Imagine this: your monthly electricity bill skyrockets, not because you’re using more power, but because of a tech boom that might never fully materialize. That’s the reality millions of families could face if the AI data center craze goes bust. Here’s the deal: the race to build data centers powering artificial intelligence tools like ChatGPT and Claude has sent electricity demand soaring after two decades of stagnation. But here’s where it gets controversial—what if this AI gold rush is just a bubble? And this is the part most people miss: if demand projections fall short, consumers could be left footing the bill for billions in unnecessary power infrastructure.
Take Ashburn, Virginia, home to the world’s largest data center hub. Residents across 13 states served by the PJM Interconnection grid—the largest in the U.S.—are set to pay a staggering $16.6 billion to secure power supplies for data centers from 2025 to 2027. A whopping $15 billion of that is for data centers that may never fully materialize. Monitoring Analytics, PJM’s watchdog, calls this a ‘massive wealth transfer’ from everyday consumers to the data center industry. Abe Silverman, former general counsel for New Jersey’s public utility board, puts it bluntly: ‘It’s a little scary if you don’t really have faith in the load forecast.’
Already, residential electricity prices are climbing. In September, Illinois saw a 20% spike, Ohio 12%, and Virginia 9% compared to the previous year. These states are among the top U.S. markets for data centers, and the correlation is hard to ignore. Joe Bowring, president of Monitoring Analytics, explains it simply: ‘When wholesale power costs go up, people pay more.’
But is this demand real, or just hype? PJM forecasts 30 gigawatts of extra demand from data centers by 2030—enough to power 24 million homes. Yet, developers are shopping projects around, likely inflating these numbers. Cathy Kunkel, a consultant at the Institute for Energy Economics and Financial Analysis, warns of duplication in forecasts. ‘We’re in a bit of a bubble,’ Silverman admits. Even industry giants like Constellation Energy and Vistra Corp. have sounded the alarm, suggesting demand could be overstated by three to five times.
The risk? ‘Stranded costs,’ where utilities build expensive infrastructure that ends up unused. If data centers don’t materialize or use less power than promised, consumers—residential, commercial, and industrial—are stuck with the bill. Take Ohio, where American Electric Power (AEP) saw data center requests drop by half after implementing stricter rules requiring upfront payments for energy commitments. The Data Center Coalition calls these rules ‘discriminatory,’ but AEP argues they’re necessary to avoid speculative projects.
There’s another concern: grid reliability. Adding 13 gigawatts of data center demand—equivalent to a dozen nuclear plants—requires massive infrastructure. If PJM doesn’t manage this carefully, the grid could become less stable. Monitoring Analytics suggests a solution: reject data center connection requests unless they bring their own power generation. This would clear up demand uncertainty and incentivize serious commitments.
So, is the AI data center boom a game-changer or a costly gamble? As consumers, we’re already paying the price. But should we be? Let’s spark a discussion: Are we overinvesting in a tech trend that might not deliver? Or is this the cost of innovation? Share your thoughts below—this is one debate you won’t want to miss.
