From: Utility Dive
Written by: Paige Knutsen & Erin Kempster
Electricity prices surged nearly twice as fast as inflation in 2025, making energy costs a political issue for voters. At the same time, power demand from AI and data centers is expanding at an unprecedented pace. Industry projections suggest these facilities could account for nearly 9% of U.S. electricity usage by 2030 — roughly double their current share.
The traditional instinct to build our way out of rising demand faces a gauntlet of challenges: long lead times, higher financing costs, siting and community pushback, and uncertainty around where new load will actually materialize. The December 2025 PJM Interconnection capacity auction — which maxed out the FERC-approved price cap while still falling short of the preferred 20% reserve margin — served as a warning of the risks ahead: a projected $70 monthly bill increase per household by 2028 or potential rolling blackouts. Conversely, overbuilt or misallocated projects can leave ratepayers responsible for costs that persist for decades, often leaving utilities with fixed generation assets geographically divorced from load centers.
Prioritizing demand side management as the first resource before committing billions to new infrastructure mitigates many of these risks. To be clear, this is not a call for excluding new infrastructure, but a recommendation for sequencing. Efficiency-first means prioritizing and exhausting cost-effective demand-side capacity before committing to the multi-billion-dollar, long-lived supply or transmission projects that define traditional grid expansion. While some grid expansion is inevitable for AI loads, these investments should occur only after the potential for cost-effective demand management has been fully accounted for. Compared to supply-side projects, RMI posits that demand-side measures can increase grid capacity at roughly half the cost and five to 10 times the speed.
Over nearly two decades of flat load growth, we have had the luxury of adding capacity at a measured pace. Now, as hyperscalers prioritize “speed-to-power,” fossil-fired stationary generation is the most familiar solution, and regulatory structures still favor traditional capital projects. Yet, the industry is moving toward more agile solutions.
A landmark 2026 framework report by the Alliance to Save Energy and The Ad Hoc group, “Bridging the Load Gap,” introduces the Bring Your Own Distributed Capacity (BYODC) model. Sponsored by Google, this framework allows large-load customers to fund strategic demand-side investments on the distribution network in return for capacity credits. By pivoting to BYODC, hyperscalers can secure the reliability and social license they need while delivering direct benefits to local customers.