AI's Next Phase: Shifting Focus from Chips to Power Supply

As AI expands, the bottleneck shifts from chip production to power supply. Energy companies like Brookfield Renewable, NextEra Energy, and Bloom Energy are poised to capitalize on massive electricity demand from AI infrastructure.

The AI infrastructure bottleneck is shifting from chip supply to electrical power, with data centers projected to consume over 90 TWh of electricity annually by 2026 . Morgan Stanley forecasts U.S. data center power demand could reach 74 GW by 2028, against a projected shortfall of roughly 49 GW in available capacity. Hyperscalers including Microsoft, Google, Amazon, and Meta are collectively on track to spend over $600 billion on AI infrastructure this year.

Energy companies like Brookfield Renewable (BEP), NEE (NextEra Energy), and Bloom Energy (BE) are positioned to benefit from this structural shift . Interconnection backlogs, power transformer shortages, and construction timeline mismatches mean capital alone cannot resolve supply constraints in the near term. On-site power generation, advanced cooling systems, and renewable energy digital twins are emerging as critical solutions.

The theme was central to NVIDIA GTC 2026, where energy efficiency and power infrastructure were featured alongside chip innovation. For investors, this represents a broadening of the AI investment thesis beyond semiconductor names into utilities, renewable energy, and industrial infrastructure. The companies that can deliver reliable, scalable power to data centers may capture outsized returns as AI training and inference workloads continue to scale.