US AI commission makes government a larger buyer of AI

A 756-page report asks Washington to double AI research funding each year to $32 billion and secure chip supply. Public money becomes an AI demand driver.
The Defense Department should spend $8 billion a year on AI research by 2025, up from about $1.5 billion, under the plan in the 756-page final report that the National Security Commission on Artificial Intelligence approved on 1 March for Congress and the President. The commission, created under the 2019 defense authorization act, warns that without action the US stands to lose its AI leadership to China within the next decade.
The report is a set of recommendations, not law, but it sets the agenda for the budget and chip bills Congress will write, and its direction is plain: the state is turning into a large buyer of AI. If adopted, federal spending becomes a durable source of demand for AI research, compute and applied systems. Government money behaves differently from private money. Procurement is slow, but contracts run for years and do not follow funding cycles, so AI-first companies able to sell to government build revenue less exposed to swings in private capital.
The targets show the size of the shift. Non-defense federal AI research funding would double every year to reach $32 billion by fiscal 2026. The defense target alone implies more than a fivefold rise in AI research spending, and the department and the intelligence agencies are to be AI-ready by 2025. National security now shapes AI markets as directly as any product cycle.
Chips carry the second half of the plan. The commission wants the US to stay at least two generations ahead of rivals in microelectronics, with tax incentives for leading-edge chip plants at home, because reliance on Taiwan for advanced chips is growing into a strategic risk. A week earlier the President ordered a 100-day review of supply chains, semiconductors first among them. Public money is starting to address how concentrated advanced chip production has become.
Selling to this buyer takes a particular kind of company. Founders who meet government security, procurement and testing requirements find a customer with multi-year budgets, at the price of long sales cycles and strict compliance. The report also points to gaps in talent and tooling, including a proposed digital service academy to train civil servants in technology. Dual-use products stand out: tools to test and evaluate machine learning systems, secure data pipelines and models that run on devices at the edge.
Commissions write reports; budgets make markets. The microelectronics advice becomes policy only if Congress funds incentives for domestic chip plants, and the research targets count only if the next federal budget request moves AI funding toward them. For startups the deciding change is quieter: whether defense agencies alter how they buy software so that young AI companies can win contracts directly.
The state is becoming a long-term buyer of AI and a co-investor in the chips it runs on. Fanaura counts that public demand as a base layer in its venture thesis, and these running notes on AI markets will track whether the money arrives. The commission has said what Washington should buy; the next budget will show what it pays for.
Building an AI-native company? Apply in five minutes. →
More from Fanaura
Building dual-use machine learning products that can pass government procurement? Send us the project.
Fanaura invests its own capital and the capital of its limited partners (LPs), usually through a separate SPV for each deal. If we see a fit, we get in touch.
Get the news by email
One short note each weekday. No marketing, and you can unsubscribe at any time.