06-reference

moonshots how the white house plans to 10x scientific productivity

2026-08-04·reference·source: Peter Diamandis (Moonshots) (YouTube)·by Peter Diamandis / Michael Kratsios
ai-policygovernmentscience-policygenesis-missionagent-economicsl5-north-star

"How the White House Plans to 10x Scientific Productivity | Michael Kratsios | EP #276" — Peter Diamandis (Moonshots)

Why this is in the vault

This is the most senior US government AI/science-policy voice — the White House OSTP Director — laying out an AI-native reimagining of the scientific process itself (the Genesis Mission, autonomous self-driven labs, agent/DAO-run research marketplaces) and naming explicit 2x-to-10x productivity targets. It's directly relevant to RDCO's read on government AI adoption pace and to the L5 north star thesis that AI-native institutional redesign, not incremental tooling, is the real endpoint.

Episode summary

Peter Diamandis interviews Michael Kratsios, White House Director of the Office of Science and Technology Policy, at the White House. They cover the administration's "born free vs. born in captivity" regulatory philosophy, the six national technology missions (AI/Genesis, quantum, fusion, lunar, robotics, semiconductors), and a deep dive into the "Science: A New Golden Age" report — its diagnosis of declining scientific productivity and its proposed fixes (long-duration grants, fast grants, "golden ticket" review, meta-science units, and a speculative AI-agent/DAO-driven science marketplace).

Key arguments / segments

Notable claims

Guests

Michael Kratsios — 13th Director of the White House Office of Science and Technology Policy and science adviser to President Trump. Described in the episode as principal architect of America's AI Action Plan, the Genesis Mission (a Manhattan-Project-style AI-for-science initiative), and the "Science: A New Golden Age" report/blueprint for accelerating scientific discovery. Previously worked on AI policy dating to the first Trump administration's 2019 AI executive order.

Sponsorship

Three distinct third-party ad reads appear within the main interview: Google for Startups [~00:08:00] (generative AI model access for startups), Blitzy [~00:26:00] (autonomous AI software development agents), and Voice Run [~00:50:00] (voice-agent development platform). A fourth, Fountain Life [~01:10:00], sponsors Moonshots' separate recurring health segment appended after the Kratsios interview closes. This is distinct from the house self-promotion (Peter's "metatrends" newsletter link in the description), which does not itself trigger the sponsored flag.

Mapping against Ray Data Co

Kratsios's account is a useful upper bound on how fast institutional AI adoption can move even inside the most change-resistant large organization there is: LLMs are still banned on White House systems under records-retention law, grant review cycles run 18 months by default, and the flagship AI-for-science program's own official target is a modest 2x over a decade — yet the same official, under mild public pressure, concedes 10x is the right aim given six months of frontier-model progress. That gap between official target and stated belief is itself a data point: government AI adoption is bottlenecked by institutional/legal inertia (Presidential Records Act, grant bureaucracy, tenure-and-publish incentive structures) rather than by capability or ambition, which is consistent with the vault's existing "governments would rather benchmark it than license it" reading of AI policy (see [[2026-06-03-innermost-loop-benchmark-not-license]] cross-reference) but adds a sharper mechanism: the golden-ticket/meta-science/fast-grant proposals are literally an attempt to import startup-style fast-iteration and agent-economics thinking into a 1950-vintage institutional design. The AI-agent/DAO scientific-marketplace vision Kratsios endorses (bounty markets, agents hiring autonomous labs, smart-contract milestone payments) is structurally the same claim RDCO's L5 north star makes about AI-native institutional redesign generally — that the endpoint isn't AI-assisted versions of existing processes but agents replacing the coordination layer itself. This episode extends, rather than challenges, that thesis: if the White House's own science-policy shop is reasoning in these terms, the L5 trajectory is less contrarian and more consensus-forming among serious institutional actors than the vault's internal framing has assumed to date — worth flagging as evidence the timeline may be compressing.

Related