Google is putting $40 million of AI credits into America's national science labs

Key points

• Google announced a $40 million commitment of AI tokens and cloud credits for researchers on November 20, 2026 at the Department of Energy Genesis Mission Summit.
• All 17 US Department of Energy National Laboratories will get early access to Google DeepMind's science tools, including AlphaFold 3, AlphaEvolve, AlphaGenome, WeatherNext and AlphaEarth Foundations.
• Tens of thousands of DOE staff will receive Gemini for Government seats for one year.
• Pacific Northwest National Laboratory is using AlphaEvolve to search complex mathematical systems too big to explore by hand.
• The National Laboratory of the Rockies cut microscope calibration time from over 90 minutes to about 13 minutes using Gemini.

Google is handing US national scientists a very big bag of AI credits.

At the Department of Energy's Genesis Mission Summit this week, Google committed $40 million in AI tokens and cloud credits to researchers working under the Genesis Mission, the White House programme that aims to double the pace of American scientific discovery within a decade.

Google DeepMind, the company's main AI research lab, announced the plan alongside Google Public Sector. It builds on an earlier promise from December to help the mission along.

So what is actually in the box?

First, Genesis Mission awardees get access to Google's science-focused AI models. That includes AlphaFold 3, which predicts the shape of proteins and how they interact (useful for drug discovery). AlphaEvolve, a Gemini-powered coding agent, meaning software that can plan and carry out multi-step programming tasks on its own, aimed at designing new algorithms. AlphaGenome, which studies how tiny variations in DNA lead to disease. WeatherNext, a set of AI weather forecasting models. And AlphaEarth Foundations, a general-purpose model for mapping the planet.

Second, tens of thousands of staff across the 17 National Laboratories get Gemini for Government seats for a year. Gemini is Google's main chatbot and assistant, and the Government version runs on a secure platform cleared for federal work.

What does this actually change for scientists?

Quite a lot, if the early stories hold up.

At Pacific Northwest National Laboratory, senior scientist Dr Henry Kvinge is pointing AlphaEvolve at huge mathematical systems that no human could fully explore by hand. He said the tool finds hidden patterns automatically, turning what would be years of work into something much shorter.

At the National Laboratory of the Rockies, Dr Steven R. Spurgeon's team plugged Gemini directly into their laboratory instruments. The result, in his words: microscope calibration dropped from over 90 minutes to about 13 minutes, roughly eight times faster. The number of manual steps needed to focus a single image fell from as many as 50 down to two.

That is the kind of change working scientists actually feel. Less fiddling with hardware. More time on the science.

Why does this matter beyond the labs?

The National Laboratories are where a lot of the country's biggest science happens. Fusion energy research, new battery materials, climate modelling, national security work. If AI shaves months off routine steps, the pipeline of real-world results, cleaner power, better medicines, faster forecasts, moves quicker too.

There is also a competitive angle. China, the UK and the EU are all pouring money into AI-for-science programmes. A $40 million in-kind commitment from one company is not going to win that race on its own, but it does put Google's best tools inside the buildings where the work is done.

A quick note on the fine print. "In-kind" credits means researchers are not getting cash. They are getting compute time and model access on Google's systems, which also means their queries and data flow through Google infrastructure. For federal science work that runs on Gemini for Government, a secure environment built for that purpose, but it is worth knowing where the pipes go.

Google says more details for researchers will come at its Public Sector Summit in October. If you work at a National Lab, your IT team is probably already getting the emails.