Senator Bernie Moreno’s (R-OH) recently introduced legislation, the RECLAIM Act (S. 5689), seeks to dramatically reshape the economic landscape for artificial intelligence (AI) data centers across the United States. The bill proposes a stringent 100 percent federal tax on any state or local "incentive" received by large AI data center operators. This measure, by its current reading, aims to effectively "top up" the sales tax in states that currently exempt data center servers and chips from such levies, by imposing a federal tax equivalent to the exempted amount. This "top-up" mechanism draws a striking parallel to the principles of global minimum taxation, where a federal layer ensures a baseline tax burden regardless of local concessions.
The proposal has ignited a significant debate regarding state fiscal autonomy, sound tax policy, and the federal government’s role in local economic development. At its core, the bill challenges the long-standing practice of states using tax exemptions and other incentives to attract high-tech investments, arguing that these provisions amount to taxpayer subsidies for large, profitable corporations. However, critics contend that the bill misunderstands the fundamental principles of a well-designed sales tax, particularly concerning the taxation of business inputs.
Understanding the RECLAIM Act’s Core Provisions
The RECLAIM Act specifies that the federal tax would be "equal to 100 percent of the amount of any prohibited State or local incentive received by the data center operator." This applies exclusively to "covered data centers," defined as certain large facilities primarily focused on AI operations. The legislation also includes a federal component, which would impact the federal tax treatment of these data centers, though the most contentious aspect lies in its direct intervention in state and local tax policy.
A close examination of the bill’s language reveals its potentially far-reaching scope. The definition of "prohibited state or local incentives" is exceptionally broad, encompassing "any state or local grant, loan, loan guarantee, cooperative agreement, tax credit or deduction, exclusion, exemption, abatement." This expansive wording suggests that the bill’s impact could extend beyond mere sales tax exemptions, potentially recharacterizing a covered data center’s income tax into something akin to a tax on gross receipts. For instance, standard business deductions for wages or capital depreciation, typically allowed in income tax calculations, could theoretically be interpreted as "deductions" or "exemptions" falling under the federal top-up. If such an interpretation were to hold, it would fundamentally alter the taxation of these businesses, imposing a punitive and economically distortive regime. To avoid unintended consequences, clarification on which types of deductions or exemptions are not covered by the policy would be crucial.
Senator Moreno’s Stated Intent and Background
Senator Moreno has been explicit about the rationale behind his legislation. In a press release announcing the bill, he stated unequivocally: "Taxpayers will not subsidize data centers, period." He further elaborated that the bill would specifically target "any state or local grant or tax exemption that hyperscalers use for data centers." Prior to introducing the RECLAIM Act, Moreno had publicly criticized specific sales tax exemptions, notably singling out a $4.5 million exemption on equipment for a data center project, which he controversially labeled a "$4.5 million subsidy" for a project creating "just 10 jobs." This perspective underscores his belief that such exemptions are not merely sound tax policy but rather constitute unwarranted corporate welfare, shifting the tax burden onto ordinary citizens.
This framing positions the bill as a measure to ensure that large AI data centers, which are often operated by highly profitable technology giants, contribute their "fair share" to public coffers, rather than benefiting from what Moreno perceives as preferential treatment. The rapid growth of the AI sector and the substantial investments in data center infrastructure have brought increased scrutiny to the incentives offered by states vying for these facilities. These centers, while critical for technological advancement, are often capital-intensive but not necessarily high-employment generators, leading some policymakers to question the return on investment for significant tax breaks.
The "Top-Up" Mechanism: A Global Precedent
The RECLAIM Act’s "top-up" mechanism bears a strong resemblance to the global minimum taxation framework (Pillar Two of the OECD/G20 Base Erosion and Profit Shifting, or BEPS, project). Under global minimum taxation, if a multinational corporation pays less than a globally agreed minimum effective tax rate in a particular jurisdiction due to local incentives or low statutory rates, other jurisdictions (typically the parent company’s home country) can impose a "top-up" tax to bring the effective rate up to the minimum.
Similarly, Moreno’s proposal aims to ensure that if a state or locality provides an incentive that effectively reduces the tax burden for a large AI data center below a certain federal threshold (in this case, by exempting sales tax on inputs), the federal government would step in to collect the difference. The underlying logic is that if states choose to offer such exemptions, the revenue saved by the data center would not simply disappear but would instead be captured by the federal government. This creates a powerful incentive for states to reconsider their exemption policies, as they might prefer to collect the revenue themselves rather than allowing it to flow to the federal treasury.
Impact on States: A Widespread Effect
Analysis suggests that the RECLAIM Act would have a profound impact on at least 31 states. These are jurisdictions that currently provide sales tax exemptions for data center equipment, often through conditional or discretionary programs. Such policies are clearly identifiable as "exemptions… for the development or commercial operations of a covered data center" under the bill’s proposed language.
States like Ohio, the home state of Senator Moreno, have already begun to pause new exemptions, indicating an awareness of the shifting political and economic landscape surrounding these incentives. However, even in such states, exemptions previously granted would still fall under the purview of the RECLAIM Act.
A smaller number of states (two, by some assessments) have policies structured closer to "by-right" exemptions, making it less clear whether they would be interpreted as "incentives" under the proposed law. This ambiguity highlights another area where legislative clarification would be necessary to avoid protracted legal challenges and uncertainty.
Conversely, 18 jurisdictions would likely face no immediate top-up on new projects. This group includes states like California and New Jersey, along with 11 others (and the District of Columbia), which already tax data center equipment at full sales tax rates. Additionally, five states, including Oregon, operate without a statewide sales tax altogether, thus having no sales tax exemptions to be "topped up" by the federal government. This disparity underscores the uneven impact of the bill, potentially penalizing states that have actively used such incentives to build their tech economies.
The Economic Rationale: Exempting Business Inputs from Sales Tax
A central point of contention raised by critics of the RECLAIM Act is its fundamental misunderstanding of sales tax design. From an economic perspective, a retail sales tax is intended to be a tax on final consumption, applied only once at the point of sale to the end-user. A well-designed sales tax system explicitly exempts all intermediate business inputs.
The reason for this exemption is to prevent "tax pyramiding." Tax pyramiding occurs when a sales tax is applied to business inputs at various stages of the production process. When this happens, the tax is embedded into the price of those inputs, and then taxed again when the final product or service is sold to the consumer. This leads to a cascade of taxes, where the effective tax rate depends on the number of transactions in the supply chain, making the tax system non-neutral and opaque. Consumers ultimately pay a hidden, higher tax burden, and businesses with longer supply chains are disproportionately penalized.
Data center equipment, including servers, networking gear, and specialized chips, are unequivocally business inputs. They are essential tools for producing services (data processing, cloud computing, AI model training) that are then sold to other businesses or consumers. Most states already recognize this principle for other industries; for example, 41 states exempt manufacturing machinery and equipment from sales tax by right. The distinction for data centers often lies merely in the legal form of the exemption – frequently granted through special, conditional policies rather than broad, by-right provisions for all business inputs. Economically, however, the principle remains the same: taxing these inputs distorts investment decisions and increases the cost of doing business.
The Tax Foundation’s State Tax Competitiveness Index, a widely respected measure of state tax policy, consistently scores states more highly when they exempt data center equipment from sales tax. Their rationale is that such purchases constitute "a business-to-business transaction that should rightly be exempted along with other business inputs." By targeting these exemptions, the RECLAIM Act risks counteracting states’ efforts to adopt economically sound sales tax treatment for a critical component of the modern economy.
The Unique Burden on Fast-Depreciating Assets: AI Chips
The economic impact of applying sales tax to data center equipment is further exacerbated by the rapid depreciation and technological obsolescence of AI-specific hardware, particularly advanced chips. A sales tax on equipment can be elegantly analogized as a uniform negative investment tax credit. While an investment tax credit at rate k effectively cuts the price of a machine by k percent, a sales tax at rate s conversely raises it by s percent.
Economists like Emil Sunley (1976) and David Bradford (1978) have demonstrated that such uniform credits or taxes are not neutral across assets with different lifespans. They disproportionately favor or penalize short-lived assets, respectively. A short-lived asset, such as an AI chip that becomes obsolete or wears out quickly, incurs the sales tax much more frequently each time it is repurchased. This translates into a significantly higher effective tax burden over the asset’s economic life.
Consider a typical sales tax rate of 7.5 percent. While this might appear modest, its application to rapidly depreciating AI chips – which often need replacement every 3 to 6 years due to technological advancements or wear-and-tear – accumulates rapidly. Using economic models that calculate the marginal effective tax rate, it can be shown that this 7.5 percent sales tax alone could amount to an effective tax rate of approximately 37 percent on the return on an AI chip. This is a stark contrast to federal tax rates on new equipment investment. For comparison, the Congressional Budget Office (CBO) estimates that federal taxes impose a marginal effective tax rate of about 4 percent on new investment in equipment in 2026, largely due to the allowance of full expensing under the 2025 One Big Beautiful Bill Act. Even under prior law with 20 percent bonus depreciation, the rate was around 16 percent. This substantial discrepancy highlights how a seemingly small state sales tax can impose an enormous effective burden on the cutting-edge, short-lived assets central to AI innovation.
Broader Implications: Fiscal Federalism and Economic Development
Beyond sales tax exemptions, the RECLAIM proposal’s broad language could also target other state and local incentives, such as property tax abatements. While some property tax abatements might be seen as non-neutral tax subsidies that are difficult to justify economically, the question of whether the federal government should intervene in such cases touches upon fundamental principles of fiscal federalism. The United States has a long tradition of states and localities having significant autonomy in shaping their tax policies to attract investment and stimulate local economies.
The debate around the RECLAIM Act thus becomes a microcosm of larger discussions about federal overreach versus state sovereignty. States argue that they are best positioned to understand their local economic needs and to design incentive packages that foster job creation and technological growth. Federal intervention, particularly when it targets what many economists consider sound tax policy (like exempting business inputs), could be seen as an undue imposition that stifles state-led economic initiatives.
The AI data center industry represents a massive, rapidly expanding sector. Attracting these facilities is often viewed as a win for states, bringing significant capital investment, though not always a large number of direct jobs. The energy demands of these centers are also substantial, leading to other debates around infrastructure and environmental impact. The RECLAIM Act, if enacted, could fundamentally alter the calculus for states considering hosting AI data centers, potentially deterring investment in states that previously relied on tax incentives to compete. This could lead to a concentration of AI infrastructure in states with already high tax burdens or no sales tax, rather than a more dispersed, geographically diverse development.
Conclusion and Future Outlook
Senator Moreno’s RECLAIM Act presents a bold challenge to current state and local tax incentive practices for large AI data centers. While the stated intent is to curb what he views as unwarranted subsidies, the bill’s broad language and its targeting of economically sound sales tax exemptions for business inputs raise significant concerns among tax policy experts and state governments. The potential for the federal government to "top up" state sales taxes, effectively forcing states to either abandon their exemptions or lose revenue to the federal treasury, would mark a significant shift in fiscal federalism.
As the bill progresses through the legislative process, further clarification on its scope and definitions will be crucial. The debate will likely center on the economic principles of sales tax design, the balance between federal intervention and state autonomy, and the role of tax incentives in fostering innovation and economic development in the rapidly evolving AI landscape. The outcome of the RECLAIM Act could set a precedent for how the federal government interacts with state tax policies, with far-reaching implications for both the tech industry and the future of intergovernmental fiscal relations in the United States.








