RECLAIM Act
- Sponsor
- Sen. Moreno, Bernie [R-OH]
- Last action
- Oct 15, 2026
Bottom line
The RECLAIM Act seeks to end federal, state, and local financial incentives for large, AI-intensive data centers by prohibiting federal aid and taxing state/local subsidies.
What it actually does
This bill, known as the RECLAIM Act, prohibits federal financial assistance and tax incentives for large data centers that primarily utilize artificial intelligence (AI) accelerator chips. It also imposes a 100 percent excise tax on any state or local incentives received by such data centers, effectively nullifying the benefit of those incentives. The bill aims to eliminate subsidies for a specific class of high-power-demand AI data centers.
Proponents argue
Supporters argue that this bill will restore tax neutrality, protect public revenues, and eliminate distortions in business location decisions that currently disadvantage domestic manufacturing and other job-creating investments. They contend that large, AI-focused data centers, often operated by highly profitable corporations, do not require or deserve taxpayer-funded incentives, which they characterize as 'excessive corporate largesse.'
Opponents contend
Opponents would likely argue that prohibiting and taxing incentives could stifle innovation and investment in critical AI infrastructure within the United States, potentially driving data center development and associated jobs overseas. They might also raise concerns about federal overreach into state and local economic development policies, asserting that states should retain the autonomy to offer incentives to attract businesses and create jobs.
The bill is very short and straightforward, making it highly accessible for any reader to understand its core provisions and implications.
Section 2(a)
Prohibition on Federal Incentives
This provision explicitly states that 'covered data centers' are ineligible for any form of federal financial assistance or tax incentive. This includes grants, loans, loan guarantees, cooperative agreements, tax credits, deductions, exclusions, exemptions, or abatements. The aim is to prevent federal funds or tax breaks from supporting the development or commercial operations of these specific types of data centers.
Supporters argue
Proponents argue that federal incentives for large, AI-intensive data centers are an unnecessary subsidy to profitable corporations, distorting markets and diverting public funds that could be used for other priorities. Eliminating these incentives promotes fiscal responsibility.
Critics contend
Opponents might contend that federal incentives are crucial for fostering technological leadership and ensuring the U.S. remains competitive in AI development. Removing them could disadvantage American companies and encourage investment abroad.
Tradeoffs
The provision balances the desire to reduce corporate subsidies and federal spending against the potential for slowing domestic investment in advanced computing infrastructure.
Section 2(b) and new Chapter 50B of the Internal Revenue Code of 1986
Excise Tax on Prohibited State and Local Incentives
This section amends the Internal Revenue Code to impose a 100 percent excise tax on the amount of any state or local incentive received by a covered data center operator. This means that if a state or local government provides a grant, tax credit, or other financial benefit to a covered data center, the federal government will collect an equivalent amount in tax, effectively nullifying the incentive for the data center operator. This provision aims to discourage states and localities from offering such incentives.
Section 5000E(b)(2) of the Internal Revenue Code of 1986 (as added by Section 2(b))
Definition of 'Covered Data Center'
This provision defines what constitutes a 'covered data center' for the purposes of the bill. It includes facilities or integrated campuses that are designed to have a contracted peak power demand of at least 100 megawatts and where at least 50 percent of the installed compute accelerators are 'AI accelerator chips' used for AI training, fine-tuning, or inference. This specific definition targets large-scale, AI-intensive computing facilities.
Section 5000E(b)(2)(B) and Section 5000E(c) of the Internal Revenue Code of 1986 (as added by Section 2(b))
Regulatory Authority and Definitions
This provision grants the Secretary of the Treasury, in conjunction with the Secretary of Commerce, the authority to promulgate regulations. Specifically, they must define 'AI accelerator chip' within one year, considering various semiconductor devices like GPUs and ASICs. They also have the authority to adjust the power demand and AI accelerator chip thresholds for high-density AI facilities and to exclude de minimis computing sites. Further regulations are required to determine the value of incentives, coordinate tax liability, and specify the timing of incentive recognition.
Section 2(c)
Sense of Congress on State and Local Incentives
This section expresses the sense of Congress that states and their political subdivisions should promptly repeal or reform any tax incentive programs specifically designed for data centers. The stated goal is to restore tax neutrality, protect public revenues, and eliminate distortions in business location decisions that might disadvantage other sectors like domestic manufacturing.
The effective date of the entire Act is contingent upon the issuance of final regulations by the Secretary of the Treasury under Section 5000E(c).
Section 2(d)
Why it matters:This is standard legislative practice for bills that require significant regulatory implementation before they can take effect, ensuring the framework is in place. However, it also means the bill's actual impact is delayed and dependent on agency action.
Case for: Supporters would argue that this phased implementation is practical and necessary, allowing agencies sufficient time to develop the complex definitions and rules required for the bill to function effectively and fairly. Rushing implementation without clear guidance could lead to confusion and legal challenges.
Case against: Critics might argue that making the effective date contingent on future regulations introduces uncertainty and delays the intended policy outcome. It also gives significant power to unelected officials to determine when and how the law takes effect, potentially allowing for political maneuvering or indefinite delays.
Estimated impact: This provision means that the federal prohibitions and the excise tax will not be enforced immediately upon enactment, but only after the Treasury and Commerce Departments complete their regulatory work, which could take over a year.
The Secretary of the Treasury, in conjunction with the Secretary of Commerce, is mandated to promulgate regulations within one year to define 'AI accelerator chip' and must consider whether the definition should include graphics processing units (GPUs), tensor processing units (TPUs), application-specific integrated circuits (ASICs), and other semiconductor devices designed primarily for AI training or inference.
Section 5000E(b)(2)(B)(i) and (ii) of the Internal Revenue Code of 1986 (as added by Section 2(b))
Why it matters:This is a necessary delegation due to the highly technical and rapidly evolving nature of AI hardware. It allows expert agencies to craft a precise definition that can adapt over time, which is standard for technical legislation. However, it also means the ultimate scope of the bill is not fully determined by the legislative text itself.
Case for: Proponents argue that this delegation is essential for the bill's long-term effectiveness, as it allows the definition to remain current with technological advancements without requiring constant legislative updates. It leverages agency expertise for technical specificity.
Case against: Critics might express concern that this delegates significant policy-making power to unelected officials, allowing them to determine which technologies fall under the scope of the bill, potentially leading to arbitrary distinctions or unintended impacts on the AI industry.
Estimated impact: The precise definition of 'AI accelerator chip' will be a primary determinant of which data centers are classified as 'covered' and thus subject to the bill's prohibitions and taxes. This definition will directly shape the bill's economic impact on the AI sector.