Two congressional proposals targeting artificial intelligence data centers would eliminate certain tax deductions and could raise $29.9 billion over 10 years while shrinking the economy, according to an August 2026 analysis from the Tax Foundation. The proposals from Sen. Mark Warner (D-VA) and Senate Finance Ranking Member Ron Wyden (D-OR) would deny new AI data centers the ability to immediately deduct the full cost of machinery and equipment, a tax benefit known as full expensing. The report warns that blocking cost recovery for these investments could push AI development overseas, taking jobs and tax revenue with it.
Warner's Data Center Tax Accountability and Disclosure Act, introduced in July 2026, would prevent businesses from claiming full expensing for any property used in an "AI data center"—defined as facilities where at least 20 percent of activity involves developing or operating artificial intelligence. Companies could still claim the deduction if their data centers earn Leadership in Energy and Environmental Design (LEED) certification at Platinum or Gold levels. Wyden's proposal goes further, denying full expensing to all new data centers while also blocking Opportunity Zone funding and real estate investment trust benefits. His plan includes a gross receipts tax at an undefined rate on data center operators, with exemptions for "internet infrastructure," corporate IT departments, "small" data centers, and pre-2024 facilities for all but "the largest actors." The Tax Foundation's central estimate projects Warner's bill would raise $29.9 billion from 2027 to 2036 on a conventional basis, though accounting for economic contraction reduces that figure to $18.2 billion. The range spans from $17.5 billion in a low-investment scenario to $46.5 billion if more data centers fall under the 20 percent AI threshold. A seemingly modest gross receipts tax can translate into punishing rates on actual profits—the report illustrates that a 4 percent levy on gross receipts could mean an effective tax rate reaching 40 percent or even 200 percent on net income, depending on a company's expense structure.
The report argues that neither proposal addresses non-U.S. investment in data centers, which may inadvertently drive data center construction abroad and pull broader AI investment along with it. According to the Tax Foundation, both plans rest on the assumption that the current tax system can't capture economic gains from AI investment—a premise that "has been scrutinized by tax experts across the ideological spectrum." The analysis notes that existing corporate income, capital gains, and property taxes are likely to collect any exceptional returns from AI companies and their data centers, weakening the justification for AI-specific taxes. The report also points out that switching depreciation deduction timing doesn't generate long-run revenue; it merely shifts the same tax dollars forward in time rather than increasing total collections. Warner's proposal would reduce the long-run size of the economy by less than 0.05 percent by raising the cost of capital for firms investing in certain data centers, the Tax Foundation estimates.
The report recommends that policymakers avoid using the tax code to capture AI returns in ways that discourage investment, warning that denying cost recovery and imposing new excise taxes would add complexity, distort investment decisions, and risk pushing activity abroad. The Tax Foundation cites extensive economics research finding that the benefits of new technologies spread throughout the economy rather than being captured primarily by innovators—a rationale often used to justify favorable tax treatment of research and development, not higher taxes. A neutral tax code that lets businesses recover investment costs while taxing resulting profits is better suited to capturing economic gains from AI without undermining U.S. competitiveness, the report concludes. None of the scenarios modeled result in sustained increases in federal tax revenues over the long term, since denying bonus expensing is primarily a timing shift rather than a permanent revenue boost.