Genesis Mission — national-lab data into models to double US scientific productivity
Genesis Mission — national-lab data into models to double US scientific productivity
Vintage: 2026-07. Primary: michael-kratsios interviewed 2026-07-29 (Moonshots, published 2026-08-04). Policy/capability claims — treat the $5B-grants-last-week and "double over a decade" target as a late-July 2026 snapshot.
One-line summary: The Trump administration's Manhattan-project-style AI-for-science program: make 70 years of US national-lab data AI-ready, put it into models, and double the productivity of the entire US scientific enterprise over the next decade, with ~$5B in grants announced the week of the Genesis Mission summit (many biotech-nested). Signed as a presidential program in December of the prior year.
The insight
michael-kratsios (OSTP) frames Genesis as the only White-House-level way to push frontier AI into agencies at the new velocity of change — "the most important thing for the nation is to make sure that we're applying this unbelievable technology called artificial intelligence to scientific discovery. And it's not just at one agency, it's across all of government."
The load-bearing mechanism is data, not a new model: "the US national labs have a tremendous amount of scientific data across a wide variety of domains that have not been made AI ready, that have not been uploaded to models." Unlocking that archive is the claimed 2× productivity lever. Kratsios himself flags the 2× as a number from last year when they were building the program, and that they internally discussed whether to raise it for the Golden Age report but kept it because it was already public.
Canonical chain: genesis-mission-to-ai-for-science-acceleration.
The chain
US national-lab scientific archives have not been made AI-ready; Genesis Mission (Manhattan-style, signed December, $5B grants the week of the summit) puts that data into models with the stated goal of doubling US scientific productivity over a decade.
Canonical: genesis-mission-to-ai-for-science-acceleration.
Evidence
- michael-kratsios in 2026-08-04-podcast-moonshots-michael-kratsios-on-the-new-golden-age-of (recorded 2026-07-29): "the most important thing for the nation is to make sure that we're applying this unbelievable technology called artificial intelligence to scientific discovery. And it's not just at one agency, it's across all of government."
- michael-kratsios (same source): "We essentially want to double the productivity of the entire scientific enterprise in the United States over the next decade."
- michael-kratsios (same source): "I think we announced about $5 billion worth of genesis mission grants just last week at the Genus mission summit. And so many of them relate to biotech."
- michael-kratsios (same source): "Our whole insight was that the US national labs have a tremendous amount of scientific data across a wide variety of domains that have not been made AI ready, that have not been uploaded to models and have not been part of the scientific process. And we have this huge opportunity before us to actually bring all that great data from 70 years of scientific discovery in the US into an AI model to be able to accelerate discovery."
- peter-diamandis (same source, show framing): "Genesis mission, a Manhattan Project style effort to accelerate breakthrough discoveries."
- michael-kratsios (same source) on the 2× target's vintage: "that's actually a number from, from last year when we were actually building what ultimately was building the program at the president signed as a Genesis mission in, in December."
Why it matters to this thread
- Directly in-scope: AI-for-science policy. Stock-market attached the $5B grants to a quantum-chips question; this page is the program, not a ticker.
- Complements agi-definitions-and-benchmark-saturation — Wissner-Gross (May 2026) already pushed unsolved-problem benchmarks at a DOE Genesis Mission event; Kratsios is the administration primary that the program exists and is funded.
- Complements automated-ai-research-llm-capability-boundary / ai-math-capability-jaggedness: Genesis is a bet that verification-rich scientific domains (chemistry, physics, biology, math) are where AI floods first — the same split Sanderson describes.
Why it matters to politics
Trump-era executive industrial policy: a presidential program (signed December) with ~$5B in grants the week of the summit, aimed at putting 70 years of national-lab data into models. Politics home for the program-as-policy is this page plus ai-action-plan; do not duplicate. $5B "last week" is a late-July 2026 snapshot from the program's architect — re-validate against a DOE/Treasury list. Sibling quantum EO is not Genesis.
Contradictions / tensions
- 2× vs moonshot 10×. Diamandis calls 2× "anti" in a moonshots frame; Kratsios kept it because it was already public. Not a factual contradiction — ambition vs already-announced target.
- Single administration primary. Interested party (OSTP director selling his own program). $5B "last week" is a dated claim (week of ~2026-07-22) — re-validate against a Treasury/DOE grant list.
- Quantum EO ("scientifically relevant quantum computer by end of term") is a sibling national mission, not Genesis itself — do not collapse them. Stock-market tracks the quantum leg.