The Full AI Stack Export Promotion Act directs federal agencies to actively promote the export and global adoption of the "U.S. full AI stack" — defined to include U.S.-designed AI semiconductors, cloud services, data centers, trained models, and technical standards. The bill tasks the Departments of Commerce and State, in coordination with Defense, Energy, the Intelligence Community, and industry consortia, to create programs, diplomatic strategies, security protocols, and measurement tools that accelerate foreign deployment of U.S. AI infrastructure while attempting to prevent illicit access by foreign adversaries.
This is a targeted industrial-diplomatic package: it couples export-promotion mechanisms (industry consortia, hotlines, diplomacy, and confidence-building standards) with security-oriented requirements (security measures for foreign purchasers, standardized deployment practices) and oversight tools (studies and a biannual "AI export success tracker"). The measure signals a federal preference for U.S.-led AI stacks and creates new reporting and coordination duties for multiple agencies — which matters for exporters, cloud operators, chip designers, allied governments, and national-security policymakers assessing tradeoffs between economic influence and risk of technology transfer.
At a Glance
What It Does
The bill requires Commerce to run a program that vets industry consortia proposals to facilitate exports of U.S. AI hardware, cloud services, models, and standards, and directs State to produce a diplomatic strategy to open foreign markets. It mandates interagency studies and reports, establishes security obligations for foreign purchasers, and creates a public, biannual metric (the "AI export success tracker") to quantify U.S. share of global AI capacity.
Who It Affects
U.S. semiconductor designers and manufacturers, cloud providers and data-center operators headquartered in the U.S., AI model developers, and allied governments targeted as purchasers are directly implicated. Federal agencies (Commerce, State, Defense, Energy, DNI) must coordinate new programs and produce multiple reports and studies.
Why It Matters
The bill moves AI industrial policy into diplomacy and export promotion rather than pure trade or export-control channels. That creates new levers for U.S. firms to expand abroad, but it also formalizes government involvement in standards-setting, security conditions on purchasers, and public measurement of market share — all of which will reshape commercial strategy and national-security assessments.
More articles like this one.
A weekly email with all the latest developments on this topic.
What This Bill Actually Does
The Act establishes an active federal program to push U.S.-developed AI infrastructure into allied and partner markets. Commerce must set up a competitive process for short-lived industry consortia that present proposals meeting U.S. security requirements; those consortia can only be formed for purposes of participating in the program, limiting standing trade associations from simply enrolling as-is.
State is directed to develop a diplomatic strategy that both identifies priority export markets and lays out how U.S. officials will persuade foreign governments and international bodies to favor U.S. technology and standards.
On the security side, Commerce — working with State, Defense, and Energy — must define and negotiate security measures with foreign purchasers to prevent unauthorized access or transfer to foreign adversaries. That includes efforts to detect and eliminate adversary hardware and software in buyer supply chains, and to require deployment practices that minimize avenues for remote exfiltration.
The bill contemplates both unclassified reporting to Congress and classified annexes where necessary.The legislation also standardizes how the federal government will measure global AI deployment. It codifies technical definitions (for example, "national computing capacity" measured in FLOP/s normalized at a precision set by Commerce, and "national memory bandwidth" measured in sustained bytes/second under benchmark conditions) and requires an "AI export success tracker" with estimates of installed capacity, share of U.S.-designed AI ICs, proportion of U.S.-operated data centers, token-processing share for U.S. models, and cloud revenue shares.
Commerce, the DNI, and State must publish that tracker biannually for five years, producing a public view of U.S. traction in the global AI market.Finally, the bill asks agencies to produce studies and "confidence" standards: Commerce must work with the public and identified consortia to develop product practices, offerings, or standards that assure large foreign purchasers about privacy, confidentiality, security, and effectiveness. Several reporting deadlines are short (many actions and reports are required within 180 days), which pressures agencies to move quickly to operationalize both promotion and security components.
The Five Things You Need to Know
Multiple required reports and program launches must occur within 180 days of enactment — the bill repeatedly uses a 180-day deadline for strategy, program status, studies, security plans, and confidence standards.
Commerce must form a program that accepts proposals only from industry consortia created solely to participate in the program, preventing pre-existing groups from automatically qualifying.
The AI export success tracker must be produced biannually for five years and include estimates of national computing capacity (FLOP/s), national memory bandwidth, the share of AI integrated circuits designed by U.S. firms, U.S.-operated data-center share by region, and model-token processing share attributable to U.S. firms.
Reports under the Act must be submitted in unclassified form but may include classified annexes, allowing sensitive security details to be withheld from the public while meeting congressional reporting requirements.
Definitions are technical and prescriptive: 'national computing capacity' is to be normalized at a precision level set by Commerce, and 'U.S. artificial intelligence semiconductor products' require that design activities occurred in the United States.
Section-by-Section Breakdown
Every bill we cover gets an analysis of its key sections.
Statement of policy: prioritize U.S.-led full AI stack
This section lays out the bill’s national-policy objectives: dominance in global AI deployment, adoption of the U.S. full AI stack by allies, denying adversaries access to advanced compute, and keeping a majority of compute capacity in the U.S. These policy declarations guide agency discretion but do not, by themselves, create enforceable sanctions or procurement requirements — instead they act as the statutory rationale for the programs, diplomatic efforts, and reporting mandates that follow.
Industry consortia program to package and export the stack
Commerce must create a program that solicits and evaluates proposals from industry consortia set up expressly for participation. The provision limits eligible consortia to entities formed only to engage with the program, a device meant to encourage tailored public–private project proposals (e.g., bundled offers of chips, cloud services, and deployment services) rather than existing trade groups. Commerce also must report back to Congress within 180 days on the program’s status and results, forcing an early accountability checkpoint.
Diplomatic strategy and industry-facing barrier removal
State, with Commerce’s input, must run outreach to clear export barriers: industry listening sessions, a hotline for export issues, and elevation of problems through diplomatic channels. The statute requires State to produce a diplomatic strategy within 180 days that identifies how to ease market access for U.S. AI firms, promote U.S. standards in international fora, and prioritize target countries — transforming export promotion into an explicit diplomatic line item.
Interagency study on global deployment impacts
State, DNI, and Commerce must complete a study on economic, diplomatic, and security impacts of global deployment of the U.S. full AI stack and deliver an unclassified report (with a classified annex option) within 180 days. The study’s scope includes benefits to partner countries, comparative competitiveness, security implications for U.S. and allied qualitative military superiority, and prioritization of regions for export efforts — providing a foundation for the diplomatic strategy and security measures.
Security of U.S. AI semiconductor products deployed abroad
Commerce, coordinated with State, Defense, and Energy, must negotiate security measures with foreign purchasers to prevent unauthorized access and transfer to foreign adversaries. The mandated reporting must describe standardized security requirements, agreements with foreign governments, measures to block remote access transfer, and analysis of adversary components in buyer supply chains — effectively exporting a U.S. playbook for secure deployment.
Confidence initiative for major purchasers
Within 180 days Commerce must develop, with other agencies and public input, generally applicable practices, product offerings, or standards to reassure large foreign buyers about privacy, confidentiality, security, and effectiveness. This is a market-facing effort meant to lower commercial friction — for example, model-licensing terms, audit capabilities, or technical assurance packages — but details and enforceability are left to agency implementation.
AI export success tracker: biannual metrics and public reporting
Commerce, DNI, and State will estimate and publish several quantitative metrics every six months for five years: installed AI compute and memory bandwidth by country, share of AI ICs designed by U.S. firms, share of data centers owned/operated by U.S. companies, token-processing share for U.S. models, and cloud revenue/capacity share. Reports are unclassified with the possibility of classified annexes. The provision formalizes specific measurement axes that will shape policy evaluation and industry incentives.
This bill is one of many.
Codify tracks hundreds of bills on Foreign Affairs across all five countries.
Explore Foreign Affairs in Codify Search →Who Benefits and Who Bears the Cost
Every bill creates winners and losers. Here's who stands to gain and who bears the cost.
Who Benefits
- U.S. AI semiconductor designers: The bill prioritizes 'U.S.-designed' integrated circuits and asks agencies to quantify and promote that share, increasing market opportunities and diplomatic leverage for U.S. chip IP owners.
- U.S.-headquartered cloud providers and data-center operators: The statute explicitly aims to expand the footprint of U.S.-operated data centers and cloud services abroad, creating new commercial opportunities and possible government-supported deals in targeted partner countries.
- U.S. AI firms and export consortia: The consortia program, export hotlines, and diplomatic strategy lower transactional and regulatory barriers for bundled offers of chips, models, and services to allies.
- Allied governments seeking secure, interoperable AI infrastructure: Partners that prefer U.S. standards and secure deployment will gain access to packaged offers, security protocols, and confidence standards tailored to reassure large national purchasers.
- Standards organizations and professional bodies: The emphasis on U.S. technical standards and the confidence initiative creates demand for interoperable, auditable standards and certification services supplied by U.S.-aligned bodies.
Who Bears the Cost
- U.S. firms that export the stack: They will face new government-prescribed security expectations, reporting touchpoints, and possible conditions negotiated by Commerce or State with foreign buyers, increasing compliance and transaction costs.
- Foreign purchasers: Buyers accepting the U.S. full AI stack may have to implement supply-chain security measures, accept U.S.-approved deployment practices, or host oversight tools — imposing operational and procurement burdens.
- Federal agencies (Commerce, State, Defense, Energy, DNI): Agencies must deliver multiple studies, strategies, programs, and biannual trackers on accelerated timelines, requiring staffing, technical capacity, and interagency coordination.
- Non-U.S. competitors and international standards diversity: The Act’s clear preference for U.S.-led stacks and standards reduces market space for other architectures, potentially increasing geopolitical friction and prompting retaliatory industrial policy.
- Privacy advocates and civil-society groups in purchaser countries: Confidence requirements and security controls could prioritize access and surveillance tradeoffs that domestic advocates might view as constraints on sovereign data governance or transparency.
Key Issues
The Core Tension
The central dilemma is this: accelerate global adoption of a U.S.-led AI infrastructure to maximize economic influence and allied interoperability, while simultaneously preventing that very diffusion from enabling adversary access or eroding U.S. security advantages — a classic trade-off between wider deployment for strategic influence and tighter controls to limit proliferation.
The bill packages promotion and security together, but executing both simultaneously creates real operational tensions. Rapid export and market-share gains require streamlined commercial offers and low friction in procurement; by contrast, robust protections against illicit access often mean intrusive technical controls, export conditions, and ongoing verification measures that raise costs and slow deployments.
Agencies must design standards that are both convincing to security-conscious partners and commercially viable for vendors.
Measurement and attribution are another practical headache. Codifying metrics like FLOP/s normalized at a Commerce-determined precision and sustained memory bandwidth under benchmark conditions gives the government tools to quantify market share, but these metrics can be gamed by vendors optimizing for benchmarks rather than real-world utility.
Token-processing share and ownership definitions (e.g., what counts as a model 'owned or operated' by a U.S. firm) will require careful methodology; otherwise the tracker risks producing numbers that mislead policymakers or skew incentives. Finally, the combination of public reporting with classified annexes creates an accountability gap: Congress and the public will see headline figures, while crucial security tradeoffs and bilateral agreements may be concealed, making independent assessment difficult.
Try it yourself.
Ask a question in plain English, or pick a topic below. Results in seconds.