Genesis Mission: DOE Selects First 278 AI-for-Science Projects, Backed by Google, Microsoft, and Meta

The US Department of Energy (DOE) announced on July 22, 2026, the first 278 projects selected under the Genesis Mission, the national initiative to accelerate scientific research with AI¹. Of these, 168 are led by universities, 87 by DOE and National Nuclear Security Administration (NNSA) national laboratories, 19 by companies, and 4 by nonprofits, with research spanning nuclear energy, critical minerals extraction, chip design, and commercial fusion¹².

The announcement also clarified how major AI companies are taking part. Google committed $40 million in AI tokens and cloud credits³, Microsoft is reported to be contributing $40 million in Azure and AI computing credits plus $20 million in solution engineering services², and Meta has already deployed its open-source vision models to a flagship national-lab project⁴. AI for science is moving from individual corporate efforts to a national-program scale.

What Is the Genesis Mission? Doubling the Pace of Discovery in a Decade

The Genesis Mission is a national effort launched by the White House in November 2025², with the goal of using AI to double the pace of American scientific discovery within a decade³. The DOE’s announcement described it as an effort to combine America’s scientific capabilities, advanced AI, high-performance computing, and leading researchers to transform how scientific discovery is conducted¹. Energy Secretary Chris Wright said, “America has no shortage of bold ideas or talented scientists, and the response to the Genesis Mission proves that. The 278 projects selected today represent the very best of our nation’s scientific enterprise.”¹

The first cohort involves 16 DOE and NNSA national laboratories, 142 universities, 157 companies, and 13 nonprofit organizations¹. Genesis Mission consortium members have reportedly secured more than $500 million in contributions overall².

Google’s Contribution: From AlphaFold 3 to Gemini for Government

At the DOE’s Genesis Mission Summit 2026, Google announced $40 million in AI tokens and cloud credits for researchers³. The commitment includes access to Google’s frontier AI-for-science portfolio: AlphaEvolve, a Gemini-powered coding and discovery agent for designing advanced algorithms; AlphaFold 3, which predicts the structures and interactions of proteins and other biomolecules; AlphaGenome, for understanding DNA variation and its impact on biology and disease; WeatherNext, its AI weather forecasting models; and AlphaEarth Foundations, a foundational model for planetary mapping³.

Google will also provide tens of thousands of users across DOE national laboratories with Gemini for Government seats and tokens for one year³. Early results are already being reported: at the National Laboratory of the Rockies, deploying Gemini in instruments cut microscope calibration time from over 90 minutes to about 13 minutes—eight times faster³. Google first stated its commitment to the Genesis Mission in December 2025³; this announcement puts concrete resources behind it.

Meta’s Contribution: SAM 3 and DINOv3 on 300 GPUs

Meta announced on July 21 that it contributed two open-source AI models to SYNAPS-I, one of the flagship Genesis Mission projects⁴: Segment Anything Model 3 (SAM 3), which draws precise pixel-level boundaries around objects in images, and DINOv3, a self-supervised vision model that learns visual patterns without human labeling. Both models were fine-tuned on scientific imaging data and deployed across 300 A100 GPUs at national supercomputing facilities⁴.

SYNAPS-I is led by Lawrence Berkeley National Laboratory, with 60 researchers across five national labs including Argonne, Brookhaven, Oak Ridge, and SLAC⁴. In the pipeline, SAM delivers precise boundaries while DINO provides global context; analysis that previously required a month of expert annotation per time step now takes roughly 15 minutes from data acquisition to fully reconstructed, semantically labeled 3D volumes⁴.

Separate Threads Assemble Into a National Program

The ties between AI and scientific research have thickened rapidly this year. In the private sector, UK-based CuspAI raised $450 million and launched the 45-plus-member AI Materials Foundry, while Google has been building out its science portfolio alongside its recent Gemini 3.6 Flash model releases. The Genesis Mission’s scope includes nuclear power and fusion—a reminder that AI’s energy demands and power generation have become two sides of the same coin, a dynamic we covered in our profile of nuclear startup Oklo.

This announcement shows those separate threads being assembled into a national division of labor: national-lab infrastructure, university research capacity, and private-sector AI models and compute.

New Touchpoints Between R&D and Procurement

For readers outside the US, direct participation opportunities are limited, but there are two practical implications. First, science-oriented AI models like AlphaFold 3 and SAM 3 will mature into battle-tested tools through national-project validation—useful evidence for R&D teams in materials, drug discovery, and manufacturing weighing their own adoption. Second, US AI policy is advancing not only through defensive measures like export controls and sanctions but also through large-scale investment in scientific infrastructure. Organizations procuring research computing or selecting partners should factor in how deeply the major AI companies are now tied into national programs.

The DOE is expected to publish results as the selected projects progress, and we plan to revisit the story as individual project outcomes emerge.