Key facts
- Community opposition, labor shortages, and equipment delays are hindering AI data center development.
- Over $170 billion in proposed AI data center capacity has been stalled since January 2024.
- US data center power demand is projected to more than double to 66 GW by 2027.
- Texas has paused new data center interconnections pending an audit of its extensive queue.
- Utilities are implementing measures like higher study fees and stricter contract terms to manage speculative load requests.
The rapid expansion of AI data centers in the U.S. is facing a significant reality check due to a confluence of challenges, including growing community opposition, construction labor shortages, and long lead times for critical electrical equipment. These hurdles are causing delays and cancellations, with more than $170 billion in proposed data center capacity stalled since January 2024, according to energy advisory firm Relae.
Despite these obstacles, market analysts anticipate the AI-driven construction boom to continue. Goldman Sachs projects U.S. data center power demand to more than double from 2025 levels to 66 GW by 2027. However, only about half of the capacity scheduled for the next one to two years is expected to come online on time. The Electric Power Research Institute forecasts that data centers will account for 9% to 17% of U.S. electricity demand in 2030, potentially reaching up to 20% by 2035.
Texas is at the forefront of states asserting control over data center development. The state's grid operator, ERCOT, has paused new data center interconnections pending an audit of its massive 474-GW queue, which far exceeds the grid's peak demand. ERCOT's president and CEO, Pablo Vegas, expressed skepticism about the projected load growth, citing uncertainties in planning methodologies. Thomas Gleeson, chairman of the Public Utility Commission of Texas (PUCT), was more direct, calling many of the load projections speculative.
To manage speculative requests and ensure grid reliability, Texas has implemented measures such as large-load tariffs requiring minimum contract durations and upfront payments for impact studies. Governor Greg Abbott signed Senate Bill 6, imposing stricter requirements on prospective large-load customers, including site control and contributions to grid upgrades. The PUCT also adopted a cluster study framework for loads over 75 MW to streamline the interconnection process. However, Governor Abbott later directed ERCOT to freeze new interconnection approvals until an audit of all data centers in the queue is completed, a pause that Bloomberg NEF estimates could delay 20% of the total U.S. data center pipeline.
Similar challenges are present in regulated utility markets, where vertically integrated utilities are tasked with distinguishing between speculative and genuine projects. Utilities are responding by increasing load interconnection study fees and structuring service agreements to prevent large loads from negatively impacting reliability or stranding costs. They are also excluding potential generation for new customers from load forecasts until contracts are near execution, ensuring a higher degree of confidence in project viability before committing to new infrastructure.