EXIM has more than $100bn of unused statutory lending capacity that the White House wants channelled into US-built, full-stack AI export packages, with the Commerce Department running a public solicitation for industry-led consortia.
The Trump administration is moving to underwrite US-made AI exports with billions of dollars of federal export financing, the latest in a string of policy instruments that have shifted the US government’s AI strategy from one defined by export controls to one defined by export promotion.
The vehicle is the Export-Import Bank of the United States, which has launched a dedicated Powers American AI Exports programme calibrated to channel statutory lending capacity into full-stack AI deals abroad.
EXIM’s capacity for this is structurally large. The bank holds a $135bn statutory ceiling on outstanding loans and currently has roughly $34.1bn drawn, on the Institute for Progress’s running analysis.
The unused headroom of more than $100bn is the capacity pool the White House is now positioning the AI Exports programme to consume.
EXIM is also up for reauthorisation in 2026, with the proposed package potentially lifting the lending cap to $205bn, which would expand the available envelope materially before the AI-export draw begins to bite.
The programmatic structure is being run through the Commerce Department.
Paul Hastings’s client-alert summary of the three Trump AI executive orders describes a sequence in which the Secretary of Commerce, in consultation with the Secretary of State and the Director of the OSTP, was directed by 21 October 2025 to establish and implement the Export Program supporting the development and deployment of US full-stack AI export packages.
The Commerce solicitation, as the Institute for AI Policy and Strategy has tracked, is calibrated around industry-led consortia rather than individual-company applications.
Each proposal must cover AI-optimised hardware and infrastructure (chips, servers, accelerators), data-centre storage, cloud services and networking, and the application-software layer that runs on top.
The strategic logic is the part the administration has been most explicit about. The policy is a deliberate inversion of the China industrial-policy playbook, with the US government taking the role of underwriter for export-financed AI infrastructure deployments in third-country markets.
The instrument sits inside the broader AI-policy moves the administration has made across the past month. The voluntary 90-day pre-release model-disclosure framework expected to be signed this week is the domestic-side complement to the export-financing track.
The Trump-Xi Beijing summit on AI guardrails and the H200 licensing dispute have set the bilateral context that the export-financing programme is designed to operate inside.
The strategic-customer geography is, on the available materials, the Asia-Pacific and Gulf markets that have been the most receptive to US-built AI infrastructure deployments outside the China block.
Specific named customers have not yet been disclosed in the EXIM materials, but the policy framework maps onto the deals visible through the recent commercial-market record.
OpenAI’s $235m Singapore applied-AI lab and the wider Singaporean Smart-Nation procurement track are the kind of bilateral footprint the programme is calibrated to underwrite.
The European competitive frame, where France’s $10bn AION gigafactory bid sits inside the EU’s €20bn InvestAI envelope, is the most-visible alternative state-financing track the US AI-Exports programme is positioning against.
What the administration has not yet disclosed is the specific dollar allocation for the first AI-Exports tranche, the named consortia that have already responded to the Commerce solicitation, the destination-market priorities for the first cohort of approved deals, or the interest-rate and tenor terms EXIM will offer against the AI-export collateral.
The next visible proof point will be the first named consortium approved under the AI-Exports programme, expected before the end of the third quarter on the administration’s timeline.
