La science comme nouvelle architecture de la puissance americaine
The White House Office of Science and Technology Policy (OSTP) report of 21 July 2026 — directed by Michael Kratsios — positions itself as the heir to Vannevar Bush’s Science: The Endless Frontier (1945). Eighty years later, the ambition is to refound the American scientific ecosystem for the age of AI and global technological competition. This is not sectoral science policy; it is grand strategy.
The United States cannot retain global pre-eminence if it loses its scientific and technological edge — an edge now contested by coordinated competitors, bureaucratic inertia in its own research system, and the rupture introduced by AI. The report integrates research, industry, defence, economy and geopolitical rivalry into a single architecture of power.
The Bush model — basic research → applied research → development → industrialisation — no longer fits. The boundary between basic and applied research has blurred, and private industry has become the primary engine.
In 2024, the United States devoted about $886 billion to R&D, of which nearly $700 billion was funded by private enterprise — more than federal government and higher education combined.
Consequence: the centre of gravity of innovation shifts from the university to a hybrid ecosystem associating state, industry, national laboratories, venture capital and new tech actors.
American science must no longer only produce knowledge, but convert it rapidly into economic, industrial and military capabilities — shortening the distance between discovery, experimentation, production and deployment.
Critical technologies developed on American soil must also be prototyped, manufactured and deployed at scale domestically. The report also advocates applying the scientific method to science itself — experimenting with funding mechanisms, evaluating results and reforming what no longer works — and broadens “scientific capital” to engineers, technicians and entrepreneurs.
The OSTP does not see AI as a productivity tool for researchers, but as a factor that may transform the very nature of scientific production.
Formulate hypotheses, design experiments, write software, pilot laboratory equipment, contribute to proofs of theorems — even closed-loop autonomous laboratories where AI adjusts its own experimental protocols.
Genesis Mission (launched November 2025): linking supercomputers, AI models, scientific instruments and national-lab databases into a national discovery infrastructure, with the stated ambition to double American scientific and technical productivity in a decade.
Advanced materials, biotechnologies, semiconductors, quantum technologies, fission and fusion — all fields deemed strategic and targeted by the Genesis infrastructure.
The report’s wager: as AI narrows the gap between top researchers and their competitors, the decisive advantage shifts to those who translate ideas into operational technology fastest.
A state can have excellent universities and still lose the technological competition for lack of large-scale production capacity; a powerful industry dependent on foreign technology remains vulnerable. Mastery of the whole chain is the condition of technological power.
The report develops a doctrine of permissionless innovation aimed at reducing regulatory obstacles to experimentation via controlled-environment testing mechanisms before generalisation — a deliberate bet that speed, managed correctly, outweighs precautionary delay.
Sovereignty is no longer limited to territory, natural resources or borders. It now encompasses models, data, compute, supply chains and the ability to keep the translation chain onshore.
The report’s architecture — top-tier universities, abundant venture capital, big tech, national labs, deep financial markets and military power — is hardly reproducible elsewhere. The question for others is not to copy it, but to identify transposable elements adapted to their own means: sovereign translation capacity, hybrid funding, and control over models, data and compute as components of sovereignty.
1. Bloc logic: emphasis on “domination” risks fragmenting international scientific cooperation (AI, biotech, quantum, autonomous weapons).
2. Externalities: permissionless innovation may underprice negative externalities — speed cannot fully substitute for control mechanisms.
3. New dependence: by placing AI at the centre of the scientific infrastructure itself, the model creates dependence on control of models, data and compute, with concentration risks.
4. Reproducibility gap: the American model’s breadth makes direct imitation illusory; strategy lies in selective adaptation.
The report nevertheless exceeds the American framework alone: for middle powers, it is a reminder that in the age of AI, drones, autonomous systems and quantum technologies, scientific research can no longer be treated as sectoral policy.
Author: Andre Hubert Onana Mfege, PhD in Military History and National Defence Studies.
Tribune: No. 1867, published 23 September 2026 — Defence Nationale, OSTP Report under Michael Kratsios, Science: A New Golden Age (21 July 2026), heir to Vannevar Bush, Science: The Endless Frontier (1945).
Source: www.defnat.com — QuarkXPress PDF, 4 pages (FR).
