The AI boom is driving up electricity costs for millions of Americans, with the PJM Interconnection, a regional transmission organization, forecasting an imminent jump in demand. This surge is primarily due to data centers powering artificial intelligence, with 64% of ballooning capacity payments related to actual and projected datacenter demand.
Residents in several states are already feeling the pinch, with New Jersey residents seeing a 13.3% bill increase, the third-highest in the US, while New York bills surged 14.4%. In fact, average annual power bills in 9 out of 13 Midwestern and mid-Atlantic states, plus DC, increased faster than the national average of over 6%.
The main culprit behind the increased energy demand is the energy-intensive nature of training and using AI models, which requires significantly more power than conventional internet browsing. The capacity auction system, which sets payments to generators, has also been criticized for its role in driving up costs.
Lawmakers are proposing reforms to address the issue, including requiring data centers to secure their own power through long-term purchase agreements. Other potential solutions include reforming capacity auctions and improving energy efficiency. For instance, splitting the capacity market into existing and new plants could offer more pricing certainty, while committing to long-term agreements for new plants could increase grid resilience. AI giants can also deploy software to use the grid more efficiently.
As the demand for AI-powered services continues to grow, it's essential to find ways to mitigate the impact on the power grid and ensure that the benefits of AI are not outweighed by the costs.
