President Trump's record-breaking 107-minute State of the Union address mentioned artificial intelligence only three times. He:

  • Announced a "ratepayer protection pledge" meant to make major tech companies cover the power needs of their new AI data centers.
  • Highlighted First Lady Melania Trump's support for the TAKE IT DOWN Act, a law targeting AI-generated deepfakes and non-consensual explicit images.
  • Briefly described the Presidential AI Challenge, a nationwide competition for K-12 students and educators to use AI tools to solve a community challenge.

This post unpacks what the President said about AI and analyzes the context, status, and implications of these policies.

1. Data Centers

Trump announced that his administration had negotiated a "Ratepayer Protection Pledge" with major tech companies, requiring them to build their own power plants for new data centers:

"Tonight, I'm pleased to announce that I have negotiated the new 'ratepayer protection pledge'. We're telling the major tech companies that they have the obligation to provide for their own power needs. They can build their own power plants as part of their factory, so that no one's prices will go up. This is a unique strategy never used in this country before. We have an old grid. It could never handle the kind of numbers, the amount of electricity that's needed. So I'm telling them they can build their own plant. They're going to produce their own electricity. It will ensure the company's ability to get electricity, while at the same time, lowering prices of electricity for you."

President Trump, State of the Union

In short, the White House is framing this pledge as a way to keep AI's massive energy demands from (1) driving up local utility bills and (2) overwhelming what Trump calls an "old grid".

The scale of U.S. data center growth helps explain why this came up in the speech:

  • Today, the U.S. hosts an estimated 4,000 to 5,000 data centers – roughly half of global capacity
  • Nearly 3,000 more are either under construction or in the planning pipeline
  • The biggest spenders include AWS, CoreWeave, Meta, Microsoft, Oracle, and xAI, with large U.S. tech firms projecting around $600 billion in data-center investment in 2026 alone

These facilities aren't evenly distributed. They cluster in a few major hubs:

  • Virginia: 663 operational centers, with nearly 600 more under construction
  • Texas: 405 existing centers, 440 planned or under construction
  • California: 300+ operating centers, 19 new projects underway
  • Georgia: 162 now, 285 more planned
  • Pennsylvania: 98 now, 184 planned
Map showing current and planned data centers across U.S. states as of Oct. 2025

States like Georgia and Pennsylvania are on the cusp of especially rapid data-center expansion. That means big new energy loads on regional grids – and explains the political value of telling voters that tech giants, not households, will foot the bill for that power.

AI data centers are the backbone of modern AI. They power the applications that will drive economic growth, boost productivity across sectors, and fuel new waves of innovation. National security officials also see strong domestic data center capacity as critical to U.S. competitiveness in AI.

In the short term, data centers bring money and jobs. They generate significant tax revenue for state and local governments and create thousands of jobs in construction, engineering, and facility operations. A recent report from the pro-tech American Edge Project estimates nearly $27 billion in data center tax revenue nationwide over the next decade. Virginia ($4.2 billion), Arizona ($2.6 billion), and Delaware ($2 billion) are projected to see especially large windfalls.

Critics focus on three main issues: heavy energy and water use, local quality-of-life impacts, and relatively few permanent jobs once construction ends.

Energy demands from U.S. data centers are expected to triple between 2023 and 2028, according to Food & Water Watch. By 2028, AI-driven data centers alone could:

  • Use 720 billion gallons of water a year just for cooling – enough to meet the indoor needs of 18.5 million U.S. households
  • Consume around 300 TWh of electricity annually – roughly what it takes to power more than 28 million U.S. homes

Local residents also worry about noise, truck traffic, new transmission lines, and industrial-scale facilities dropped into or near residential areas.

Trump's "ratepayer protection pledge" lands in the middle of a growing backlash against data center build-out.

Since 2023, about $162 billion in data center projects have been blocked or delayed, according to Data Center Watch, as local organizing in opposition to data centers ramps up. In December, more than 230 national, state, and local groups signed a letter to Congress calling for a nationwide moratorium on new data centers, citing unsustainable energy and water use and rising utility costs for families and small businesses.

Voters aren't uniformly anti-data center, but support is highly local. Polling suggests:

  • People are more open to data centers in the abstract than in their own backyard
  • Only 17% expect data centers to matter in this year's elections (though that jumps to 30% for people living within a mile of a data center)
  • 57% think data centers will eventually become a campaign issue where they live

Perhaps the biggest takeaway from recent polling is there is a large swath of indifferent or undecided voters on this issue: A recent POLITICO poll found 37% of voters would support new data center development, 28% would oppose, and 36% are on the fence. Interestingly, many see "new jobs" as the main benefit – even though, long term, data centers are not big job creators.

Bar chart: voter views on local data center development, split by 2024 vote

Still, the issue is already showing electoral teeth. In New Jersey and Virginia, Mikie Sherrill and Abigail Spanberger won gubernatorial races in part by promising tougher data center regulation. In Georgia, Democrats flipped two statewide utility regulatory seats after running on putting guardrails around data center growth.

The political factions on this issue are still fluid. Progressives and moderate Democrats are split on whether to pause new construction, and some Republicans are also calling for stricter rules on AI infrastructure:

  • Sen. Bernie Sanders (I-VT) has called for a federal moratorium on AI data center construction, warning about runaway expansion and huge energy loads.
  • Florida Gov. Ron DeSantis has become a vocal critic, unveiling a "bill of rights" last year aimed at protecting residents from data center impacts.
  • In states seeing rapid build-out, leaders are trying to split the difference. Pennsylvania Gov. Josh Shapiro, a likely 2028 contender, has proposed new safeguards for ratepayers and resources, while still backing new construction.

"This is a political bomb waiting to go off. You've seen data centers go from a third-tier issue to a top-tier issue in politics in the span of a year," said Jared Leopold, a Democratic strategist and co-founder of Evergreen Action.

Microsoft quickly praised the agreement Trump announced. Other major AI firms, including Google and Anthropic, rolled out their own cost-conscious pledges ahead of the speech.

Experts, though, questioned how much the pledge really addresses the full costs of data center build-out.

"Most of today's cost pressure is coming from transmission, distribution, and system readiness, not energy supply. Those costs remain even if a data center self-supplies generation," wrote Brandon Owens, a grid expert and founder of AIxEnergy.

"There's a lot of ways that these can cause increased rates that are outside of just paying for the very obvious, direct interconnection and potential new generation costs. As to how those costs are ultimately distributed among millions of ratepayers, those details are hashed out by utilities and state regulators — not the White House and tech companies," said Catherine Casomar, co-founder of the Better Data Center Project.

Ari Peskoe, director of the Electricity Law Initiative at Harvard Law School, argued that the pledge is aimed at the wrong actors – focusing on data center developers rather than the utilities and grid operators who shape rates and infrastructure.

Other advocates doubt Big Tech's incentives altogether. "Let's be honest: Big Tech isn't going to do anything for the benefit of anyone but itself," said Mitch Jones, managing director of policy and litigation at Food & Water Watch. "And with the Trump administration's complete unwillingness to hold corporations accountable for any number of harmful impacts on society, it's imperative we halt new data center construction now."

Sen. Mark Kelly (D-AZ) echoed the skepticism on X: "A handshake agreement with Big Tech over data center costs isn't good enough. Americans need a guarantee that energy prices won't soar and communities have a say."

In the speech, Trump didn't name the companies that signed the "ratepayer protection pledge," or spell out how it would be enforced. According to press reports, the White House plans to host companies in early March to formalize the effort. Energy Secretary Chris Wright said after the address that the administration had reached agreements with all the "brand-name" AI companies.

It's also unclear how much this pledge will actually change existing behavior. Major firms were already moving toward "bring your own power" models or claiming to cover their full energy costs:

  • PJM Interconnection, the country's largest grid operator, recently proposed that big new power users either bring new generation with them or curb usage when the grid is stressed.
  • Google says it already covers its own data center energy costs.
  • Meta has similarly said it pays the full cost of the electricity used by its data centers so it isn't passed on to consumers.

Finally, even if the pledge works as advertised and shields local ratepayers from higher electric bills, it leaves big issues untouched: air and noise pollution, water depletion, land use conflicts, and the risk of stranded projects if local resistance grows. And it doesn't grapple with the downstream impacts of accelerating AI more broadly – from job displacement and environmental strain to expanded surveillance and an information ecosystem saturated with AI slop.

2. Take It Down Act

President Trump's second AI mention was more of an aside, tying First Lady Melania Trump to "AI legislation":

"Over the past year, [Melania] has had an incredible impact championing AI legislation."

President Trump, State of the Union

He was referring to the TAKE IT DOWN Act, though he did not name it. The bipartisan law, sponsored by Sens. Amy Klobuchar (D-MN) and Ted Cruz (R-TX), was signed by Trump on May 19, 2025. It:

  • Criminalizes the publication of non-consensual intimate imagery (NCII), including AI-generated deepfakes.
  • Requires social media platforms and other online services to remove reported non-consensual sexual imagery within 48 hours of a valid request.

Companies are currently being advised to update their policies and terms of service, build dedicated reporting tools, and prepare for the 48-hour removal requirement. The criminal penalties for publishing NCII took effect immediately upon signing. Platforms have until May 19, 2026 to fully comply with the takedown requirements. The Federal Trade Commission (FTC) is responsible for enforcing compliance and going after platforms that don't meet the standards.

The law passed with broad bipartisan and industry support, endorsed by more than 100 organizations, including Google, Meta, Amazon, and TikTok.

Supporters frame the law as long-overdue protection against tech-enabled sexual abuse:

"The TAKE IT DOWN Act gives victims of online abuse legal protections and tools for when their intimate images, including deepfakes, are shared without their consent, and enables law enforcement to hold perpetrators accountable," said Sen. Amy Klobuchar.

"Predators who weaponize new technology to post this exploitative filth will now rightfully face criminal consequences, and Big Tech will no longer be allowed to turn a blind eye to the spread of this vile material," said Sen. Ted Cruz.

Civil liberties groups, including the Electronic Frontier Foundation, warn that the law may overshoot. They argue that a rigid 48-hour notice-and-takedown rule is likely to push platforms to remove anything remotely risky, which could incentivize over-removal of lawful speech and be abused by bad actors to take down legitimate content.

3. Presidential AI Challenge

President Trump's final AI mention was the Presidential AI Challenge:

"Students and educators in every state have joined the First Lady's efforts in the presidential AI challenge, keeping America's next generation positioned to succeed, and strongly succeed in the future. Tonight, we welcome two young people whose lives reflect the First Lady's impact: Sierra Burns and Everest Nevraumont. Thank you both. And Melania, thank you. I know how hard you worked on it."

President Trump, State of the Union

The Presidential AI Challenge is a nationwide K-12 competition created by executive order last year. It asks students and educators to:

  • Identify a real community problem
  • Use AI tools to create, study, or deploy a solution

Educators can also:

  • Propose new ways to teach an AI concept
  • Design an AI tool to manage some part of classroom life that couldn't be done without AI

On paper, the goal is to make sure young people are not just users of AI, but builders in the future AI economy.

Key dates:

  • Submissions were due January 20, 2026
  • State-level winners are expected to be announced early this year
  • National finalists will be invited to a three-day White House showcase in June 2026
  • National champions will receive $10,000 for their school or team

The AI competition is part of a broader push by the administration to infuse AI into K-12 education. But critics argue it's more symbolism than substance, given the administration's other moves on schooling and tech.

Over the past year, the administration has:

  • Terminated more than 400 federal grants aimed at advancing STEM education in K-12 schools and universities
  • Cut Department of Education staff roughly in half
  • Eliminated the Office of Educational Technology, which helped states and districts adopt new technologies in classrooms

So while the AI Challenge encourages students to experiment with cutting-edge tools, it's being rolled out alongside deep cuts to the very infrastructure that would help most schools teach AI in a serious, sustained way.

What the President Did Not Say

Americans are uneasy about an AI-dominant future. That makes it notable what Trump's SOTU didn't touch on, especially around kids, jobs, and rules for industry.

President Trump highlighted the TAKE IT DOWN Act and the K-12 AI Challenge, but he didn't propose any new measures to address broader AI-related risks to children – one of the few issues that unites voters across the spectrum:

  • 78% of Trump voters want tech companies held liable for harms to children, according to the Institute for Family Studies.
  • Among women 50+ – a high-turnout group – 90% are concerned about the lack of national AI standards to protect kids, and 70% say they're very concerned.

Members of both parties have been sounding the alarm:

  • A bipartisan Senate bill, the Guard Act (Sens. Josh Hawley, R-Mo., and Richard Blumenthal, D-Conn.), would bar companies from providing AI chatbot "companions" to minors.
  • In a recent hearing, Sen. Ted Cruz, R-Texas, argued that heavy device use in schools, even for learning, is fueling a teen mental health crisis and undermining academic outcomes.

None of these child-safety concerns or proposals came up in the speech.

Public anxiety about AI and jobs is intense – and remains largely unaddressed.

A recent YouGov poll found:

  • Only 7% of Americans think AI will increase the number of jobs
  • 63% think AI will reduce jobs in the U.S.

An Edelman report similarly found that about two-thirds of low-income Americans expect to be left behind by generative AI rather than benefit from it.

This is a ripe area for concrete policy. Lawmakers who can talk honestly about job loss and back it up with specific plans have room to connect with voters. Examples of bills currently in play:

  • Workforce of the Future Act – Directs the Secretaries of Labor, Commerce, and Education to analyze AI's impact on the economy and authorizes $90 million in grants for training workers most at risk of displacement.
  • Investing in American Workers Act – Aims to help workers gain skills for an AI-driven economy, including through incentives for employer-funded training.

None of this job-focused agenda made it into Trump's AI remarks.

Polling is consistent on one point: Americans want laws, not just voluntary pledges, to govern AI.

  • In a recent Gallup survey, 80% of Americans said they want rules for AI, even if that slows the technology's development.

Voters aren't asking for more executive orders or handshake agreements with tech companies; they're looking for clear, enforceable guardrails written into statute.

So far, however, the Trump administration has shown little appetite for supporting comprehensive AI legislation. That gap between public demand for hard rules and the White House's focus on voluntary commitments is one of the most striking omissions from the SOTU's AI section.

Conclusion

Trump's State of the Union treated AI as a backdrop rather than as a structural force reshaping the economy, childhood, and democracy. The policies he highlighted are not trivial: the ratepayer pledge nods to real tensions around AI infrastructure; the TAKE IT DOWN Act offers long overdue protections against image-based abuse; and the Presidential AI Challenge signals that AI fluency matters for the next generation. But set against public anxiety over jobs, child safety, and the absence of hard rules for industry, his brief AI remarks feel more like branding than a governing agenda.

The political opportunity — and risk — now lies with whoever is willing to move beyond pledges and contests to make concrete, enforceable choices about how AI will be built, paid for, and constrained in American life.