Stretching from Washington D.C. clear to Chicago, PJM is the largest electric power market in the U.S. and the epicenter of a nationwide struggle. For the last several months, the grid operator has worked overtime to solve two intertwined crises.
The first, highlighted in recent statements by Pennsylvania Governor Joshua Shapiro and New Jersey Governor-elect Mikie Sherrill, is the painfully high cost of the power PJM delivers to nearly 70 million customers. The second is the grid operator’s seeming inability to accommodate a tidal wave of demand from new large energy users, particularly data centers powering artificial intelligence.
While data center growth is a flashpoint, PJM’s sharply higher prices and reliability challenges are also caused by capacity-market mechanics colliding with a system that’s moved from surplus toward shortage. Because of its size and role in powering the AI boom, how PJM approaches the dual issues of demand and market design to address the affordability and reliability of electricity could influence policy beyond its borders, potentially prompting other markets and regions to follow its lead.
But several of the current proposals that PJM is weighing threaten to pinch off one of the best tools in our arsenal for lowering the cost of power and rapidly maximizing the capacity of the grid: demand response. That would be an enormous loss, given that the tool has proven to be instrumental in maintaining system reliability and preventing deeper emergency actions by PJM during peak events.
We should be encouraging and harnessing more demand response through virtual power plants to deliver reliable service and affordable bills for all energy users, big and small — plus speedy power to AI data centers and other large loads that are critical to economic growth and global competitiveness.
These tools, which can take decades to play and build out the old-fashioned way, call on a diverse array of grid end users — such as industrial plants, hospital and college campuses, large batteries, and millions of smart devices — to reduce their energy use when the grid is most stressed. There are roughly 8 GW of demand response in PJM, but if implemented poorly, the ongoing policymaking effort could drive away much of that capacity, which is vital for closing the widening gap between demand and supply.
As it stands now, new large loads are coming online faster than new supply. This means that commercial and industrial demand response customers are likely to be dispatched much more frequently, and manual load shedding is likely to become more common. In time, this could discourage the largest customers from participating in demand response at all.
Losing that participation would be costly and unnecessary, just when even some data center operators are beginning to show what “real” flexibility can look like. Google, for example, has expanded efforts to temporarily reduce data center electricity use by shifting or curtailing certain computing workloads during periods of grid stress.
To effectively build policy around flexible load, PJM needs to answer: What, exactly, can the grid operator require of a customer in real time, and how is performance verified?
How to meet the moment
There are four keys to PJM getting out of this dilemma and delivering plentiful, affordable and reliable power for all.
First, large loads that don’t commit enough flexible capacity to offset their needs at peak demand times need to be the first customers off the system when there is an impending shortage of electricity — before demand response customers must curtail their usage. These commitments can include the customer’s direct participation in demand response, as well as securing commitments from other large load customers to provide specific amounts of grid relief when needed through private VPPs.
That’s basic fairness and reliability: Customers who have made firm commitments to provide flexibility shouldn’t be penalized ahead of new entrants that haven’t supplied comparable system value. If PJM doesn’t establish this clear curtailment hierarchy, it risks weakening the very resources it depends on most during emergencies.
Next, if large loads participate in demand response, those customers should be told to use less electricity at the same time as other demand response participants rather than after, as some large loads have proposed via a “go last” model. Dispatch has to be synchronized, so the grid gets relief when it’s needed most, not after system operators have already burned through reserves or escalated to emergency actions.
By contrast, a “go last” model effectively asks existing demand response providers to shoulder the early, high-pressure response while giving new large loads the benefits of participation without the same operational responsibility.
Also, if there is a requirement to bring one’s own capacity, any capacity should be eligible, be it from generation elsewhere in a service territory or demand response. Sometimes called “private VPPs,” this flexible capacity solution would allow an inflexible data center to essentially purchase flexibility from other market participants. In developing a private VPP, the large load or an aggregator acting on its behalf would secure accredited capacity and confirm its demand-response participation.
Finally, to the extent new demand response products are created, they should be open to all loads on a comparable basis.
Ultimately, we need more education and incentives for customers, like the ability for large loads to buy flexibility from other demand response or price-responsive demand providers, to increase the number of large energy users that participate in demand response.
Only by spreading grid relief across more sites and avoiding customer fatigue will we continue reaping the benefits of demand response and ensure that PJM gets every megawatt of supply it needs.
Michael Smith is CEO of CPower, a leading virtual power plant platform with 6.7 GW of customer capacity at more than 23,000 sites. The opinions represented in this contributed article are solely those of the author, and do not reflect the views of Latitude Media or any of its staff.
