AI's Disinflationary Promise Runs On Inflationary Energy
The article discusses the significant inflationary impact of AI data center growth on electricity costs for consumers, contrasting with AI's long-term promise of disinflation. It highlights rising power costs, widespread local opposition, and moratoria considerations in several US states, leading to project delays and cancellations. The piece suggests decentralized computing as a potential solution to mitigate these issues.
Dr. Jemma Green, cofounder and chairman of Powerledger.io, highlights a growing paradox: while AI promises future disinflation, its infrastructure buildout is causing immediate inflationary pressures, particularly on electricity bills in states like Virginia, Indiana, and Ohio. This immediate cost is largely borne by households, while the long-term benefits are projected for a different population and timeframe.
The article points to substantial evidence of these costs, noting that annual capacity charges in the PJM territory, serving 13 states, surged from $2.2 billion in 2023 to $14.7 billion in 2025, alongside a 75.5% jump in wholesale power costs this year. Bill Fehrman, CEO of AEP, expressed low confidence in resolving interconnection issues for the 300 gigawatts of projects proposed to PJM.
Public backlash is escalating, with $156 billion in data center projects cancelled or delayed in 2025 due to local opposition. Eleven states are considering moratoria, and 42 activist groups are campaigning in Virginia alone. A nationwide poll shows only 44% of Americans welcome nearby data centers, and Ohio voters may consider a statewide ban on large data centers. The North American Electric Reliability Corporation (NERC) issued a Level 3 Essential Actions Alert after data centers caused significant grid disruptions, signaling a structural mismatch in energy supply and demand.
NVIDIA CEO Jensen Huang emphasizes energy as the foundational layer of AI infrastructure, suggesting current energy needs are 1,000 times greater than available supply. The article suggests decentralized computing, like Google DeepMind's DiLoCo algorithm and Prime Intellect's approach, offers a solution by moving AI to existing energy sources, thereby mitigating community impacts related to noise, water use, grid strain, and rising rates. This emerging model could bridge the gap between AI's inflationary present and disinflationary future.