.

AlphaFold 3 predicts protein and molecular structures, work that has already reshaped drug discovery and structural biology. WeatherNext handles advanced weather forecasting. AlphaGenome targets DNA research. AlphaEvolve is an agent that designs algorithms. AlphaEarth Foundations maps the planet using satellite data.

Maureen Costello, vice president for the UK, Ireland and Sub-Saharan Africa at Google Cloud, wrote in Google’s announcement that the tools will help teams “solve hard problems more quickly,” from forecasting extreme weather to mapping molecules.

One wrinkle is worth knowing. Academics have been able to download AlphaFold 3’s code and weights for non-commercial work since November 2024. So the genuinely new part of this offer isn’t the model alone. It’s the combination: direct access to the full set of tools, plus credits to actually run them, inside a European research program.

Why Google is doing this now

Europe has spent the last two years trying to close a compute gap with the United States and China. Its researchers produce world-class science, but many teams can’t get enough processing power to run the latest models at scale. That gap is the problem RAISE was built to address.

RAISE launched as a virtual institute in November 2025 with €107 million in pilot-phase EU funding, according to reports on the program. It sits inside a much larger European push: Horizon Europe has committed €600 million to expand computational access for scientists and startups, and a separate €20 billion fund set up in February 2025 aims to build up to five AI gigafactories on the continent.

Google’s timing fits that push. The company gets to place its best-known scientific tools at the center of Europe’s research infrastructure, build relationships with the scientists who will use them, and show goodwill in Brussels at a moment when the EU is writing the rules for AI. That last part matters. Europe’s AI Act is already in force, and regulators there are deciding how the toughest rules apply to the most capable models.

For Google, $3 million in credits is a small price for that kind of presence.

What the announcement doesn’t say

For all the detail on the tools, Google has left the practical questions unanswered. It hasn’t said how researchers will be selected, when access begins, or how the credits will be divided.

That matters more than it sounds. Split evenly across 200 researchers, $3 million comes to $15,000 each. That’s real money for testing an idea, running a batch of experiments, or prototyping a new approach. It won’t fund large-scale model training or a months-long research program. Researchers who need serious compute will still need their own grants, institutional resources, or a slot on one of Europe’s planned gigafactories.

Access also isn’t the same as results. Teams still have to prepare their data, validate model outputs, and decide whether a given tool fits their specific question. A protein-structure prediction is only useful if a lab can check it, reproduce it, and build on it. Google’s announcement doesn’t change that work. It just removes one barrier at the start of it.

There is also the question of what happens after the credits run out. If the program works, demand will exceed supply fast. Google hasn’t said whether it plans to expand the cohort, renew the credits, or treat this as a one-time pilot.

Why this matters

Small access programs like this one can punch above their weight in science. AlphaFold itself started as a research project before it became a tool thousands of labs depend on, and its creators won a Nobel Prize in 2024. The next breakthrough in protein science, weather forecasting, or genomics could come from a team that today can’t afford the compute to test its idea.

That is the bet Google is making. And it is a bet with a competitive edge: the researchers who learn to work with Google’s tools are more likely to build their future projects around them.

For US readers, the story cuts two ways. American labs still hold the deepest bench of AI talent and compute, and US companies dominate the market for frontier models. But Europe is moving fast to build its own scientific AI capacity, and programs like RAISE give European teams a reason to stay in Europe rather than decamp to Silicon Valley or Boston. Talent follows tools and funding, and both are now on offer in Brussels as well as in California.

Still, keep the scale in view. Two hundred researchers is a pilot, not a transformation. Europe’s compute gap is measured in billions of euros and gigawatts of data-center power, not in a single $3 million credit pool. If this program stays at its current size, it will be remembered as a goodwill gesture. If it grows, it could become the template for how frontier AI companies work with public science.

The test will be in the details Google hasn’t shared yet: who gets picked, how fast they get access, and what they produce with it.

Frequently asked questions

What is Google offering European researchers?

Google DeepMind and Google Cloud will give up to 200 European researchers direct access to five frontier science AI tools, plus $3 million in compute credits, through the EU’s RAISE initiative.

Which Google AI models are included?

The program covers AlphaFold 3, WeatherNext, AlphaGenome, AlphaEvolve, and AlphaEarth Foundations, spanning protein structures, weather forecasting, DNA research, algorithm design, and satellite mapping.

How much is the compute credit worth per researcher?

Google has not said how the $3 million will be divided. Split evenly across 200 researchers, it would come to $15,000 each, enough to test ideas but far from funding large-scale model training.

When will European researchers get access?

Google has not published a start date or selection criteria yet. Those details will decide whether the program is a quick win for working scientists or a slow-moving announcement.

Sources: Reuters, Unite.AI.