Key Takeaways
Introduction
The White House recently released an AI Action Plan alongside three executive orders aimed at accelerating AI deployment across industry, infrastructure, and government. The AI Action Plan outlines more than 90 federal government actions to be taken across three core areas—innovation, infrastructure, and global competitiveness—with an emphasis on streamlining data center permitting, supporting exports of U.S.-origin AI systems, and prohibiting the procurement of “woke” large language models (LLMs) by the federal government. Together, the package is an attempt to use federal AI policy to support faster implementation of AI technologies, shifting the focus from an emphasis on potential risks to an emphasis on deployment, investment, and international positioning while reducing regulatory friction.
The first core area of innovation includes deregulatory initiatives, workforce training, AI safety research, and expanded federal use of open-source models. The second core area of infrastructure includes backing for large-scale data center and related AI infrastructure development, grid modernization, and secure compute for both civilian and military use. The third core area of global competitiveness sets out a framework for advancing U.S. interests abroad through export controls, standards diplomacy, and compute restrictions tied to national security.
Three executive orders detail specific steps that the administration is directing agencies to take in these three core areas:
- Executive Order 14318, Accelerating Federal Permitting of Data Center Infrastructure, directs agencies to streamline permitting for “qualifying” data centers, defined as those with more than 100 megawatts of new electric load, with at least $500 million in capital investment, that protect national security, or that are otherwise identified by designated federal agencies. It prioritizes development on federal lands, brownfields, and superfund sites. It extends to related components such as transmission lines, dispatchable baseload energy sources (including backup power systems) constructed or used principally to serve the project, and semiconductor materials.
- Executive Order 14319, Preventing Woke AI in the Federal Government, directs agencies to apply “unbiased AI principles” when evaluating LLMs for federal use. The order prohibits procurement of models that incorporate content related to diversity, equity, and inclusion (DEI) or that generate outputs deemed ideologically skewed.
- Executive Order 14320, Promoting the Export of the American AI Technology Stack, establishes a new framework for exporting integrated U.S. “AI stacks,” including hardware, cloud services, data infrastructure, and application-layer models. It directs agencies to coordinate trade and diplomacy efforts to promote adoption of U.S. technical standards abroad.
Agencies are expected to begin implementing these directives through rulemaking and formal guidance over the coming months. Oversight will be coordinated by the Office of Management and Budget (OMB) in consultation with a new federal AI implementation task force.
Below is a more thorough discussion of the administration’s three new core areas of focus for AI: (1) innovation, (2) infrastructure, and (3) global competitiveness.
1. Accelerating U.S. Innovation
Relying on executive branch authority over how the federal government procures technology, the AI Action Plan promotes new funding opportunities, imposes procurement restrictions, and implements new procurement regulations and policies. These actions are intended to promote private sector-led innovation so long as the innovation aligns with the administration’s policy objectives that seek to deregulate AI and reduce perceived “woke” AI requirements.
Funding Opportunities
The government aims to incentivize the private sector’s development of AI by providing access to nontraditional funding opportunities. For example, the government plans to use the National Science Foundation’s National AI Research Resource Pilot—an AI research program open to federal grant recipients—to promote financial markets for startups and academics. In addition, the AI Action Plan encourages agencies to develop and scale foundational and translational manufacturing technologies through sources often reserved for developmental programs or emergent circumstances, such as the Small Business Innovation Research program, the Small Business Technology Transfer program, research grants, Creating Helpful Incentives to Produce Semiconductors and Science (CHIPS) research and development programs, Stevenson-Wydler Technology Innovation authorities, Title III of the Defense Production Act, Other Transaction Authority, and potentially other authorities. The AI Action Plan also emphasizes the need to fund cloud-based research, products, and services, including a direct call to the U.S. Department of Defense (DOD) to prioritize agreements with cloud-service providers and related entities. Entities working in the AI arena should be on the lookout for new grants and alternative contracting opportunities as sources to fund new AI initiatives.
Procurement Restrictions
The AI Action Plan recommends actions to disincentivize AI innovation that contradicts the administration’s policies. For example, according to EO 14319, the administration has determined that DEI ideologies displace truth and create bias, and in line with this view, the AI Action Plan calls for an update to federal procurement guidelines “to ensure that the government only contracts with frontier LLM developers who ensure that their systems are objective and free from top-down ideological bias.” The AI Action Plan also states that the administration has determined that bureaucratic red tape and excessive regulation hinder AI development, limiting America’s ability to lead AI innovation on the global stage. Therefore, the AI Action Plan calls upon federal agencies with AI-related discretionary funding to reduce funding to states with burdensome AI regulatory regimes. These tactics are similar to those used in other recent executive orders that call for the review and possible termination of federal funding that the administration determines does not align with its policies.
New Procurement Regulations and Policies
The AI Action Plan calls for the development of new AI policies and guidance, revisions to existing AI standards, and industry input on each. Notably, the AI Action Plan directs the General Services Administration (GSA), in coordination with the OMB, to create an “AI procurement toolbox” to create uniformity across the federal government when agencies are procuring or choosing between compliant AI models. Relatedly, the AI Action Plan calls on GSA to also implement an Advanced Technology Transfer and Capability Sharing Program for interagency transfer of advanced AI capabilities and uses. The development of an AI procurement toolbox aligns with previous guidance issued by OMB in M-25-22, which instructs agencies on AI acquisition standards. Additionally, the AI Action Plan requires revisions to any federal regulations that unnecessarily hinder AI development or deployment and specifically calls for revision of the National Institute of Standards and Technology (NIST) AI Risk Management Framework to eliminate references to misinformation, DEI, and climate change. In total, these changes to the procurement regulatory landscape may lead to new requirements in future solicitations and may result in modifications to existing contracts.
Open‑Source Licensing and Intellectual Property Exposure
The AI Action Plan states that open-source and open-weight models are critical for innovation, academic research, and global competitiveness. It directs the National Telecommunications and Information Administration to analyze licensing constraints and intellectual property risk factors that may limit adoption, especially for smaller developers, academic users, and U.S. export partners. The AI Action Plan does not prescribe solutions but signals potential policy guidance or support to reduce legal uncertainty and promote reuse of public models. These reviews could inform future agency action on model licensing, particularly in contexts involving shared infrastructure, federal funding, or international deployment.
Security Considerations
The AI Action Plan calls on government agencies to “collaborate with leading American AI developers to enable the private sector to actively protect AI innovations from security risks, including malicious cyber actors, insider threats, and others,” but it imposes no obligations or standards that would ensure AI developers adopt best practices to mitigate security risks.
Focus on Deepfakes
The AI Action Plan directs NIST and the U.S. Department of Justice (DOJ) to expand on the TAKE IT DOWN Act to address the growing risks posed by synthetic media (such as deepfakes) used as evidence in the legal system, signaling increased interest in regulating the proliferation of deepfakes. The AI Action Plan directs NIST to build on existing efforts by formalizing guidelines and voluntary forensic benchmarks to provide courts and agencies with reliable tools for detecting and evaluating AI-generated media. Additionally, DOJ is encouraged to issue guidance to agencies on adopting evidentiary standards for deepfakes, modeled after proposed Federal Rule of Evidence Rule 901(c).
State Preemption
The AI Action Plan tasks the Federal Communications Commission (FCC) with preempting overly burdensome state AI regulation. Under the FCC’s organic act, the Communications Act of 1934, as amended, the FCC has limited authority to preempt state laws that impair market entry by telecom service providers. It is unclear how this authority could be legally sufficient to preempt state AI laws, but following the AI Summit, FCC Chairman Brendan Carr announced that he would have the FCC’s Office of General Counsel identify state AI laws that would be worthy of FCC preemption. Relatedly, the AI Action Plan also calls for withholding discretionary federal funds from states that have overly burdensome AI regulations. While acknowledging that states have a right to protect their residents from harms and that the administration does not seek to prevent all state AI regulation, the AI Action Plan articulates the executive branch’s clear intent to use discretionary authority as a mechanism to discourage states from enacting overly burdensome AI laws and regulations.
2. Building AI Infrastructure
Under the AI Action Plan, the administration will seek to streamline environmental regulatory processes and incentivize building AI infrastructure on federal lands. Specifically, EO 14318 instructs multiple agencies to take certain actions for “qualifying projects,” including “data center projects” and “covered components.” As expected, the order identifies natural gas, nuclear energy, coal, and geothermal as the generation sources for the data centers, as well as other “dispatchable baseload energy sources.” The EO revokes President Biden’s EO 14141, which directed agencies to take similar actions but focused on clean energy to power the projects.
Siting on Federal Lands
EO 14318 directs the U.S. Department of the Interior (DOI) and the U.S. Department of Energy (DOE), after consulting with industry and the U.S. Department of Commerce (DOC), to offer authorizations to develop qualifying projects for sites DOI and DOE have identified under applicable authorities.
On July 24, 2025, the day after publication of EO 14318, DOE announced the selection of four sites—Idaho National Laboratory, Oak Ridge Reservation, Paducah Gaseous Diffusion Plant, and Savannah River Site—to move forward with plans to invite private sector partners to develop AI data center and energy generation projects. The site selection was informed by a Request for Information on Artificial Intelligence Infrastructure on DOE Lands, and it builds on an earlier EO directing the secretary of energy to designate DOE sites for use by privately funded nuclear reactor technology to power AI infrastructure. DOE will release more details on the project scope, eligibility requirements, and submission guidelines with site-specific solicitations in the coming months, with partners to be selected by the end of the year.
EO 14318 further directs DOD to identify sites on military installations for covered component infrastructure uses and to competitively lease available lands for qualifying projects to support DOD’s mission. This move aligns with a DOD Request for Information earlier this year seeking industry interest in leasing a plot of 620 acres on the Davis-Monthan Air Force Base being considered for development into a large-scale data center for commercial use and associated infrastructure and generation to support energy reliability and resilience.
Streamlining Regulatory and Environmental Review
Consistent with the administration’s priority of reducing regulatory review for any type of infrastructure project, EO 14318 seeks to expedite permitting for data centers, energy generation and electricity transmission, and additional infrastructure to serve those projects. As have past executive orders, EO 14318 focuses on categorical exclusions as one means to eliminate more extensive review under the National Environmental Policy Act (NEPA). Here, EO 14318 directs relevant federal agencies to identify any categorical exclusions already established or adopted that could facilitate construction of qualifying projects and to coordinate with the Council on Environmental Quality to establish new categorical exclusions to cover data center-related actions that “normally do not have a significant effect on the environment.” EO 14318 gives agencies wide discretion on the justification for establishing these new categorical exclusions; the only direction given is to “rely on any sufficient basis … as each agency determines” is appropriate. Moreover, based on the NEPA definition of “major [f]ederal action,” EO 14318 excludes loans, loan guarantees, grants, tax incentives, or other forms of federal financial assistance from environmental review where federal financial assistance represents “less than 50 percent” of total project costs. The percentage threshold is a significant expansion of existing authority and past precedent.
The U.S. Environmental Protection Agency is directed to:
- Expedite environmental permitting for qualifying projects by streamlining or reducing regulations promulgated under the Clean Air Act; the Clean Water Act (CWA); the Comprehensive Environmental Response, Compensation, and Liability Act; and other relevant laws. This is particularly significant given qualifying projects extend to “covered components,” including natural gas pipelines and electricity transmission, that could be located far from the data center project being powered and certain baseload power for which the administration is seeking to dramatically revise or rescind existing regulations.
- Identify brownfield and superfund projects for use by qualifying projects and develop guidance to help expedite environmental reviews for qualified reuse. Within 180 days, the Army Corps of Engineers (Corps) is directed to review nationwide permits issued under section 404 of the CWA and section 10 of the Rivers and Harbors Act to determine whether a nationwide permit (NWP) for qualifying projects would expedite construction rather than a developer having to seek multiple individual permits. The AI Action Plan further states that if a NWP for qualifying projects is adopted, the Corps should ensure that this permit does not require a pre-construction notification and covers development sites consistent with the size of a modern AI data center. Notably, a proposed rulemaking to reissue and modify nationwide permits is underway. Some commenters on the AI Action Plan made recommendations consistent with the EO’s directives, which suggest that the Corps may indeed determine that a new NWP is needed.
For any sites identified by DOI and DOE, the action agency is directed to initiate a programmatic consultation under the Endangered Species Act with DOI, DOC, or both for common construction activities that will occur over the next 10 years.
Finally, the executive director of the Federal Permitting Improvement Steering Council has the authority to designate the project as a “transparency project” under section 4370m2(b)(2)(A)(iii) of the Fixing America’s Surface Transportation Act of 2015 (FAST-41) upon identification by the relevant agency. Once designated, the qualifying project can be published on the Permitting Dashboard, which tracks project schedules and milestones. This action mirrors the administration’s previous actions for critical minerals. The executive director also is directed to expedite the transition of eligible qualifying projects to FAST-41 covered projects.
Financing Qualifying Projects
EO 14318 further directs the secretary of commerce, in consultation with the director of the Office of Science and Technology Policy (OSTP) and other relevant executive departments and agencies, to launch an initiative to provide financial support for qualifying projects, which could include loans and loan guarantees, grants, tax incentives, and offtake agreements. Notably, qualifying projects do not include those powered by wind and solar projects, consistent with an earlier EO seeking to eliminate any subsidies for those projects. Agencies are further directed to submit to the director of OSTP a list of existing financial support that can be used to assist qualifying projects, consistent with the protection of national security.
Integrating Cybersecurity With AI Planning
The AI Action Plan’s calls to “Bolster Critical Infrastructure Cybersecurity,” “Promote Secure-By-Design AI Technologies and Applications,” and “Promote Mature Federal Capacity for AI Incident Response” generally update existing cybersecurity approaches, such as information-sharing, secure-by-design principles, and promulgating incident-response best practices, to account for AI. The measures are not revolutionary, but they will help the government and industry proactively focus attention on AI in security planning instead of waiting for a specific problem to emerge or worsen.
This call to action, however limited, puts AI on the list of considerations that government agencies, cybersecurity professionals, and critical infrastructure operators must incorporate. Although these provisions recognize that critical infrastructure owners often “operate with limited financial resources,” the AI Action Plan does not offer direct financial support to critical infrastructure security, much of which is privately owned. And while the AI Action Plan calls for the development of standards and expansion of information-sharing, it lists recommendations and goals. It does not propose security mandates, and it does not build on the Biden administration’s proposals to incentivize secure development by shifting liability to producers and distributors of hardware and software. In sum, while the AI Action Plan highlights important cybersecurity considerations, it relies on voluntary information-sharing and adoption of best practices without providing incentives or imposing mandates.
3. Advancing International Competitiveness
The AI Action Plan directs the government to “drive adoption of American AI systems, computing hardware, and standards throughout the world.” Through the AI Action Plan and the accompanying EO 14320, the administration emphasizes countering foreign influence over AI by tightening some export controls, such as over semiconductors, while promoting the proliferation of AI abroad through “full-stack export packages.”
Structure of the American AI Exports Program
EO 14320 directs the DOC to implement within 90 days an “American AI Exports Program” to support the development and deployment of “full-stack AI export packages.” DOC is to consult with the U.S. Department of State and OSTP to develop the program. DOC is instructed to identify domestic suppliers through a request-for-proposals process from industry-led consortia for possible inclusion in the program.
Required Components of an “AI Stack” and Consortium Eligibility
Proposals must include a full-stack AI package submission including: (1) AI optimized hardware, data center storage, cloud services, and networking, with a description of extent manufactured in the United States; (2) data pipelines; (3) AI models and systems; (4) security and cybersecurity for AI models and systems; and (5) use case-driven AI applications (e.g., agriculture, healthcare, etc.). Proposals must also identify countries/regions for export engagement, describe the business and operational model, and detail requested federal support. Proposals selected will be considered “priority AI export packages” and given access to tools such as financing, resources to accelerate deployment, export promotion of American AI technologies, and other marketing, financial, and promotional tools.
Export Compliance Implications for U.S. Developers and Cloud Vendors
EO 14320 does not specify what controls will apply to tech-stack exports, other than stating that exports must comply with U.S. export control laws and guidance from DOC’s Bureau of Industry and Security (BIS). However, it is possible that BIS may relax applicable controls on specific elements of such full-stack exports to promote export growth under the program.
Federal Funding for Scientific Research To Promote Biosecurity
With respect to concerns over biosecurity, the AI Action Plan implements new requirements for recipients of federal funding “to use nucleic acid synthesis tools and synthesis providers that have robust nucleic acid sequence screening and customer verification procedures” to promote biosecurity developments and reduce related threats. Note that the government intends to verify compliance through enforcement mechanisms rather than voluntary attestations.
Conclusion
The three AI-related executive orders signed by the president to support the AI Action Plan are largely focused on directing the federal government’s actions in the areas of procurement, national policy priorities, and matters of foreign affairs, international trade, cybersecurity, and national security. In many cases, we will have to await further actions by federal agencies to carry out the intent of the executive orders. Nevertheless, the AI Action Plan gives fairly clear direction on what we can expect from federal agencies in the coming months and years. Parties with business interests implicated by the AI Action Plan, such as those who (1) sell AI systems to the federal government, including, in particular, the DOD; (2) export AI systems; or (3) will be involved in the development of new AI data center projects, including on federal land, should actively monitor opportunities presented by federal agencies in the coming months. In the meantime, stakeholders should be aware that while many state AI laws and regulations might be in the crosshairs of the administration, they will remain legally binding unless they are preempted, repealed, amended, or overturned. Further, developers of AI systems who sell to the federal government should remain mindful of any material biases that may call into question the objectivity of their LLMs.
[View source.]
