What Is Blocking U.S. Power Expansion for AI—And What Could Unlock It by 2030?

Why It Matters

Unlocking this capacity is critical for AI data centers and broader economic growth, while providing utilities and investors clear pathways to capture new revenue streams.

Key Takeaways

• Interconnection reforms could add 65–130 GW, largest potential gain
• Permitting streamlines could free up 16–54 GW of capacity
• Grid‑enhancing technologies may contribute up to 7 GW
• Supplemental generation reforms could yield 11–106 GW
• Coordinated action across four clusters could unlock 92–297 GW

Pulse Analysis

The rapid expansion of artificial‑intelligence workloads is turning U.S. data centers into some of the largest single electricity consumers. While utilities have long‑term generation plans, the pace at which new capacity can be brought online is throttled by a web of regulatory and operational bottlenecks. RAND’s 2026 research brief quantifies those constraints, isolating four clusters—permitting, interconnection, underutilized transmission, and supplemental generation—that together could limit or enable an additional 92 to 297 gigawatts of reliable power by 2030. Understanding these levers is essential for any stakeholder betting on AI‑driven growth.

Interconnection emerges as the most lucrative lever, with streamlined studies and surplus‑interconnection pathways projected to unlock 65–130 GW—roughly half of the upper‑range scenario. Permitting reforms, such as programmatic environmental reviews on federal lands, could contribute another 16–54 GW, while incentives for grid‑enhancing technologies (dynamic line ratings, power‑flow control) add up to 7 GW. The least certain but still significant source is supplemental generation; relaxing backup‑generation rules and encouraging on‑site clean resources could free 11–106 GW, especially for data‑center clusters that already maintain standby power.

Policymakers face a narrow window: coordinated reforms across all four clusters must commence within the next few years to meet the 2030 target. Federal agencies—DOE, FERC, and the interior department—can drive uniform generator‑replacement rules, publish real‑time transmission capacity maps, and create incentive structures for grid‑enhancing pilots. For investors, the upside is clear: regions that clear interconnection backlogs and unlock supplemental generation will attract AI‑focused data‑center projects, translating into higher asset valuations and new revenue streams for utilities willing to modernize. The study underscores that piecemeal fixes will fall short; only a holistic approach can sustain AI‑driven electricity demand.