The Midcontinent Independent System Operator released its first comprehensive long-term load forecast on April 13, projecting that peak demand across its 15-state region will grow from 121 gigawatts in 2025 to 163 gigawatts by 2035. That is a 35% increase in ten years. Data center development drives almost all of it. Under MISO’s current trajectory, data centers will account for a fifth of all MISO electricity by 2030. By 2040, that share reaches a quarter.
The forecast marks, specifically, a significant departure from MISO’s historical load trajectory. Between 2009 and 2024, peak load across the MISO region remained essentially flat. The 2026 forecast projects a compound annual growth rate of 2% through 2046. That compares to the 0.5% rate that defined the prior 15-year period. MISO expects, moreover, 8 to 14 gigawatts of data centers to come online across its territory in 2026 and 2027 alone. It describes that range as a key near-term indicator of how quickly AI-driven demand will materialise.
The Midwest as the Next Major AI Infrastructure Market
That geographic concentration reflects several factors. Available land, supportive utility relationships, and grid interconnection timelines shorter than those in saturated markets like northern Virginia all favour the Midwest. MISO noted that much of the projected demand will come from high-confidence projects. It defines those as facilities with interconnection agreements in place, regulatory awareness, and active construction underway.
Data center load has, however, a fundamentally different demand profile from the residential and commercial load that has historically dominated MISO’s planning horizon. Hyperscale facilities operate at roughly 90% load factors. They draw near-constant power around the clock regardless of season or time of day. That flat demand profile consequently raises MISO’s system load factor and changes the economics of generation investment across the region. As we have covered in our analysis of how AI infrastructure is permanently restructuring the US power sector, the arrival of flat, high-density industrial load at this scale changes the fundamental economics of the regional grid in ways that conventional utility planning models were not designed to accommodate.
The Uncertainty MISO Could Not Resolve
MISO was, however, direct about the limits of its forecast. It cautioned that future data center demand is highly dependent on the AI industry finding routes to profitability. Rapidly rising announcements and project submissions, moreover, widen uncertainty around both volume and realisation timelines. The 8 to 14 gigawatt range for 2026 to 2027 data center additions reflects that uncertainty directly. Even at the low end of that range, the load additions would, notably, represent the largest two-year demand increase in MISO’s history.
That uncertainty has, in turn, direct planning implications for utilities operating within MISO’s footprint. They are, consequently, being asked to commit to generation and transmission investments sized against a load forecast that may prove materially wrong in either direction. As we have covered in our analysis of the demand response opportunity AI data centers keep refusing to take, the gap between announced data center capacity and actually energised capacity is one of the most consequential uncertainties facing grid planners in every major US market. MISO’s forecast makes that uncertainty explicit at a regional scale for the first time.
