PG&E’s flexible grid connection program is starting to pick up momentum — and the utility now has its sights set on standardizing it for all new loads facing long wait times for grid upgrades.
It’s been ten months since the first “Flex Connect” site came online in California. At the end of last year there were just two live sites; today there are five. In the first few quarters of 2025, PG&E conducted around 35 site analyses per quarter, and is on track to have 10 sites online by the end of the year.
The pilot program is designed as a workaround to distribution grid congestion, for load that can be flexible during peak periods, such as electric vehicle charging hubs. It’s been in the works for well over a year now, at first somewhat quietly, as the utility has worked to iron out the technical challenges that come with offering capacity on an as-available basis.
Now, though, most of those challenges have been solved. The sites that are online are working as expected, including responding to forecasted heat events already in 2025, Alex Collins, a senior manager overseeing the Flex Connect program, told Latitude Media. For most sites, he added, their energy use isn’t even being curtailed during heat events when PG&E caps the capacity the site can use.
“Really what Flex Connect is doing is it’s reducing the maximum amount a site can use,” Collins explained. “Even if you lower that ceiling, it will likely have no impact on your business…unless you’re in that absolute peak loading condition, which most of our customers are not.”
The utility’s current data set is somewhat limited because only a handful of sites are live, but Collins said the sites in the pipeline are expected to come online an average of a year and a half earlier than they would have via the traditional interconnection process. Pepsi, for example, used Flex Connect to expand charging hours for its fleet of electric semitrucks at a bottling facility in Fresno. Now the fleet can access as much as 4.5 megawatts most of the time — including during previously restricted daytime hours — without the 18-month delay that would have been caused by waiting for long-term capacity upgrades.
The majority (approximately two-thirds) of the projects in the Flex Connect pipeline continue to be EV chargers for both public and commercial fleets, as well as on-grid batteries participating in the wholesale market.
The goal for 2026, Collins said, is making Flex Connect a standard offering for certain types of load seeking grid connection but facing long wait times.
But PG&E isn’t there quite yet. In fact, the next several weeks are critical for proving out the concept and the tech: August and September evenings, between 4 p.m. and 9 p.m., are when PG&E expects to call on their Flex Connect sites the most.
“We want a full year of field data just to verify that all the operational fail safes work as intended and our operational processes work as intended” during those peak times, Collins explained. “We want to make sure that we are meeting or exceeding those forecasts, otherwise a customer could be in a bad spot where they built something and the value doesn’t come at the end.”
But so far this year the utility has been able to deliver more flexibility and additional capacity than initially predicted via the pilot, Collins said: “If that continues to be the case, which we expect it to, then we can have more confidence in using that forecasting methodology to give customers the best information so they can make a decision to opt into the program.”
Proof for planning
Collecting that additional data is largely about proving the program’s effectiveness to PG&E’s distribution planners than demonstrating its value to customers.
Distribution engineers currently want to see real-world performance data that confirm that the key technologies — like flexible energization controls, improved load forecasting, and on-the-ground monitoring of equipment capacity — are working as intended before making site flexibility a standard part of the distribution planning and energization process.
It’s a significant departure from the traditional approach, which Collins described as “just a thumbs-up-thumbs-down with a timeline.”
“We’re also building the tooling that allows that team to do this as part of their standard work,” he added, including new internal technology and types of analysis that can forecast load over a typical year and feed that insight into upgrade decisions.
PG&E planners have been “supportive” of the Flex Connect effort so far, Collins said, and in some cases have brought projects to the pilot team for review. Those include cases in which the planners aren’t confident about the projected load, but models are nudging them toward requiring new load to wait for upgrades to avoid overloading the equipment.
That cooperative approach means PG&E can gather real-world data on how much load the site actually draws and how much it impacts the grid, then feed that information back to the planners, who can make more accurate and timely decisions about whether, when, and how much to upgrade.
The challenge of pitching flexibility
But standardizing Flex Connect isn’t just a technical or grid-planning challenge; it’s also a marketing one.
For most customers, the idea of a flexible service connection is still unfamiliar, and opting into the pilot program involves making choices about both technology and integration: Does a site want to build their own in-house controls, adopt a solution from a pre-approved vendor, or pursue fully custom integration to meet needs like market bidding and dispatch?
Within Flex Connect’s existing customer base, capabilities vary widely, even between similar categories of load, like EV charging, for example. Some EV sites, which Collins described as a “perfect fit” for Flex Connect, use a local site controller that manages all chargers under a single power limit for the entire site. On the other end of the spectrum are sites where each charger connects individually to a cloud-based system, which typically requires a more customized approach to make Flex Connect work.
Then there’s a small class of customers that are even more complex: front-of-meter battery developers who participate in the wholesale market. Those projects come with a different set of demands, because the controls must integrate not only with PG&E’s distributed energy resource management system, but also with the CAISO market systems they use for bidding and dispatch. Those large, multimillion-dollar projects almost always opt for fully custom integrations, Collins said.
All that means that the early onboarding process can be a heavy lift on the PG&E side, Collins said. To mitigate that, the utility is preparing simple, plain-language marketing materials that will illustrate what qualifies as a “simple” versus “complex” site, outline the trade-offs between custom and ready-made solutions, and set clear expectations about likely cost and scheduling.
The Flex Connect team is also hoping that their field data from live sites — which they plan to publish after the August and September peak data is in — will help convince more customers to opt in. That can be tricky, given that customers in the pilot must also cover the costs associated with installing the necessary load control technology that enables communication with PG&E’s DERMS.
PG&E is hoping to counter that risk, Collins said, with performance data showing that sites are getting their desired capacity the vast majority of the time, even during stress events like heat waves, and that the limits imposed by Flex Connect won’t disrupt daily operations.
Field results like that, coupled with easier, cheaper ways to flexibly plug in, will be crucial to persuading more developers to sign on, he added.
