The demand for electricity from data centers focused on artificial intelligence has resulted in an additional $23.1 billion in costs for PJM Interconnection, the largest grid operator in the U.S. This is detailed in a report by Monitoring Analytics.
According to the document, the load from computing centers is reshaping the power market. Over three auction periods (from 2025/2026 to 2027/2028), the anticipated increase in consumption has generated billions in additional revenue for generators, which ultimately falls on consumers.
Experts highlighted that in just the last two auctions, end-user bills increased by $13.8 billion. This occurred despite existing price caps established by an agreement between Pennsylvania and PJM.
The report for the first quarter of 2026 indicates that the average wholesale electricity price in the region surged by 75.5% year-over-year, reaching $136.53 per MWh. The most significant impact was seen in capacity payments (generation reservation), which skyrocketed by nearly 400%.
Analysts at Monitoring Analytics described the current trend as "unique and unprecedented." They noted that the power market was originally designed for gradual demand growth, while AI projects require massive energy connections in a short timeframe.
As a solution, Monitoring Analytics proposed tightening regulations for data center operators:
- Connect large facilities only when new generating capacity is brought online;
- Implement full load-shedding protocols during stress situations for the power system;
- Create a separate queue for technical connections for projects that generate their own power.
Experts warned that the issue is systemic. While bitcoin miners traditionally sought cheap surplus energy and were prepared for outages, AI centers demand high reliability. This raises the question for regulators: who should bear the cost of infrastructure development—tech giants or ordinary households?
As a reminder, in May, TeraWulf acquired a site in Eastern Kentucky for developing HPC infrastructure with a potential capacity of over 1 GW.
