Navigating the AI Boom in the Texas Energy Market
By Maiya Olsen
Artificial Intelligence is becoming deeply ingrained in business operations and even our daily lives. The rate at which this is occurring is being aggressively accelerated by the Trump Administration’s 2025 AI Action Plan, which proclaims that the United States is “in a race to achieve global AI dominance.” However, this expansion, fueled by deregulatory measures, is occurring at an unsustainable rate. There needs to be greater consideration of the long-term consequences of current operations before the federal government’s drive for rapid AI development outpaces accountability and safety for citizens and the environment. There may even be signs that it is already happening.
Data centers consume a tremendous amount of electricity, which is problematic considering ERCOT (the operator of the Texas electricity grid) is unprepared for such a surge in demand. Its current capacity is 40-50 new large loads per year, but ERCOT currently has 225 new interconnection requests, 70% of which are data centers. This surge in requests can be attributed to companies drawn to Texas’s cheap land and electricity costs, enabled by the state’s deregulated power market, forcing ERCOT to spend big just to prepare the grid to accommodate the significant energy demands to come. One such project, costing $9.4 billion, was to improve power distribution by building 1,109 miles of transmission lines.
Utility companies seeking to expand to accommodate increasing demand are doing so with nonrenewable sources of energy, mainly natural gas, with assistance from the Texas Energy Fund. This fund is providing companies with $10 billion in low-interest loans and grants to build gas-fueled power plants, which are driving up greenhouse gas emissions and local air and water pollution from associated fracking.
There have even been recent actions by the federal government to keep coal and gas plants that were previously slated for retirement online. Federal officials are doing so by invoking section 202(c) of the Federal Power Act, which is typically reserved for “emergency situations.” For example, in 2025, the U.S Department of Energy ordered five coal plants to keep running despite contributing to the deaths of 124 people since 2019. These policies not only further contribute to emissions but place a burden on consumers due to the costs associated with restaffing and continued maintenance of outdated facilities. U.S. Energy Secretary Chris Wright told Reuters a major motivation behind these actions is the expansion of AI, and many from the Department of Energy have said, “This is just the beginning.”
With all these projects underway, energy demand estimates may be inaccurate, as companies can submit requests to build data center(s) in multiple prospective cities and states. According to Grid Strategies, national utility load forecasts might be overestimated by as much as 40%. These trends can even lead to the construction of electricity infrastructure solely for potential projects rather than for viable projects. However, there have been initiatives to help address these issues through policy, such as Senate Bill 6. The bill, which was passed by the Texas Legislature in June of last year, introduced requirements for data centers in with a capacity of more than 75 GW. These facilities must pay a non-refundable $100,000 fee for an initial screening transmission, provide proof of land ownership or lease, and commit to depositing funds for infrastructure built to serve their electricity demands. This is a step in the right direction, especially for newer data centers in Texas that typically exceed the 75MW requirement. However, because ERCOT keeps deal terms confidential, it is difficult to determine whether ratepayers will be protected from higher electricity prices associated with this new infrastructure.
The expansion of energy infrastructure in states to accommodate the surge in interconnection requests would be passed to customers through higher monthly utility bills. For some perspective, these rate increases were expected to affect 40 million customers across the U.S. in 2025 and would have been in addition to rate increases approved by state energy commissions. In states with extensive AI infrastructure, such as Virginia, electricity rates have increased by 267% over the last five years, which makes these additional rates that much more burdensome. These trends highlight data centers’ responsibility to ensure their electricity consumption doesn’t increase costs for residential customers, but the rapid rise in data centers is making that nearly impossible, especially without proper legislation in place.
It is important to note that the Trump Administration emphasized the need for companies to pay for their own power generation in this year’s State of the Union Address, as part of what is known as the “ratepayer protection pledge.” No details were provided on how this proposal would work. There is also sentiment that “handshake agreements” between the government and Big Tech are not good enough to ensure their commitment to reducing costs for communities. According to Travis Kavalla, a former Montana utility regulator, even if companies were to produce their own energy, they would still have to use power from the grid, straining the infrastructure and triggering capital investments that would inevitably be passed on to consumers.
The policies and choices made about how data center electricity demand will be met will dictate the future of AI. However, the rapid expansion of AI is making it difficult to ensure this technological transformation is carried out in a way that maintains the interests of the American people.
Maiya Olsen is a student studying Environmental Science and Government at the University of Texas at Austin and is a Policy and Advocacy intern with Public Citizen’s Texas office.