Akshat Singh — August 4, 2026
Resolved: The United States federal government should enact a moratorium on hyperscale data center construction.
Background
The data center buildout happening right now is unprecedented in scale. BloombergNEF finds that capex (capital expenditure, or money spent on infrastructure like buildings and equipment) from the 14 largest publicly traded data center operators is nearing $750 billion in 2026, up from under $450 billion the year before. Accuris confirms the top five hyperscalers (companies that operate massive, internet-scale data center networks) alone—Amazon, Microsoft, Google, Meta, and Oracle—will spend over $600 billion on infrastructure this year, a 36% jump from 2025, with roughly 75% of that money going straight into AI.
All of that money buys machines, and machines need power. Data centers are electricity-hungry, and not like a normal building. A single facility can pull as much electricity as a small city. Gartner finds that global data center electricity use will grow 26% in 2026 alone, and is projected to roughly double again by 2030. That growing demand doesn’t stay contained to the tech industry; it competes with regular households for space on the power grid. In the mid-Atlantic and Midwest, that competition is already showing up on people’s electric bills. IEEFA reports that PJM, the operator that manages the regional power grid across 13 states, saw prices spike so sharply that they pinned 63% of the increase on data centers alone. This entails a $9.3 billion cost that gets passed down to everyday electricity customers, and some households are already paying $16 to $21 more per month because of it.
Those kinds of costs are exactly why a moratorium, which is a temporary, legally imposed halt on new construction, has gone from a vague idea to a real piece of federal legislation. The Cato Institute confirms that Senator Bernie Sanders introduced the federal AI Data Center Moratorium Act in March 2026, and by the end of June, 116 cities and towns had already passed versions of their own. The Rockefeller Institute finds that most of these local moratoriums come down to the same handful of community complaints, including strained power grids, rising bills, excessive water use, noise, and pollution. Supporters of a moratorium see all of this as reason enough to slow down; and many opponents see it as America handing China a head start in the AI race. That’s the fight that this topic aims to settle.
AFF
Ratepayer Burden
Data center demand is driving up electricity costs for ordinary Americans, not the tech companies building the facilities. IEEFA finds that PJM’s independent market monitor (PJM manages the power grid across 13 mid-Atlantic and Midwestern states) attributed 63% of a massive regional price spike directly to data centers, adding $9.3 billion in costs in a single year, passed straight down to ratepayers (electricity customers).
That number keeps climbing. Introl finds PJM’s capacity prices (the rate charged to keep enough power plants on standby for peak demand) jumped from $28.92 per megawatt-day to $329.17 in two years, an increase of over 1,000%. Cumulative costs could reach $100 billion to $163 billion by 2033. Washington D.C. customers are already paying $21 more a month. Ohio customers are paying $16 more. Maryland customers are paying $18 more. The average PJM household could face a $70 monthly increase by 2028.
Rising utility costs carry real consequences. Families that struggle to afford electricity cut back on other necessities like food, medicine, or heat. The burden of the AI boom lands hardest on the households least able to absorb it.
Community and Environmental Strain
Beyond the electric bill, hyperscale data centers are reshaping the physical communities around them, and residents in a lot of these towns never agreed to that trade. Good Jobs First finds that at least 63 local moratorium actions have already passed across dozens of towns and counties, and a March 2026 Pew Research poll found seven in ten Americans nationwide don’t want a data center built anywhere near their community.
Water use is one of the biggest flashpoints. ElectricChoice finds that in The Dalles, a town in Oregon, Google’s water use grew 316% while the town’s population grew only 12%, a direct sign that the strain data centers put on local resources far outpaces any benefit brought to the town itself. The same report finds that roughly 80% of a data center’s total water footprint doesn’t even come from onsite cooling, it comes from the power plants generating its electricity upstream, meaning the true water cost is often hidden from the communities footing the bill.
Denver offers a clean case study of what this looks like on the ground. According to Data Center Watch, residents in the Globeville-Elyria-Swansea neighborhood organized against a CoreSite data center expansion, arguing the facility added to a neighborhood already burdened by highways, industrial zoning, a nearby refinery, and a Superfund site. The same pattern is playing out nationwide: in Box Elder County, Utah, protesters gathered outside a 40,000-acre AI data center project over concerns about water use, energy demand, and environmental harm.
These aren’t isolated incidents. Data Center Watch finds that in 2026 alone, more than $130 billion worth of proposed data center projects have already been defeated or canceled due to this kind of organized local opposition, which shows the scale of the backlash is not a fringe phenomenon, it’s a nationwide pattern repeating state after state.
If you want to scale this contention up for a bigger-picture judge, the environmental justice angle is strong: many of the neighborhoods bearing the heaviest burden, like Globeville-Elyria-Swansea, are already low-income and already overburdened by industrial pollution, meaning the costs of the data center boom are landing disproportionately on the communities least equipped to fight back or relocate.
The AI Debt Bubble
The AI data center buildout isn’t just expensive, it’s being financed in a way that makes the whole industry more fragile than it looks from the outside.
Introl finds hyperscalers raised $108 billion in debt during 2025 alone to fund this construction boom, with projections suggesting $1.5 trillion in debt issuance over the coming years. That same report finds capital intensity has reached 45% to 55% of revenue for major hyperscalers (capital intensity means the share of a company’s revenue that gets poured back into infrastructure spending), a level historically unthinkable for technology companies and closer to what utility or industrial companies carry.
That level of debt-fueled spending only makes sense if AI demand keeps growing exactly as projected, and a lot of analysts aren’t convinced it will. BloombergNEF finds three of the four major hyperscalers lost market value following recent earnings calls specifically because investors are worried about the size of these capital commitments, even as the companies themselves keep raising their spending targets.
If AI adoption or revenue growth ends up slower than expected, hyperscalers are left holding massive debt tied to physical infrastructure that can’t easily be repurposed. A data center built for AI training isn’t easily converted into something else if the demand doesn’t materialize. That combination, high debt plus an asset that loses value fast if the underlying business case weakens, is exactly the kind of structural risk that has caused broader financial slowdowns in the past, like the overbuilt fiber-optic networks of the dot-com era that left companies drowning in debt for infrastructure nobody was using yet.
A moratorium (a temporary halt on new construction) doesn’t eliminate this risk entirely, but it does slow the pace at which this debt keeps piling up, giving markets and regulators more time to assess whether the spending is actually sustainable before it gets even bigger.
NEG
The AI Race with China
This is the strongest, most sourced NEG contention, and it directly answers everything AFF says about grid strain and rising bills.
Kevin O’Leary, the businessman and investor, warns that China is getting more power sooner and using it to train its own AI systems while American states slow down construction here at home. The concern isn’t hypothetical. Pro-AI experts warn that if the movement for a moratorium keeps growing and gets backed by a major political party, temporary pauses could become permanent, handing China a clear advantage.
The national security concern isn’t limited to commentators either. A local news outlet in Seattle reports that when the city council voted to ban new data centers, a federal Department of Energy official warned the moratorium raises national security and data sovereignty concerns, arguing it weakens America’s position in the broader technology race with China.
There’s also a deeper problem worth flagging: some of the loudest opposition to data centers may not be organic. The Bitcoin Policy Institute finds that Chinese state media outlets and a CCP-aligned nonprofit network have been running coordinated campaigns against American AI infrastructure, and traces over $2 billion routed through foreign-tied charitable vehicles into the same advocacy groups pushing the moratorium movement. If foreign adversaries are actively encouraging the U.S. to slow itself down, that’s a serious reason to be skeptical of the resolution.
The impact here scales as high as any judge wants it to. Whoever wins the AI race gains the upper hand economically and militarily going forward. Falling behind isn’t just a policy setback, it’s a long-term geopolitical loss that’s difficult to reverse.
Jobs and Tax Revenue
Even though data centers don’t employ huge numbers of people permanently, the economic footprint they leave behind is real and substantial, and a moratorium takes that away from communities that could use it.
The Associated General Contractors of America finds that a single mid-sized data center project can support nearly 9,700 construction jobs during the build phase, generate $2.4 billion in total economic output, and contribute $1 billion to state GDP. The same report finds these projects generate roughly $84 million in annual peak state and local tax revenue, money that goes directly toward schools, roads, and other public services in the surrounding community.
That tax benefit isn’t just theoretical either. According to a jobs-focused analysis of the industry, data centers in Loudoun County, Virginia alone contributed over $600 million annually in local tax revenue, a figure that dwarfs what most other kinds of commercial development could provide to a single county.
Communities that pass moratoriums aren’t just pausing construction, they’re turning away this kind of investment altogether, often for projects that have already secured financing and are ready to break ground elsewhere the moment they hit local resistance. Once a developer redirects a project to a more welcoming state or county, that community doesn’t get a second chance at the same investment. The jobs, the tax base, and the economic activity simply move somewhere else, while the underlying national demand for AI infrastructure doesn’t go away, it just relocates.
Federalism and Mechanism Problems
Even setting aside whether a moratorium is a good idea, there’s a real question about whether the federal government even has a clean way to enact one.
Land use, zoning, and retail electricity regulation are overwhelmingly state and local powers, not federal ones. A debate-focused analysis of the topic finds that a federal moratorium has no clean legal lever to work with, and instead has to rest on contestable constitutional hooks like the Commerce Clause or federal permitting authority, which is exactly why nearly all real-world moratorium action so far has happened at the state and local level instead of in Congress.
This matters because it means the affirmative world isn’t as simple as flipping a switch. The Cato Institute finds that even local moratoriums are legally limited: they only apply to new or unapproved projects, meaning facilities that already secured zoning permits, submitted complete applications, or began construction can continue regardless. A federal version would likely face the same limitation, meaning the moratorium wouldn’t actually stop the huge wave of data centers already in the pipeline, it would only apply to future projects that haven’t broken ground yet.
There’s also the question of what happens when the pause ends. A moratorium is temporary by definition, and the debate-focused analysis above notes that a real moratorium bill needs to define how long it lasts and what criteria have to be met before it’s lifted. Without a clear answer, a moratorium doesn’t actually solve the underlying problems AFF raises, like grid strain or rising bills, it just delays them, while creating additional uncertainty for an industry that other countries would be happy to take business away from in the meantime.
Read More Here:
