The Data Center Bargain: Can Taxing AI Infrastructure Fund the Workers It Displaces?
Senator Mark Warner wants to extract a pound of flesh from data centers. The phrase is deliberately Shakespearean – and the trade-off it implies deserves unpacking.
At the Axios AI Summit in Washington, D.C. this week, the Virginia Democrat floated an idea that sits uncomfortably between two camps in the AI infrastructure debate: tax the data centers powering the AI boom, and use that revenue to help workers navigate the economic transition these same technologies are accelerating.
The proposal arrives at a moment when the politics of AI infrastructure have become genuinely scrambled. On one side, Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced legislation calling for a national moratorium on new data center construction until strong national safeguards are in place. On the other, the Trump administration is urging Congress to preempt state AI laws it views as too burdensome. Warner's position occupies neither pole – and that's precisely what makes it worth examining.
The Disagreement Beneath the Disagreement
The surface-level debate is about data centers: their noise, their water consumption, their strain on electrical grids. According to Data Center Watch, $64 billion in U.S. data center projects have been blocked or delayed amid local opposition since May 2024. Communities from Virginia to Wisconsin to Oklahoma have organized against facilities that promise tax revenue but deliver, in the eyes of many residents, rising utility bills and transformed landscapes.
But Warner's framing suggests something deeper is at work. The fear of AI-related job loss, he told TechCrunch, is palpable. A venture capitalist recently told him he's writing software investments down to zero due to the capabilities of Anthropic's Claude. A major law firm said it's not hiring first-year associates because AI can now handle much of the work once assigned to junior lawyers.
This is where the debate gets interesting – and where the positions need disentangling.
Position A holds that data centers are the problem. Stop building them, and the harms stop accumulating. This is the Sanders-AOC view, and it has the virtue of simplicity. As Sanders put it: We cannot sit back and allow a handful of billionaire Big Tech oligarchs to make decisions that will reshape our economy, our democracy and the future of humanity.
Position B holds that data centers are infrastructure for a race the United States cannot afford to lose. Warner articulated this view bluntly: A data center moratorium simply means China is gonna move quicker, and this is one where we can't lose. Senator John Fetterman agreed, writing on X that he refuse[s] to help hand the lead in AI to China.
Position C – Warner's actual proposal – accepts that data centers will be built, but argues the communities bearing their costs deserve compensation that addresses the underlying anxiety: not just noise and water, but the economic displacement these facilities enable.
The question worth asking: are these three positions actually in conflict, or are they arguing about different things?
What the Evidence Shows About Job Displacement
The job displacement fears are not imaginary, though their scale remains contested. Research from GovAI and the Brookings Institution estimates that approximately 6.1 million U.S. clerical workers face high AI exposure combined with low adaptive capacity – meaning fewer transferable skills and financial resources to navigate a career transition. Women make up about 86 percent of those most vulnerable workers.
Analysis from the Federal Reserve Bank of Dallas offers a more nuanced picture: AI appears to be simultaneously aiding and replacing workers. Employment in the computer systems design sector has declined 5 percent since ChatGPT's release, while wages in the same sector have risen 16.7 percent. The explanation? AI may be automating codified knowledge – the book learning of entry-level workers – while complementing the tacit knowledge of experienced professionals.
The pattern is consistent with what Warner heard from that law firm: not mass layoffs, but a closing of the entry-level pipeline. Young workers with primarily codifiable knowledge face challenging job markets. Experienced workers, particularly those in occupations with high experience premiums, may actually benefit.
This distinction matters for policy design. A program that retrains displaced secretaries for nursing careers addresses a different problem than one that helps new law school graduates find alternative paths when firms stop hiring first-year associates.
The Mechanism Question
Warner's proposal raises a practical question he acknowledges hasn't been fully answered: who should pay?
Should it be the chip makers, Jensen [Huang, Nvidia's CEO]? Should it be the large language model companies? Should it be the Goldman Sachs of the world who are using these tools to cut back on a number of first-year associates?
His answer – that the easiest place to extract the pound of flesh is probably going to be from the data centers – is pragmatic rather than principled. Data centers are physical, locatable, and already subject to local taxation. They're easier to tax than software or algorithms.
But this creates a potential mismatch. The communities hosting data centers are not necessarily the communities experiencing AI-driven job displacement. A data center in rural Indiana generates tax revenue for St. Joseph County, but the law firm not hiring first-year associates might be in New York or Chicago. The secretary whose job is automated might live in a suburb with no data center in sight.
Warner points to Henrico County, Virginia, which used data center tax revenue to fund affordable housing. That's a tangible community benefit, but it's not directly connected to AI job displacement. The question is whether the connection needs to be direct, or whether any mechanism that extracts value from AI infrastructure and redirects it toward worker support counts as progress.
The Precedent Problem
There's a deeper tension in Warner's framing that deserves acknowledgment. The pound of flesh metaphor comes from The Merchant of Venice, where Shylock's demand for literal flesh becomes a symbol of justice perverted into cruelty. Warner presumably means something more benign – a fair extraction of value from those who benefit most from AI.
But the metaphor reveals an assumption: that data centers owe something to the communities they affect, beyond what they already pay in taxes and jobs. Amazon's recent $15 billion investment in Northern Indiana includes an energy agreement with NIPSCO that the company claims will provide approximately $1 billion in cost savings to existing Indiana residents and businesses over 15 years. The company is also funding workforce development programs, including data center technician training and fiber optic fusion splicing workshops.
Is this enough? The answer depends on what obligation one believes tech companies have to communities affected by their products – not just their facilities.
The strongest version of the industry's argument would be: data centers create jobs, pay taxes, and fund infrastructure upgrades. The AI they enable creates productivity gains that benefit the entire economy. Demanding additional pounds of flesh risks driving investment elsewhere, including overseas.
The strongest version of the critics' argument would be: the productivity gains from AI are not evenly distributed. The workers displaced by AI-enabled automation don't automatically benefit from aggregate economic growth. Without deliberate intervention, the gains accrue to shareholders and highly skilled workers while the costs fall on those least able to adapt.
Warner's proposal attempts to bridge these positions by accepting the premise that data centers will be built while insisting that their benefits be more deliberately shared.
What Would Have to Be True
For Warner's approach to work, several things would need to be true:
First, data center taxation would need to generate sufficient revenue to fund meaningful worker transition programs. The scale of potential displacement – millions of workers over the coming decade, according to various estimates – suggests the funding requirements would be substantial.
Second, the programs funded would need to actually help displaced workers find new employment. According to Fortune, nearly 75% of people don't apply for unemployment benefits. Any transition program would need to reach workers who might not seek help through traditional channels.
Third, the political coalition supporting this approach would need to hold. Warner explicitly declined to support the Sanders-AOC moratorium, positioning himself as a moderate alternative. But if public anger toward AI and data centers continues to grow – an NBC News poll found 46% of registered voters view AI negatively compared to only 26% viewing it positively – the pressure for more aggressive action may intensify.
The Question That Remains
Warner's warning is worth taking seriously: The pitchforks are coming out.
The backlash against data centers is real, bipartisan, and accelerating. According to the American Enterprise Institute, support for an outright ban on data center construction grew from 37% to 41% in successive polls just a month apart in late 2025. At least 142 activist groups across 24 states are organizing to block data center construction.
The question is whether taxing data centers to fund worker transition represents a sustainable middle ground, or merely a temporary holding pattern before the politics shift further.
What's clear is that the debate has moved beyond technical questions about grid capacity and water usage. It's now a debate about who benefits from AI, who bears its costs, and what obligations the builders of AI infrastructure have to the workers their products may displace.
That's a values disagreement, not a facts disagreement. And values disagreements don't get resolved by finding the right data – they get resolved by making choices about what kind of economy and society people want to build.
The conversation about AI infrastructure in Europe faces similar tensions, though the regulatory context differs. For those working through these questions – policymakers, technologists, investors, researchers – the American debate offers both cautionary tales and potential models. The challenge is distinguishing which is which.
For those seeking to engage with these questions directly, rather than through the distorting lens of polarized debate, Human x AI Europe on May 19 in Vienna offers a rare opportunity: the right people, in the right room, working through complexity rather than performing positions.
Frequently Asked Questions
Q: What is Senator Mark Warner's proposal for addressing AI job displacement?
A: Warner proposes taxing data centers and using that revenue to fund worker transition programs, including training for new careers like nursing and AI upskilling programs. He has not yet introduced formal legislation but described data centers as the easiest place to extract the pound of flesh from the AI industry.
Q: What is the Sanders-AOC data center moratorium bill?
A: The Artificial Intelligence Data Center Moratorium Act, introduced on March 25, 2026, would pause all new data center construction with peak power loads exceeding 20 megawatts until Congress enacts comprehensive AI regulation including worker protections, environmental safeguards, and civil rights provisions.
Q: How many U.S. jobs are at risk from AI displacement?
A: Estimates vary significantly. Research from GovAI and Brookings identifies approximately 6.1 million clerical workers at high risk with low adaptive capacity. The Tufts University American AI Jobs Risk Index projects 9.3 million jobs at risk of displacement in the next 2-5 years, with associated household income at risk spanning $200 billion to $1.5 trillion annually.
Q: How much data center investment has been blocked or delayed by local opposition?
A: According to Data Center Watch, $64 billion in U.S. data center projects have been blocked or delayed between May 2024 and March 2025, with at least 142 activist groups across 24 states organizing against new construction.
Q: Which workers are most vulnerable to AI job displacement?
A: Research indicates administrative and clerical workers face the highest risk, particularly those with limited transferable skills and financial resources. Women make up approximately 86% of the most vulnerable workers. Entry-level positions in fields like law, software development, and customer service are also heavily affected.
Q: What is the current public opinion on AI and data centers in the United States?
A: An NBC News poll found 46% of registered voters view AI negatively compared to 26% viewing it positively. Regarding data centers specifically, a Heatmap poll found only 44% would support a data center being built near them, while 42% would oppose it – a net approval of just +2%.