By Spencer Borison, President of the Americas, Codibly
At this summer’s PLMA Symposium in Arlington, the EV Grid Impacts track and the Data Center XChange track ran side by side in the same hotel. They opened together, in a shared plenary titled “Flexible Load at Scale: What EVs Taught Us and What Data Centers Will Challenge.” I spent the two days moving between both rooms, and I left convinced that the joint opening was the most important design choice of the event.
Most of our industry still treats these as two separate problems. They have separate teams, separate programs, separate vendor lists, and usually separate conferences. Data centers are the load nobody planned for, arriving faster than interconnection queues can clear. EVs are the load we have spent years learning to manage, one program at a time, through a lot of trial and error.
From the grid’s side, they ask the same question. A large, growing, partly controllable load shows up before the wires are ready. Can the utility connect it sooner if the load agrees to flex? And can the utility trust that the load will actually flex when it counts? Also as we heard a lot of discussion during the how, utilities must balance the immediate grid strain caused by the boom in data centers with the continuous growth from EVs, which will surpass data centers for total load on the grid in the US by around 2040.
Where the two loads already meet
The overlap is not only a matter of concept. Charging hubs for trucks, buses, and delivery fleets now ask for several megawatts of service. In some regions they are waiting on the same substations as data center requests. A distribution planner does not have one pool of capacity for EVs and another for data centers. Every megawatt promised to one is a megawatt the other cannot have, unless both can flex.
The central idea of the data center track was using load flexibility to accelerate energization: connect the site sooner, in exchange for a firm commitment to reduce load when the grid needs it. Charging depots have practiced a version of this for years. A depot connects with less capacity than its chargers could draw at once, and software keeps the total below the limit. The scale is different. The logic is the same.
What managed charging has already taught us
Four lessons from EV programs carry over directly.
A commitment is only as good as its proof. Programs that could not produce settlement-grade interval data lost time disputing baselines instead of growing enrollment. A data center that trades flexibility for an earlier connection will face the same test, with far more money attached.
The limit must hold automatically, in the worst hour. A managed charging program cannot depend on someone answering the phone during a heat event. Neither can a flexible interconnection agreement. The control has to act on its own, and it has to fail safe when a signal is lost.
The customer’s core need comes first. Drivers override charging when the car must be ready in the morning. Data center operators will protect compute and uptime in the same way. Flexibility designed around the customer’s priority lasts. Flexibility that works against it gets opted out.
Many vendors, one program. Managed charging runs across many vehicle and charger platforms, and most of the effort goes into making them work together. A data center has its own mix of systems for cooling, backup power, storage, and workloads. Once again, the integration work is where timelines will slip.
One capability, not two stacks
For utilities and program operators, the practical conclusion is to build one flexibility capability, not two. That means one way to set a limit for a site, send it, confirm the site is holding it, and settle the result. It should work for a fleet of home chargers and for a hundred-megawatt campus.
The standards to carry these signals already exist and keep maturing. OpenADR and OCPP both give utilities and site operators a way to send a limit to a site. What is usually missing is the software layer that turns a connection agreement into enforced, measured, auditable behavior at each site. Building that layer twice, once for EVs and once for data centers, doubles the cost and slows the connection of load that is already waiting.
PLMA was right to put both tracks under one roof. I would like to see more utilities put both programs on one platform.
If you are working on a flexible connection for a fleet depot, a charging hub, or a data center, I would like to compare notes on how you set the site limit, enforce it, and prove it held. You can find me on LinkedIn.
About Codibly
Codibly is a technology partner to utilities, aggregators, and energy technology companies building platforms for distributed energy resources, EV charging, and grid-edge flexibility. The company is a SunSpec Alliance Authorized Test Lab and an OpenADR Alliance Implementation Partner, and has delivered DERMS, VPP, charging, and grid-integration platforms across North America, Europe, and Australia. codibly.com