Sen. Elizabeth Warren (D-Mass.) called Wednesday for a new per-kilowatt-hour excise tax on the energy consumed by artificial intelligence data centers — and a companion wealth tax targeting AI-sector billionaires — with all revenue earmarked for workers displaced by automation. The proposal, published in a Time op-ed and reported by Axios, arrives as American technology companies have eliminated more than 113,000 jobs in 2026 and tech-sector unemployment has climbed to 5.8%, its highest level since the dot-com bust of 2001–2002.

"We can't be afraid to consider even bigger and bolder proposals to tax AI too, including ideas that sound radical today but may quickly become common sense," Warren wrote. Axios described the proposal as the latest progressive response to a technology that polls show is increasingly unpopular with voters.

The plan has two pillars. The first is a per-kilowatt-hour excise tax applied directly to the energy consumed by AI data centers — the facilities that power model training and inference for products from OpenAI, Google, Amazon, and Microsoft. The second is a wealth tax targeting AI-sector billionaires; Warren singled out Amazon founder Jeff Bezos and OpenAI CEO Sam Altman by name as examples of ultra-wealthy individuals she argues are capturing nearly all of AI's economic gains while the workers most exposed to automation lack a safety net. Revenue from both mechanisms would fund universal healthcare, free higher education, and a strengthened unemployment insurance system.

AI Data Center Energy Consumption Fuels Excise Tax Logic

The energy-based tax is designed to scale directly with AI's compute intensity. Global data center electricity demand soared 17% in 2025 — well above the 3% growth in overall global electricity consumption — and is set to double by 2030, according to the International Energy Agency. Capital expenditure from five major technology companies exceeded $400 billion in 2025 and is projected to increase by a further 75% in 2026.

A typical AI-focused hyperscaler annually consumes as much electricity as 100,000 households, according to IEA estimates cited by Pew Research Center. The largest campuses currently under construction are expected to use 20 times as much. In Virginia alone, data centers consumed approximately 26% of the state's total electricity supply in 2023, according to the Electric Power Research Institute, putting upward pressure on utility bills for residential customers across the mid-Atlantic.

Warren's per-kilowatt-hour structure mirrors the logic of existing energy excise taxes on other resource-intensive industries: the more compute power consumed, the larger the tax obligation. Critics have argued that such a levy could push data center investment overseas or slow AI development in the United States. Supporters counter that it would internalize energy and social externalities that currently go unpriced by the market.

Warren Wealth Tax: What It Targets and Who Would Pay

The wealth tax component draws from legislation Warren introduced in March 2026, the Ultra-Millionaire Tax Act of 2026. That bill would impose a 2% annual tax on net worth above $50 million and an additional 1% on billionaires — effectively a 3% annual levy on the wealthiest tier. University of California, Berkeley, economists Emmanuel Saez and Gabriel Zucman, known for their work on economic inequality, estimated the bill would raise $6.2 trillion over a decade. The legislation is co-sponsored in the House by Rep. Pramila Jayapal (D-Wash.) and Rep. Brendan Boyle (D-Pa.).

Warren's Wednesday op-ed frames the wealth tax as an AI-specific measure rather than a general inequality policy: AI productivity gains, she argues, flow almost exclusively to a narrow class of investors and executives while the workforce disruption those gains cause goes uncompensated. The senator cited warnings from tech executives themselves — including forecasts of extreme wealth concentration — to argue that the industry acknowledges the risk but has not proposed a remedy.

How Does AI Job Displacement Actually Work?

AI job displacement refers to the loss of positions when artificial intelligence tools automate tasks previously performed by human workers. It is most acute today in entry-level roles in customer service, software development, legal research, data entry, and financial analysis, where AI can perform a large share of routine tasks without human oversight.

Employment for software developers ages 22 to 25 has fallen nearly 20% from its 2024 peak, according to the Stanford Human-Centered AI Institute's 2026 AI Index, even as headcount for older developers at the same companies continued to grow. The World Economic Forum's Future of Jobs Report 2025, covering surveys of more than 1,000 employers representing 14 million workers, projected 92 million job displacements by 2030, offset partially by 170 million newly created roles — a net gain that nonetheless requires massive retraining investment and leaves a significant timing gap for displaced workers.

Fiscal Critics: Revenue Overestimates, Administration Hurdles

The proposal faces substantive opposition from fiscal policy analysts. The National Taxpayers Union Foundation, which released an analysis of Warren's 2026 tax package in April, found that her wealth tax proposals "drastically overestimate the potential revenue to be raised, handwave difficult questions of administrability, and portend significant economic consequences." The Tax Foundation has raised similar objections in prior cycles, noting that wealth taxes face acute enforcement challenges because of the difficulty of valuing illiquid assets — private company stakes, real estate, and intellectual property — on an annual basis.

Warren's op-ed does not specify the per-kilowatt-hour rate she has in mind for the data center excise tax, describing it only as "reasonable." Without a specific rate, independent revenue estimates are not yet possible. The proposal is also not yet introduced as formal legislation; it is a policy argument published in a general-audience magazine, and its path to a Senate floor vote is, at minimum, uncertain in a body where partisan division has blocked similar measures.

The Republican-controlled Congress has moved in the opposite direction: the "Big, Beautiful Bill" signed in 2025 delivered $67 billion in retroactive research tax breaks to corporations in 2026 alone, according to a Joint Committee on Taxation analysis Warren commissioned, rather than new levies on tech companies.

Political Context: Bipartisan Anxiety, Divided Prescriptions

Warren is not alone among senators in pointing to AI infrastructure as a source of public revenue. Sen. Mark Warner (D-Va.) called at the Axios AI Summit in March 2026 for extracting a "pound of flesh" from data centers to fund worker transition programs, describing the facilities as the most tractable point at which to impose AI-related costs. Warner has not introduced formal legislation either, but his remarks signal that data-center taxation has moved from a fringe position to a topic of active Senate discussion.

Bipartisan agreement exists on some elements of AI governance — including export controls on advanced chips and national security guardrails — but proposals to tax AI companies directly remain deeply contested. Warren has previously backed digital services taxes and financial transaction taxes, and the new proposal fits a pattern of using the tax code to redirect gains from capital-intensive technology sectors. Whether it advances in the current Senate depends heavily on whether the distributional argument — that AI gains are flowing to a narrow group while costs are distributed broadly — gains enough political traction to force a floor debate.

Frequently Asked Questions

What is Elizabeth Warren's AI data center tax proposal?

Warren is proposing a per-kilowatt-hour excise tax on the energy consumed by AI data centers, paired with a wealth tax on billionaires including Amazon's Jeff Bezos and OpenAI CEO Sam Altman. Revenue would fund universal healthcare, free higher education, and a stronger unemployment insurance system for workers displaced by automation. She outlined the proposal in a Time op-ed published May 27, 2026.

How much energy do AI data centers use?

Global data center electricity demand rose 17% in 2025, far outpacing overall electricity growth of 3%, according to the International Energy Agency. A typical AI-focused hyperscaler consumes as much electricity annually as 100,000 households, and the IEA projects overall data center electricity consumption will double by 2030 as AI workloads expand.

Will AI cause mass job losses?

The evidence so far shows concentrated disruption rather than mass elimination. Tech-sector unemployment has risen to 5.8% in early 2026 — its highest since the dot-com bust — and employment for software developers ages 22 to 25 has fallen about 20% from its 2024 peak. The World Economic Forum projects 92 million job displacements globally by 2030, partially offset by 170 million new roles.

Does Warren's wealth tax proposal have a chance of passing?

The proposal faces long odds in the current Congress. The companion Ultra-Millionaire Tax Act of 2026 — introduced in March 2026 — is considered unlikely to pass given partisan divisions, according to CBS News coverage. Fiscal critics including the National Taxpayers Union say Warren's wealth-tax proposals overestimate revenue and would be administratively unworkable.