The United States wants to lead the global AI race.
 |
| AI data center facing power grid |
- More models.
- More computing power.
- More data centers.
- More electricity.
But there is a problem. The physical infrastructure behind the AI boom is increasingly colliding with the interests of the people living around it. And that conflict is becoming impossible to ignore.
The AI Boom Has a Physical Problem
AI may look digital, but it ultimately depends on very physical things.
- Servers need electricity
- Data centers need land
- Cooling systems may require significant amounts of water
- Large facilities need transmission infrastructure, transformers, substations and reliable power
As AI models become more computationally demanding, the infrastructure supporting them becomes larger as well. That is why the AI race is gradually becoming an energy and infrastructure race. The U.S. Energy Information Administration expects American electricity consumption to reach record levels in both 2026 and 2027.
Recent forecasts put total U.S. electricity demand at roughly 4,270 billion kWh in 2026 and 4,349 billion kWh in 2027, with data centers among the major drivers of growth. The International Energy Agency expects global data-center electricity consumption to more than double by 2030, reaching roughly 945 TWh. The United States is expected to account for the largest share of that increase. The AI boom therefore has a simple requirement Someone has to power it.
But Americans Are Increasingly Saying: Not Here
This is where the story gets interesting. A Gallup survey found that 71% of U.S. adults oppose building an AI data center in their local area, including 48% who strongly oppose such projects. Only 27% said they favor them. That does not mean Americans are necessarily against artificial intelligence itself.
There is an important distinction between AI is useful and I want a massive AI data center next to my community. The second question produces a very different answer. And that gap could become one of the biggest political and economic challenges facing the AI industry.
Why Are Communities Pushing Back?
The concerns vary by location, but several themes appear repeatedly.
- Electricity.Large AI facilities can require enormous amounts of power. Communities are increasingly asking a simple question Who pays for the additional electricity infrastructure? If utilities need to build new transmission lines, substations or generation capacity, residents may worry that some of those costs could eventually appear in their electricity bills. That concern has already reached Washington. The U.S. House is preparing to consider legislation aimed at protecting electricity customers from costs associated with the enormous power requirements of data centers.
- Water. Cooling is another concern. Some large data centers use water-based cooling systems, creating tension in regions where water resources are already under pressure. For a community, the calculation is straightforward: If an AI company receives billions of dollars in investment and tax incentives, but residents believe their water or electricity costs could rise, the economic benefits may not feel evenly distributed.
- Land and Noise. A hyperscale data center is not a small office building. These facilities can occupy large areas and contain thousands of servers, cooling equipment, backup generators and other infrastructure. For nearby residents, that can mean construction activity, industrial noise, traffic and changes to the character of the surrounding area. The digital economy suddenly becomes very physical.
Texas Shows How Quickly the Mood Can Change
Texas is one of the clearest examples. The state spent years positioning itself as an attractive destination for data centers because of its business-friendly environment, available land and energy resources. But the political environment has changed.
In September 2026, Texas imposed a temporary halt on approvals for new data centers while the state audits whether proposed facilities are appropriate for connection to the electricity grid. The move came amid growing public and political pressure over the rapid expansion of AI infrastructure.
That is significant. Texas was not an obvious anti-technology jurisdiction. It was one of the places actively competing for technology investment. The fact that resistance is now emerging there shows that the issue is no longer simply about environmental activists opposing technology. It has become an infrastructure and political question.
This Is Not Just a Texas Problem
Texas is only one part of a broader trend. Across the United States, states and local governments have been considering restrictions, moratoriums, zoning changes and other measures affecting data-center development.
Reuters reported in August that a growing number of governments, regulators and cities around the world were moving to freeze, restrict or otherwise limit new data-center construction amid concerns over power, water and infrastructure.
Brookings has also documented the rise of data-center moratoriums at the state and local level, while emphasizing that governments need better information and planning rather than simply relying on blanket bans. This creates a new bottleneck for the AI industry. The problem may no longer be Can we build the AI model? It may increasingly become Can we build the infrastructure needed to run it?
The AI Race Is Becoming an Energy Race
This changes the investment landscape. For years, investors focused heavily on companies building AI models, chips and software. But the next bottleneck could sit further down the supply chain. AI requires:
- electricity generation;
- transmission infrastructure;
- substations;
- transformers;
- cooling systems;
- data-center construction;
- networking equipment;
- land;
- water infrastructure;
- backup power;
- and increasingly, new sources of reliable energy.
The AI economy therefore extends far beyond companies that make AI models. The infrastructure surrounding AI could become one of the most important investment stories of the decade.
Nuclear Power Is Back in the Conversation
The energy requirements of AI are also changing the conversation around nuclear power. Nuclear plants can provide large amounts of continuous electricity without the intermittency associated with some renewable sources.
That makes nuclear attractive to companies seeking reliable power for large-scale computing. But nuclear energy also faces its own challenges:
- construction costs;
- long development timelines;
- regulation;
- public acceptance;
- financing;
- and the availability of suitable sites.
AI cannot simply solve its energy problem by switching on nuclear plants overnight. The infrastructure cycle is much slower than the software cycle. That mismatch could become important. An AI company can release a new model in months. Building major electricity infrastructure can take years.
The Hidden Risk for AI Investors
This is where the story becomes more interesting from an investment perspective. AI companies may have enormous demand for computing power. But demand alone does not guarantee that infrastructure can be delivered on schedule. Projects can face:
- Grid delays. A data center can be ready to build but unable to obtain enough electricity.
- Political opposition. Local governments can delay permits or impose new restrictions.
- Higher financing costs. Investors and lenders may demand greater protection when projects face regulatory uncertainty.
- Construction bottlenecks. Transformers, power equipment and specialized construction capacity are not unlimited.
- Community resistance. Even projects with strong economic arguments can face opposition from residents.
Recent reporting suggests these risks are already affecting financing conditions for AI infrastructure projects. Reuters reported that lenders have become more cautious as projects encounter delays related to electricity access, supply chains and political opposition.
That is an important development. The AI infrastructure boom may be enormous. But enormous projects also create enormous capital requirements.
The $1 Trillion Question
- The United States has a strategic reason to continue investing in AI.
- China is competing aggressively in artificial intelligence.
- Washington wants American companies to maintain technological leadership.
That creates pressure to build more computing capacity rather than less. But local communities have their own priorities. They care about:
- electricity bills
- water availability
- land use
- noise
- jobs
- tax revenue
- and quality of life.
Those interests do not always align. And that creates a political paradox. The federal government may want more AI infrastructure, Technology companies may want more data centers,Investors may want more AI growth, But local communities may say Not here.
AI May Be Digital. Its Costs Are Not
This is perhaps the most important point. The AI revolution is often presented as something happening inside computers. But the physical reality is very different.
- Every AI query ultimately depends on hardware
- Every large model requires computing infrastructure
- Every data center requires electricity
- Every expansion requires land, cooling and connectivity.
The bigger AI becomes, the harder it becomes to hide its physical footprint. And that footprint is now becoming a political issue.
What Investors Should Watch Next
The next stage of the AI boom may therefore be determined by infrastructure rather than algorithms. Investors should watch several areas closely.
1. Power generation
Who can provide reliable electricity at scale?
2. Grid infrastructure
Who builds the transmission lines, substations and transformers needed to connect new facilities?
3. Nuclear energy
Can nuclear power become a meaningful source of reliable electricity for the AI economy?
4. Natural gas
Will gas generation fill the gap while other sources take longer to develop?
5. Renewable energy and storage
Can solar, wind and batteries contribute enough reliable capacity to support growing data-center demand?
6. Cooling and water technology
Can data centers reduce their resource footprint while increasing computing capacity?
7. Data-center financing
Which projects can actually secure power, permits and financing? These questions could determine which AI infrastructure projects survive and which remain stuck on paper.
The AI Boom Has Entered Its Reality Check
The first phase of the AI boom was largely about capability.
- Who has the best model?
- Who has the fastest chip?
- Who has the most powerful AI?
- The next phase may be different.
- It could be about physical constraints.
- Who has enough electricity?
- Who can obtain grid connections?
- Who can secure land and water?
- Who can get permits?
- Who can finance construction?
- And perhaps most importantly Who can convince communities that the benefits of AI are worth the costs of hosting its infrastructure?
America is unlikely to abandon the AI race. But the country may be entering a period where simply wanting more AI is no longer enough.
- The infrastructure has to be built.
- The electricity has to come from somewhere.
- The communities have to accept it.
And investors ultimately have to pay attention to the difference between an AI project that is announced and an AI project that can actually be built. The AI revolution may have started in the cloud.
Its next bottleneck could be on the ground.