Power prices on the PJM Interconnection grid, which covers 13 states and Washington D.C., jumped from $54 per megawatt-hour in May 2023 to $95 in May 2024—that’s a 76% spike. The jump is the steepest in PJM’s history and comes as data centers for AI training and cloud computing suck up more electricity than regulators expected. The grid operator warned last week that the imbalance isn’t temporary; it’s structural. The PJM Interconnection covers a region that includes major tech hubs like Northern Virginia, home to the world’s largest concentration of data centers, and Chicago’s booming AI corridor. The surge isn’t just a pricing problem—it’s a warning sign. PJM’s CEO Manu Asthana told regulators the grid can’t deliver the power these facilities need without major upgrades that won’t be finished for years. “We’re seeing demand growth that’s faster than we’ve ever seen before,” Asthana said. “The grid wasn’t designed for this.” That’s a polite way of saying the U.S. power grid is running on 1960s-era infrastructure trying to serve a 2020s workload. The grid’s capacity hasn’t kept pace with the explosion of AI workloads. Data centers now consume about 4% of the nation’s electricity, but that share is growing fast—some analysts project it could hit 12% by 2028. The problem isn’t just generation. It’s transmission too. New power plants take years to permit and build, and new transmission lines face lawsuits and local opposition that can drag projects out for a decade. Meanwhile, AI training clusters like Nvidia’s DGX systems and Microsoft’s AI supercomputers in Iowa and Virginia are gobbling up power faster than the wires can deliver it. Watchdogs aren’t mincing words. The Federal Energy Regulatory Commission (FERC) FERC called the situation “unsustainable” and warned that unless something changes, power price spikes could become routine. FERC chair Willie Phillips said the grid is “at a crossroads” and urged Congress to fast-track permitting reforms. The watchdog’s report didn’t just blame the grid—it pointed fingers at tech giants. Companies like Google, Microsoft, and Amazon have pledged to go carbon-free, but their rapid AI expansion is straining local grids before renewable projects can come online. In Virginia, Dominion Energy Dominion Energy has asked regulators to approve a $55 billion plan to build new power lines and substations over the next 15 years. But even that timeline is tight. Neighbors near planned lines are suing, arguing the projects will ruin property values and local landscapes. Meanwhile, in Texas, ERCOT ERCOT is scrambling to keep the lights on as data centers in the Dallas area double their power demand every two years. ERCOT’s CEO Pablo Vegas has warned that without new gas plants and batteries, the grid could face blackouts during peak demand—like last summer when a heat wave forced rolling outages. The situation is forcing a reckoning. States are starting to act. Virginia lawmakers just passed a law to streamline approvals for energy projects near data centers. New York is fast-tracking permits for renewable energy tied to AI hubs. But even with these moves, the gap between power supply and AI’s hunger won’t close quickly. Analysts say the only short-term fix might be older, dirtier power plants coming back online. That’s a bitter pill for a grid trying to go green. The long-term solution? More nuclear plants, massive battery storage, and a smarter grid that can route power where it’s needed most. But none of that happens overnight.

What You Need to Know

  • Source: TechCrunch
  • Published: May 15, 2026 at 15:45 UTC
  • Category: Technology
  • Topics: #techcrunch · #startups · #tech · #business · #economy · #power

Read the Full Story

This is a curated summary. For the complete article, original data, quotes and full analysis:

Read the full story on TechCrunch →

All reporting rights belong to the respective author(s) at TechCrunch. GlobalBR News summarizes publicly available content to help readers discover the most relevant global news.

Curated by GlobalBR News · May 15, 2026

🇧🇷 Resumo em Português

O Brasil pode parecer distante dos problemas da maior rede elétrica dos Estados Unidos, mas o impacto de um fenômeno como o aumento de 76% nos preços da energia — impulsionado pela voracidade de data centers de IA — serve como um alerta global sobre os limites das infraestruturas energéticas diante do boom tecnológico. Enquanto aqui o debate ainda gira em torno da matriz limpa e dos leilões de energia renovável, lá, na Pennsylvania, Nova York e outros estados do sistema PJM Interconnection, o problema é urgente: a demanda dos centros de processamento de dados, que consomem energia 24 horas por dia, está superando a capacidade das redes, elevando tarifas e acendendo luzes amarelas sobre a transição energética e a sustentabilidade do setor.

A situação nos EUA reflete um paradoxo da era digital: a mesma tecnologia que promete revolucionar indústrias e a vida cotidiana está pressionando sistemas elétricos que, em muitos casos, ainda dependem de fontes fósseis ou enfrentam lentidão em ampliações. No Brasil, embora o cenário seja outro — com uma matriz predominantemente hidrelétrica e renovável —, especialistas já alertam para riscos semelhantes no futuro, especialmente se a expansão da computação em nuvem e da IA acelerar sem planejamento integrado entre governo, empresas e operadoras. A questão vai além dos custos: envolve segurança energética, emissões de carbono e a capacidade de o país — que tem sido referência global em energia limpa — manter sua competitividade sem repetir os erros de nações que não anteciparam a demanda.

A expectativa agora é de que reguladores e empresas acelerem soluções, seja com investimentos em geração distribuída, armazenamento de energia ou políticas que incentivem data centers a operarem com fontes mais estáveis — um movimento que, se bem-sucedido, pode até mesmo inspirar o Brasil a revisar sua própria estratégia diante da crescente digitalização.

🇪🇸 Resumen en Español

La escalada de precios de la electricidad en el mayor sistema eléctrico de Estados Unidos, un 76% en solo un año, enciende las alarmas sobre los límites de una red que no logra seguir el ritmo frenético de la demanda impulsada por los centros de datos de inteligencia artificial.

Detrás de este repunte histórico no está solo el crecimiento exponencial de la IA, sino la combinación de factores como la lentar modernización de infraestructuras, los retrasos en permisos para nuevas líneas de transmisión y la dependencia de fuentes energéticas volátiles. Para los hispanohablantes, especialmente en sectores industriales o residenciales con facturas energéticas en dólares, esto se traduce en un aumento directo del coste de vida y de la competitividad empresarial, mientras las autoridades debaten soluciones entre recortes de demanda o inversiones millonarias que tardarán años en materializarse.