Michael Chua
PowerShift Contributor
First Published Q2 2026
For most of the last two decades, U.S. electricity demand barely moved. Utility forecasters got used to projecting half a percent of annual growth and going home. That world no longer exists.
Hyperscale data centers now routinely request 300 to 500 MW per site, and some proposals exceed 1 GW (Myszka, 2026). EVs and building electrification are piling on. Meanwhile, regulators want utilities to integrate more distributed energy resources (DERs), hold the line on customer bills, and keep the grid reliable… all at once. The concept of “flexibility” gets thrown around at every industry conference these days, but what does it mean when you’re the one trying to deliver it?
I put that question to two people at GridX who spend their working hours in the middle of it.
“You’re explaining variances to Wall Street and investors,” Pirro said. “That can drive down credit rating, share price, the whole thing. It’s essential that the load forecast is spot on. And, I think that’s become more challenging due to all the moving pieces recently, because you have so much more growth that you’re actually trying to project.”
The Shape of the Problem
Michael Pirro spent 35 years inside investor-owned utilities, like Niagara Mohawk, National Grid, PG&E, and Duke, before joining GridX as Senior Director of Regulatory and Rate Design. As recently as a few years ago, his load-forecasting team at Duke would layer in four or five “special projects” a year, each at 10 to 20 MW. “In my last two forecasts, we had 95-plus projects below the line that we had to layer in,” he said. Same team. Same spreadsheets. Roughly twenty times the work.
That kind of step change does two things to a utility. It creates excitement, with growth back after decades of flat demand, and it creates anxiety, because nobody is sure how much of that demand will actually materialize. A customer that requests 800 MW might end up at 300 MW. If you’ve planned, financed, and built around the 800, the gap shows up in the next quarterly call. “You’re explaining variances to Wall Street and investors,” Pirro said. “That can drive down credit rating, share price, the whole thing. It’s essential that the load forecast is spot on. And, I think that’s become more challenging due to all the moving pieces recently, because you have so much more growth that you’re actually trying to project.”
Regulators have caught on. The standard approach used to be a single forecast and a capital plan to match. Now commissions want scenarios – e.g., low, medium, and high – with explicit ramp schedules and exit points. As Joe Chang, Senior Director of Product at GridX, put it: “Planning, ratemaking, and stakeholder conversations used to happen sequentially. They now need to happen together.”
Most utilities still do this work in Excel. New technologies could help them adapt more quickly.
Who Pays?
The other thing regulators have caught on to is that somebody has to pay for all this new infrastructure, and ordinary ratepayers (probably) shouldn’t be the ones to do it. Affordability has become the dominant talking point at every utility conference Pirro attends. “Maybe to the point that people don’t even know what they mean when they say, ‘affordability.’”
Virginia produced the clearest test case. In November 2025, the Virginia State Corporation Commission approved Dominion Energy’s new GS-5 rate class for customers demanding 25 MW or more. This effectively established a dedicated tariff for data centers and other very large loads. Starting January 1, 2027, those customers will be required to pay a minimum of 85% of contracted distribution and transmission demand, and 60% of generation demand, with 14-year contract commitments attached.
And, Virginia doesn’t stand alone. The Smart Electric Power Alliance counts 77 large-load tariffs pending or in place across 36 states, with 29 approved in 2025 alone, more than were approved in the prior seven years combined. AEP Ohio, which adopted one of the more aggressive structures, cut its large-load forecast by more than half after the tariff took effect, suggesting that speculative requests evaporate quickly when a deposit is required (Lutz, 2026).
The execution problem is harder than the policy debate suggests. Each of these large-customer deals is essentially a custom contract – unique demand commitments, special charges, exit fees, and take-or-pay provisions. Standard billing systems weren’t built for that, and most weren’t built recently.
The Case for Dynamic Pricing
Chang argues that dynamic pricing is the most useful near-term bridge between integrating distributed resources and keeping the grid reliable – rates that reflect actual hourly grid conditions rather than averaging them across a season.
In the U.S., California’s CalFUSE framework (California Flexible Unified Signal for Energy) represents the most developed example. The common problem with hourly pricing: customers run from anything they can’t easily understand. Without a clear, simple visualization of how dynamic pricing might help, customers tend to stick with the old pricing they comprehend. Enter the ‘shadow bill,’ CalFUSE’s answer to this habitual hesitation.
With a shadow bill, customers stay on their existing rate, paying as they usually would; but, they also receive a parallel calculation showing what they would have paid under dynamic pricing. If switching plans would have saved them money under the new rate plan, customers would receive those savings via a credit back. “That removes the bill risk that usually slows adoption of complex new rates,” Chang said. “Utilities and customers can learn alongside each other before committing.”
GridX is working on three components: the underlying price-signal calculations, the shadow bills, and the customer-engagement layer that translates an hourly price into something a thermostat or EV charger can act on. None is easy on its own. But the core idea is simple. “Batteries, smart thermostats, EVs, and the customers behind them only respond to prices they can see clearly,” Chang said.
California isn’t satisfied with pilots – the state’s Load Management Standards will require the large IOUs and CCAs to offer dynamic hourly rates for all customer classes by 2027 (CEC, 2023). “Picture a grid where thousands of customers and their devices get a clear ‘demand is tight right now’ signal, respond through real economic incentives, and see a bill afterward that confirms what happened,” Chang depicts. “That’s a fundamentally different system than the one we operate today.”
“Picture a grid where thousands of customers and their devices get a clear ‘demand is tight right now’ signal, respond through real economic incentives, and see a bill afterward that confirms what happened. That’s a fundamentally different system than the one we operate today.”
The Bill is the Policy
Both men keep returning to the same idea: customers experience policy through their bills. If the bill doesn’t make sense, the policy doesn’t either. In practice, this tenet proves difficult for utilities to face, as most contend with information systems built decades ago, when residential rates had two periods and large-customer billing was simple and straightforward.
“Before GridX, there was no such thing as a sandbox in the billing system,” Pirro said. “New tariffs get designed in a rate case, then handed to the billing team months later, at which point, someone discovers the existing system can’t produce the bill the policy requires.” He sees that gap as the most underrated constraint on the industry’s policy ambitions over the next decade.
Internal silos make it worse. Load forecasting, rate design, customer outreach, and billing operate on different cycles and rarely talk until late. Asked whether the bigger barrier to innovation is regulation or something inside utilities themselves, Pirro didn’t hesitate: “internal silos.”
The constraint becomes palpable on the customer side. Pirro is blunt about what consumers should know. “Husband and wife are working. They come home, want to turn the lights on, cook dinner, and run the kids to softball and dance class. They don’t want to worry about whether they can turn the washer on now.” Automation handles part of it. Clear messaging handles the rest. “In ten seconds, you either have them or you don’t.”
Two Paths Emerge
Looking ahead five years, two theoretical, albeit not equally probable, paths emerge.
On the route where things go well, performance-based regulation continues to spread… a potential future even now hinted at by the 17 states and DC, which have already established some form of it, with another dozen currently exploring (Downard, 2025). In this world, large-load tariffs harden into something predictable, hyperscale customers trade flexibility commitments for faster interconnection, and dynamic pricing graduates from California pilots into something other states can borrow. Utilities rebuild billing systems so customers can see what’s happening on the grid through their bills and even act on it.
On the other track, adoption splinters state by state. Some utilities lean in, while others stay on cost-of-service models, producing increasingly risky decisions based on outdated rules – rooted in rewarding capital investment and cost recovery – as customers, regulators, and policy makers move toward flexibility, electrification, decarbonization, and affordability. Ratepayer backlash spikes due to unfair distribution of costs, decarbonization stalls, and utilities watch as potential competitors emerge, getting rewarded for aligning their incentives with customers and society.
Neither future is locked in. Chang and Pirro both seem to believe – though neither put it this way – that the components and technological necessities to cement the first reality already exist. What is missing is the connective tissue: between the rate case and the bill, between the forecast and the capital plan, between the price signal and the device that’s supposed to respond to it. That’s where their work matters today.
Receive the Magazine today.
