Structural Power Crunch - AI Meets the Grid

Liquidity & Infrastructure Macro
PJM Interconnection’s emergency proposal to fast-track 15 GW of new power capacity is one of the clearest signals yet that the AI data-center boom is overwhelming legacy grid planning. With U.S. data-center power demand already at 44.6 GW and hitting record monthly additions, the Mid-Atlantic grid operator is bypassing the broken interconnection queue via bilateral matching between hyperscalers and new generation plants from September 2026 onward.

This isn’t incremental growth — it’s a structural regime shift. The old slow permitting process simply cannot keep pace with AI-driven electricity demand. For liquidity and positioning, this points to sustained upward pressure on power prices, transmission bottlenecks, and attractive setups for utilities, flexible gas assets, nuclear restarts, and anyone capable of delivering actual electrons on compressed timelines.

We’ve witnessed similar infrastructure catch-up cycles. The winners are rarely the pure-play hyperscalers alone; they are the enablers who solve the physical constraint. Expect this theme to dominate real-asset and energy positioning well into 2027 and beyond.