Candice Horbacz
US companies are on the precipice of a monumental investment surge into artificial intelligence (AI), with projections indicating trillions of dollars flowing into the sector over the coming years. Goldman Sachs analysts recently underscored this trend, estimating that capital expenditures from a select group of US AI "hyperscalers" alone could reach an astonishing $581 billion by 2026. A significant portion of this investment is earmarked for the buildout of advanced data centers—the physical backbone that houses the computational power and storage necessary to process the vast quantities of information underpinning AI development and deployment. However, this unprecedented growth is not without its complexities, as lawmakers in Washington are increasingly scrutinizing the broader societal and economic impacts of this technological boom, particularly concerning the infrastructure it requires.
The Proliferation of Data Centers and Emerging Policy Concerns
The rapid expansion of AI capabilities necessitates an equally rapid expansion of data center infrastructure. These facilities are energy-intensive, consuming substantial amounts of electricity and often requiring significant land footprints, leading to concerns among members of Congress and local communities. Beyond environmental considerations, the potential for AI to reshape labor markets—creating new jobs while potentially displacing others—has also drawn legislative attention. In response to these multifaceted concerns, several proposals have emerged from Congress aimed at regulating, and in some cases taxing, new data center developments. Two prominent ideas, one from Senator Mark Warner (D-VA) and another from Senate Finance Ranking Member Ron Wyden (D-OR), specifically target the tax treatment of AI data centers, proposing to deny them the crucial benefit of full expensing for machinery and equipment.
The premise of these proposals is that the current tax framework may not adequately address the externalities associated with the rapid growth of AI infrastructure or capture the immense economic gains projected from AI adoption. However, critics warn that such measures could introduce significant complexity into the tax code, create arbitrary distinctions between different types of investments, and potentially deter AI investment in the United States, thereby sacrificing jobs, economic growth, and vital tax revenues that could benefit local communities.
Deep Dive into the Legislative Proposals: Warner and Wyden
The legislative push to re-evaluate the tax treatment of AI data centers gained momentum in mid-2026. These proposals reflect a growing awareness in Congress of the profound shifts AI is poised to bring, alongside a desire to ensure responsible innovation and mitigate potential negative consequences.
Senator Warner’s "Data Center Tax Accountability and Disclosure Act"
In July 2026, Senator Mark Warner (D-VA) introduced the "Data Center Tax Accountability and Disclosure Act." This bill represents a targeted approach to incentivize environmentally conscious development within the AI sector. The core provision of Warner’s proposal is to disallow businesses from claiming full expensing under Section 168(k) of the tax code for "any property used in an AI data center." Full expensing is a critical tax incentive that allows businesses to immediately deduct the full cost of certain investments in new or improved technology, equipment, or buildings. This accelerates cost recovery, reduces the after-tax cost of investment, and typically encourages greater capital expenditure, leading to increased productivity, wages, and job creation.
Senator Warner’s definition of an "AI data center" is crucial to understanding the scope of his bill. While the specific criteria were still being refined, the proposal indicated that a data center would fall under this designation if a significant percentage (e.g., 20 percent) of its computing power or energy consumption was dedicated to AI-related tasks. This threshold, however, raised immediate questions regarding its practical implementation and measurement, a common challenge when attempting to legislate rapidly evolving technologies.
A key feature of Warner’s bill, however, is an exemption mechanism tied to environmental performance. Businesses could still claim full expensing for their AI data centers if these structures achieve Leadership in Energy and Environmental Design (LEED) certification at the Platinum or Gold levels. LEED certification, developed by the U.S. Green Building Council (USGBC), is a globally recognized symbol of sustainability achievement and leadership. It provides a framework for healthy, highly efficient, and cost-saving green buildings. By linking tax benefits to LEED certification, Warner’s proposal aims to directly encourage data center owners to invest in advanced energy efficiency measures, renewable energy integration, and sustainable construction practices, thereby addressing some of the environmental concerns associated with these facilities.
Senator Wyden’s White Paper on Data Center Taxation
Shortly after Warner’s bill, in August 2026, Senate Finance Ranking Member Ron Wyden (D-OR) released a comprehensive white paper outlining his own set of proposals for taxing data centers. Wyden’s approach is broader than Warner’s, encompassing not only the denial of full expensing but also introducing a new gross receipts tax.
Specifically, Wyden’s proposal sought to deny full expensing to new data centers, aligning with Warner’s objective of modifying investment incentives. However, it also proposed a new gross receipts tax on data center operations. Gross receipts taxes are levied on a company’s total revenue, regardless of its profitability. While Wyden suggested this tax would be in the "low single digits," such taxes can disproportionately impact businesses with high operating costs or low profit margins, potentially leading to significantly higher effective tax rates on net income. The white paper also included a "deemed minimum regime," a complex provision designed to ensure that data center users exceeding certain asset and expense thresholds pay a minimum tax, which Wyden argued was necessary to combat tax avoidance strategies.
Wyden’s proposal, like Warner’s, included certain exemptions. For instance, "internet infrastructure" and "small local data center operators" were mentioned as potential carve-outs from the gross receipts tax. However, the lack of precise definitions for these terms within the white paper immediately highlighted potential administrative hurdles and compliance challenges. Without clear guidelines from the Treasury Department and the IRS, businesses would face considerable uncertainty in determining their tax liabilities.
Unanswered Questions and Administrative Complexities
Both the Warner and Wyden proposals, while well-intentioned in their aims to address the societal impacts of AI and data center growth, suffer from significant definitional ambiguities that could complicate their administration and enforcement. For instance, Warner’s bill does not explicitly detail how the "20 percent AI threshold" for determining an "AI data center" would be calculated or verified. Would it be based on peak load, average load, specific hardware components, or software usage? These details are critical for consistent application.
Similarly, Wyden’s white paper leaves critical terms undefined, such as "internet infrastructure" and "small local data center operators" that would be exempt from his gross receipts tax. The "deemed minimum regime" further adds layers of complexity, potentially subjecting a wide array of entities engaged in "complex financial arrangements" with data centers to the new tax, raising concerns about its broad reach and potential unintended consequences. Such vagueness creates regulatory uncertainty, which can itself be a deterrent to investment.
Economic Analysis: Implications for Investment and Competitiveness
The proposed tax changes carry significant economic implications, potentially altering investment decisions, impacting US competitiveness in the global AI race, and reshaping the domestic economic landscape.
The Nature of Gross Receipts Taxes
Wyden’s proposed gross receipts tax, even at a "low single-digit" rate, could have a profound impact on data center profitability. As illustrated by economic analysis, a seemingly modest 4 percent gross receipts tax can translate into an effective tax rate on net income that is substantially higher—potentially reaching 40 percent or even 200 percent—depending on a firm’s operating expenses and profit margins. This is because gross receipts taxes are levied on revenue regardless of whether a company is profitable, effectively penalizing businesses with high costs or thin margins. Such a tax could disproportionately affect newer, less established data center operators or those investing heavily in research and development, potentially stifling innovation.
Risk of Driving Investment Overseas
A central concern raised by both proposals is the risk of driving AI and data center investment overseas. Neither the Warner nor Wyden proposals directly address non-US investment in data centers. By increasing the cost of capital or imposing new taxes on domestic data center operations, these measures could inadvertently make other nations with more favorable tax regimes more attractive for AI infrastructure development. This could lead to a significant erosion of the United States’ currently projected advantage in attracting global AI investment, potentially pulling broader AI research, development, and deployment activities with it. Such an outcome would deny US towns and cities the jobs, economic growth, and tax revenue that flow from investment in local communities, and could also have national security implications related to data sovereignty and technological leadership.
The Broader Tax Landscape and AI’s Economic Gains
A fundamental premise underlying these legislative efforts is the notion that the current US tax system is ill-equipped to capture the economic gains from AI investment and adoption. However, this premise has been subject to scrutiny by a wide array of tax experts across the ideological spectrum. Existing tax mechanisms—including corporate income taxes, capital gains taxes, and property taxes—are generally designed to capture returns from profitable enterprises, including AI companies and the data centers they operate. If AI generates "supernormal returns," these existing taxes are inherently positioned to capture a share of those profits.
Furthermore, if the economic benefits of AI are broadly distributed across the US economy, the federal government can anticipate higher tax revenues through various channels: increased corporate profits leading to higher corporate income tax collections, rising capital gains realizations from appreciating assets, and higher wages for individuals in increasingly productive sectors, particularly those in higher marginal tax brackets.
Depreciation and Investment Incentives
The denial of full expensing (or "bonus depreciation" as it’s often called) is a change in the timing of tax deductions, not a permanent increase in tax revenue in the long run. While it brings tax revenues forward, it also discourages new investment by increasing the upfront cost of capital. Economic literature extensively demonstrates that the benefits of new technologies are often widely dispersed across the economy, not solely concentrated with the innovators. This phenomenon is frequently cited as a justification for favorable tax treatment of research and development (R&D), a stark contrast to proposals that would increase taxes on critical AI infrastructure. By making it more expensive to invest in data centers, these proposals could slow down the very technological advancements that promise widespread economic benefits.
Estimated Fiscal Impact and Economic Harm
The Tax Foundation’s Taxes and Growth (TAG) Model provides critical insights into the potential fiscal and economic consequences of these proposals. Focusing on Senator Warner’s proposal to disallow bonus depreciation for qualified AI data centers, the central estimate projects a conventional revenue increase of $29.9 billion from 2027 to 2036. This revenue, however, comes at an economic cost. The proposal is estimated to reduce the long-run size of the economy by less than 0.05 percent, primarily due to an increase in the cost of capital for firms investing in certain data centers. On a dynamic basis, which accounts for changes in economic activity, the proposal is expected to raise $18.2 billion over 10 years, reflecting reduced income and payroll tax revenue associated with a smaller economy.
To account for inherent uncertainties, particularly regarding the projected path of data center investment, the proportion of data centers meeting the 20 percent AI criteria, and the rate of LEED certification adoption, the Tax Foundation modeled three scenarios: low, central, and high estimates for the investment basis affected. The low scenario projects $17.5 billion in conventional revenue over 10 years, while the high scenario could raise $46.5 billion. Even in the higher scenario, the corresponding economic harm, though increased, remains below 0.05 percent of GDP in the long run.
A crucial finding from these models is that none of the scenarios result in sustained increases in federal tax revenues over the long term. This reinforces the understanding that denying bonus expensing is primarily a timing shift—accelerating tax revenue collection—rather than a fundamental, long-term increase in the overall tax base.
| Table 1. Revenue Estimates of Bonus Depreciation Disallowance Under the Data Center Tax Accountability and Disclosure Act, 2027-2036, in Billions (Central Estimate) | Fiscal Year | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 | 2033 | 2034 | 2035 | 2036 | 2027-2036 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Conventional Revenue | 2.5 | 3.1 | 3.6 | 4.0 | 3.8 | 3.4 | 2.9 | 2.4 | 2.1 | 2.1 | 29.9 | |
| Dynamic Revenue | 1.5 | 2.0 | 2.4 | 2.7 | 2.6 | 2.3 | 2.0 | 1.6 | 1.4 | 1.4 | 18.2 |
Source: Tax Foundation General Equilibrium Model, August 2026.
| Table 2. Range of Revenue Estimates Under the Data Center Tax Accountability and Disclosure Act, 2027-2036 | Scenario | Conventional Revenue (2027-2036, Billions) | Dynamic Revenue (2027-2036, Billions) | Long-Run GDP Impact |
|---|---|---|---|---|
| Low | 17.5 | 10.5 | < -0.05% | |
| Central | 29.9 | 18.2 | < -0.05% | |
| High | 46.5 | 28.0 | < -0.05% |
Source: Tax Foundation General Equilibrium Model, August 2026.
Industry and Expert Reactions
The legislative proposals have drawn varied reactions from stakeholders. Proponents, primarily environmental advocacy groups and some community organizations, have largely welcomed the initiatives, viewing them as necessary steps to mitigate the environmental footprint of data centers and ensure that the economic benefits of AI are more equitably distributed. They emphasize the importance of attaching social and environmental responsibilities to rapidly growing industries.
However, the technology industry, particularly companies heavily invested in AI development and data center operations, has expressed significant apprehension. Industry associations representing cloud providers and data center operators have warned that such taxes could stifle innovation, increase operational costs, and ultimately harm US competitiveness. They argue that the focus should be on incentivizing green technology adoption through direct grants or existing tax credits, rather than penalizing necessary infrastructure. Many advocate for a "technology-neutral" tax code that avoids picking winners and losers, allowing businesses to recover investment costs efficiently.
Economists and tax policy experts are largely divided. Some acknowledge the need to address externalities but caution against complex, narrowly targeted taxes that could distort investment decisions. They often advocate for broader, more neutral tax reforms that allow markets to allocate capital efficiently while addressing societal concerns through other policy levers. Others, particularly those concerned about market concentration and wealth inequality, see these proposals as a valid attempt to ensure the public benefits from technological advancements and that corporations contribute their fair share.
Broader Policy Debate and Future Outlook
The debate surrounding the taxation of AI data centers encapsulates a broader policy challenge: how to foster technological innovation and economic growth while simultaneously addressing legitimate concerns about environmental sustainability, labor market disruption, and equitable distribution of wealth. Policymakers are urged to exercise caution when using the tax code to target specific industries or technologies in ways that could discourage essential investment.
A neutral tax code, one that allows businesses to recover their investment costs efficiently and taxes the resulting profits, is generally considered more effective at capturing economic gains from innovation without undermining competitiveness. The current proposals introduce layers of complexity and create distinctions that could distort capital allocation. As AI continues its rapid ascent, the legislative landscape will likely see further attempts to balance these competing priorities. The ultimate outcome of these tax proposals will significantly influence the trajectory of AI development and the positioning of the United States in the global technological race for decades to come. The dialogue underscores the need for comprehensive, forward-looking policy frameworks that can adapt to the accelerating pace of technological change while safeguarding broader societal interests.
