Arizona Public Service Rate Case Opens May 18 With 45 Percent Data Center Hike Sitting Beside a 6 to 8 Percent Commercial Increase

Eight weeks of evidentiary hearings begin May 18 at the Arizona Corporation Commission on Arizona Public Service’s request for a 14 percent average rate increase. The proposed allocation does not distribute the burden evenly. Residential customers face a 16 percent hike. Small, medium, and large commercial customers face 6 to 8 percent. Data centers face 45 percent.

The numbers are the most explicit acknowledgement to date by a regulated utility that hyperscale load is a separate cost-of-service problem from the commercial rate class that has historically subsidized it.

The filing. APS filed the rate case in October 2025 seeking recovery on $300 billion of system investments and capacity costs tied to load growth that the utility attributes substantially to data center interconnection in the Phoenix metro. In a separate request filed March 4, 2026, the utility asked the Commission to move from periodic rate cases to annual rate adjustments, citing the pace of capital deployment.

The 45 percent figure for data centers is not a special tariff but a reallocation of cost responsibility within the existing rate structure. APS has argued that historic rate design assigned a share of generation, transmission, and capacity costs to the data center class that no longer matches the cost the class imposes. The 6 to 8 percent commercial increase reflects what remains after that reallocation, plus the underlying revenue requirement growth.

The Texas precedent. The structural argument is the same one the Public Utility Commission of Texas accepted in its El Paso Electric order on April 26, when it eliminated commercial-to-residential cross-subsidies and raised C&I demand charges to the actual cost of service. Arizona’s proceeding goes further by carving out hyperscale load as a distinct cost driver inside the existing C&I population.

If the Commission accepts the APS allocation, two regulated utilities operating in different ISO footprints will have moved within six weeks to restructure commercial rate design around explicit cost causation. That is not yet a national pattern. It is a directional signal from two states with the highest concentrations of new commercial load in the country.

The compounding problem. Phoenix-metro commercial customers on APS Schedule E-32 and E-34 already face demand charges that can exceed 50 percent of a monthly bill. A 6 to 8 percent across-the-board increase to those schedules raises the demand-charge component proportionally. For a 500 kW peak commercial customer paying roughly $7,000 per month in demand charges, the increase translates to $5,000 to $6,500 in additional annual demand-charge expense before any change to underlying peak coincidence.

The compounding matters because it stacks on top of an APS request that did not exist a year ago: annual rate adjustments. If the Commission grants the procedural request alongside the rate case, commercial customers are not facing a one-time 6 to 8 percent step but a recurring annual adjustment indexed to the utility’s capital plan. Capital plans for utilities serving data center growth corridors are growing faster than the underlying customer base.

The cost-shift argument. APS has framed the 45 percent data center increase as a defensive measure against cost-shift onto residential and small commercial customers. The framing is regulatory armor against the rising chorus of state-level legislation, including the Maryland RELIEF Act signed April 14, that bars utilities from socializing data center grid costs onto the broader rate base.

Whether the Commission accepts the framing matters less than the precedent the request creates. Once a utility files for a 45 percent class-specific increase against a 14 percent system average, the analytical method for separating large-load cost responsibility is in the record. Other utilities serving similar load growth in Virginia, Ohio, and Georgia will cite the methodology in their own filings.

The commercial position. A commercial customer on Schedule E-32 with peak demand in the 200 kW to 1 MW range sits at the meeting point of three forces. Rate-base inflation pushes the volumetric and demand components of the bill up by 6 to 8 percent in 2026 and potentially again in 2027 if annual adjustments are granted. Data center load growth, the underlying driver, continues regardless of whether the 45 percent class allocation is approved as filed. And the customer’s peak coincidence with system peak determines how much of the increase actually reaches the bill.

The peak shaving math for that customer changes sharply. A storage system sized to reduce monthly peak demand by 30 percent against an APS demand charge of $19 to $22 per kW returns roughly $68,000 to $79,000 per year on a 1 MW peak. Apply the proposed 6 to 8 percent demand-charge increase and the same kilowatt of avoided peak returns approximately $4,000 to $6,000 more annually. Apply a second adjustment in 2027 and the curve steepens.

The procedural calendar. Eight weeks of evidentiary hearings means a final order is unlikely before late summer, with new rates effective late 2026 or early 2027. The intervening period is the window during which commercial customers either lock in current demand-charge exposure or hedge it. Storage projects placed in service before the new rates take effect compound the value of the hedge.

That timing pressure exists in parallel with the OBBBA-settled 30 percent ITC plus the 10 percent domestic-content adder. A commercial project approved in summer 2026 and placed in service before the rate adjustment captures both the federal tax benefit and the pre-increase rate structure as the baseline against which storage savings are measured.

The signal. The APS filing is the second utility request in 2026 to explicitly separate hyperscale load cost responsibility from the broader C&I class. The first was El Paso Electric. The third is likely to come from a Mid-Atlantic utility with PJM-zone data center exposure inside the calendar year.

The pattern is not that data centers will pay more. The pattern is that commercial rate classes are being progressively unbundled from the load growth that drove their rate increases through 2025. The unbundling shifts the analytical question on commercial demand charges from “what is the average increase” to “what does this customer’s coincident peak actually cost the system.” Rate design built on that question rewards customers who can flatten their own peak.

Eight weeks of evidentiary hearings open in 15 days.

Sources