The energy race that is reshaping the future of artificial intelligence in the United States
Artificial intelligence is growing at a pace that few thought possible just a few years ago, and that growth comes with a cost that goes far beyond developing algorithms and models.
It comes with a cost that is literally electric. ⚡
The data centers that support all this AI infrastructure consume a staggering amount of energy, and the U.S. power grid simply was not designed to handle this kind of demand. Just to give you an idea of how big the problem is: today, more than 4,000 data centers are already operating on American soil, with another 3,000 planned or under construction. Some of these facilities consume more energy than entire cities.
It is a scenario that has turned into a race against the clock, and the U.S. government just took a step that could change the game in a major way.
In June 2026, the FERC — the Federal Energy Regulatory Commission — unanimously approved a measure to fast-track the electrical interconnection of large energy consumers, especially AI-focused data centers, to the country’s high-voltage transmission grid. The commission’s own chair described it as a historic vote.
But like any decision that involves money, power, and politics, it did not come without controversy. And the challenges left unresolved are just as big as the ones this approval is trying to address. 👇
What did the FERC actually approve?
The measure approved by FERC has a technical name, but the concept behind it is simple to understand. Before this decision, any company wanting to connect a large data center to the electrical transmission grid had to go through an approval queue that, in many cases, took years to complete. We are talking about processes that could drag on for four, five, or even six years before a single switch was flipped. For an industry that operates at the speed of artificial intelligence, that is practically an eternity.
What FERC did was direct that AI data centers and other large energy consumers be able to connect to the transmission system in a timely and orderly manner. In practice, this means data center operators will be able to, under certain technical and regulatory conditions, have their electrical interconnection processed much faster than before, reducing bottlenecks that were stalling billion-dollar projects before they even got off the ground.
The vote was unanimous among all five commissioners, which in itself is a sign that there was a rare consensus that something needed to change urgently. Laura Swett, appointed by President Donald Trump and chairing the commission, called the vote a historic action to push the country’s electricity market into the future. The market reacted positively, and tech companies that had been pushing for this kind of regulatory reform celebrated the decision. 🤔
But like any structural change of this scale, the implementation details are still being worked out, and that is exactly where most of the uncertainty lies.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The role of the Energy Secretary and the rivalry with China
This decision did not come out of nowhere. Energy Secretary Chris Wright had asked FERC to take action to help the United States compete more effectively with China for supremacy in the artificial intelligence sector, which is growing at a breakneck pace. The Trump administration’s pressure on the independent commission reflected a strategic view that energy infrastructure is just as important as the technological development itself in the global race for AI leadership.
For the administration, attracting foreign investment and maintaining U.S. economic and military relevance runs directly through the ability to deliver fast and reliable power to feed the computing complexes that support this technology. Trump even signed an executive order this month establishing a framework for the federal government to assess the national security risks of the most advanced AI systems for up to one month before their public release. The approach combines infrastructure acceleration with strategic control over the technology itself.
Tech companies and data center developers had already been applauding initiatives that make faster connections to the nation’s power supply possible. That private sector support helped create the political momentum needed for FERC to make its move.
Why did data center energy demand explode like this?
To understand the scale of the transformation underway, it helps to take a step back and look at what is driving this massive demand for energy. The answer is directly tied to the evolution of generative artificial intelligence, large language models, and the training and inference infrastructure that powers all of it. Training a single large-scale language model can consume the same amount of electricity that tens of thousands of homes use in an entire month — and that is just the training phase. When that model goes into production and starts responding to queries in real time, the consumption continues on an ongoing and growing basis.
Modern AI-focused data centers are completely different from the traditional data centers the industry knew a decade ago. They are equipped with highly specialized hardware accelerators, such as GPUs and TPUs, that require far more energy per rack than conventional servers. A standard server rack used to consume around 5 to 10 kilowatts. A rack loaded with today’s most advanced GPUs can easily exceed 100 kilowatts. Multiply that by thousands of racks inside a single facility and you start to get a real sense of what is being demanded from the power grid.
Data from the Electric Power Research Institute shows that data centers now account for about 5% of total U.S. electricity demand, but that number could triple by 2035. In Virginia, where the largest concentration of these facilities is located, data centers already account for more than 25% of total energy demand and could surpass 40% by 2030. These numbers help explain why this issue has become a regulatory priority. 💡
The promise to protect consumers
One of the central concerns around this decision is the impact on American consumers’ wallets. Laura Swett made a point of addressing this directly during the vote.
I know Americans across the country are worried about affordability, and so are we, Swett said, referring to the five-member commission. She added that she takes extremely seriously the mission Congress entrusted to FERC to ensure that rates are reasonable and that Americans pay their fair share or less.
One of the key provisions in the order approved by the commission is that data centers will pay the full cost of any grid upgrades needed to make their connection possible. In other words, in theory, residential consumers should not have to foot the bill for these infrastructure improvements. This cost separation is a fundamental point in giving the measure legitimacy.
However, the order has its limitations. It can do little to address the squeeze on energy supply that is already driving up electricity bills in some regions and triggering warnings about potential blackouts. This is happening because data center construction is moving faster than new power plants can come online to meet the additional demand.
Big Tech’s consumer protection commitment
To try to ease public fears, companies like xAI, Google, Microsoft, Meta, Oracle, OpenAI, and Amazon signed what is called the Consumer Protection Commitment proposed by Trump. In this agreement, they pledged to:
- Build or purchase new power generation sources for their data centers
- Cover the costs of infrastructure upgrades needed for interconnection
- Make backup power generation available to prevent blackouts during emergencies
- Hire local labor for the construction of their data center complexes
These commitments are a sign that tech companies recognize they cannot simply plug into the power grid without giving something back. The question that remains is how well these promises will be kept in practice and what enforcement mechanisms exist to ensure accountability.
Opposition from utilities, states, and local communities
Not everyone celebrated FERC’s decision, and it is important to understand why. Utilities, state governments, and regional grid operators expressed concern that the Republican administration’s plan could strip them of their authority to manage the interconnection process. These entities are on the front lines of daily grid operations and worry that unchecked acceleration could create instabilities.
Clean energy advocates also raised objections. They want the agency to advance, rather than undermine, state-level efforts to require the use of renewable energy. A large portion of the American power grid still relies on non-renewable energy sources, and increasing interconnection capacity for large consumers without clear guarantees about where that electricity comes from is seen as a step backward by environmental organizations.
Beyond the regulatory and environmental concerns, there is a growing backlash from local communities against data centers. Residents do not want to live near these complexes, citing fears about rising energy prices, pollution, massive water consumption, and the loss of open spaces, farmland, and the rural character of their regions. Across the United States, protests have been organized against new facilities, creating a tense social landscape that federal regulations alone cannot resolve. ⚠️
The construction projects that are falling behind
Even with accelerated regulatory approval, there is a practical bottleneck that could limit the real impact of this measure. A J.P. Morgan report published last month revealed worrying data: based on satellite imagery, more than 60% of data center capacity planned for completion by 2027 has not even broken ground yet. Another 7% is behind schedule.
The typical culprits behind the delays include:
- Permitting and licensing issues
- Delays in gas turbine deliveries
- A shortage of transformers available on the market
- A lack of qualified labor
These bottlenecks show that speeding up grid interconnection is only one part of the equation. The physical construction of data centers faces its own logistical and supply chain challenges that no single regulatory decision can solve on its own. Tech companies continue ramping up their data center investments, but the numbers show there is a real gap between planning and execution.
The steps that led up to this point
The June decision was not the first action in this direction. This latest vote came eight months after Secretary Wright asked the independent agency to take greater control to ensure that the vast network of massive computing complexes needed to power AI would be connected quickly to high-voltage transmission lines.
Back in December, FERC had already taken an important initial step by voting to allow tech companies to connect a data center directly to a power plant. This approach, known in the industry as co-location, was considered a creative workaround for transmission system bottlenecks. However, it also sparked controversy over how to ensure that power consumed directly was not being diverted from consumers who depended on the same plant.
These sequential moves show that there is a deliberate, incremental strategy to reform the country’s electrical interconnection system in favor of large energy consumers tied to AI.
What this means for the future of AI
In the bigger picture, FERC’s decision is a clear signal that governments are beginning to treat artificial intelligence infrastructure as a matter of national energy policy, not just a private tech sector issue. The implications are enormous. It means that conversations about the future of AI will increasingly overlap with debates about grid planning, transmission regulation, environmental policy, and the allocation of scarce resources.
It is a convergence that turns the tech sector into a central player in conversations that once belonged exclusively to utilities, energy regulators, and state governments.
For companies that operate or plan to operate large-scale data centers, the approval represents a real window of opportunity to move forward on projects that were stuck in the regulatory queue. The reduction in electrical interconnection timelines could mean months or years shaved off before new facilities come online, which has a direct impact on the ability to meet growing demand for AI services, cloud computing, and real-time data processing.
In a market that moves as fast as artificial intelligence, that time difference can be the factor that separates market leaders from competitors who got left behind.
And for the end user — whether you are someone using an AI assistant on your phone, a company running machine learning models, or a developer building the next product powered by natural language — the energy equation matters more than it might seem at first glance. The availability, speed, and cost of AI services depend directly on the infrastructure that supports them. When that infrastructure grows more efficiently, the benefits, in theory, trickle down across the entire chain.
The path between a regulatory decision in Washington and the end-user experience is long and full of variables, but this FERC approval is, without a doubt, a meaningful step in that direction. 🚀
