A VPP Is a Garden, Not a Power Plant: Tend It Right and It Grows Itself

Utilities keep procuring virtual power plants as if they were peakers: buy the megawatts, then expect them to stay put. ev.energy's CEO Nick Woolley argues a VPP is closer to a garden. Tend it through acquisition, activation and retention, and it becomes the only capacity on a utility's books that grows on its own.

Load is growing again for the first time in two decades, driven by data centers, electrification and heat. The traditional answers, more generation and more network, both come with a queue. For projects reaching commercial operation in 2025, the median wait from interconnection request to switch-on was more than five years, and of the capacity that asked to connect between 2000 and 2020, only 13% was operating by the end of 2025 [2].

Flexibility at the grid edge has no such queue. It grows every time a household buys an EV, a battery, or solar panels, and its value is well evidenced. ev.energy research, with research from the Brattle Group, puts the avoided system cost of one actively managed EV at $145 to $575 a year from charging alone, and finds virtual power plants deliver power at up to 60% lower cost than traditional generators [1]. On a residential feeder, coordinating EVs, batteries, and solar can cut the modeled peak by 44% with no new wires [3], worth $5 million to $20 million of deferred reinforcement per constrained circuit in ev.energy's modeling [4]. And a VPP can enroll customers and dispatch devices within weeks.

So why isn't every utility sitting on hundreds of megawatts of flexibility? Woolley's answer is that a VPP is not a power plant, and the industry keeps procuring it as though it were. A peaker stays the size you bought, where you put it. A VPP is tens of thousands of devices owned by people who move house, sell cars, end leases, and unplug things without telling anyone. The megawatts on the contract are identical. The asset is completely different to run. You do not commission a garden. You cultivate it, and tended well, it does something a power plant never will: it spreads.

First, the case for default enrollment, and where it has to happen

The National Bureau of Economic Research recently randomized enrollment offers across all 12,174 eligible EV-owning households in a California community choice aggregator's territory. With no incentive, 1% enrolled. At $40 a month, roughly 15% of a typical bill, 4.6% enrolled, too few to detectably change the group's load, even though households could collect the money without switching the feature on [5].

The researchers conclude managed charging has to stop being opt-in, and Woolley agrees. But a default has to attach to a moment, and the obvious ones are installation and commissioning: the charger, battery, or solar system goes in enrolled, and the vehicle is connected at handover. Default enrollment with an honest opt-out is the largest opportunity in the industry, and ev.energy would back the regulatory work to get there.

Almost no jurisdiction has that machinery yet, and program managers have megawatt commitments to meet this year under opt-in rules. What works today comes down to three disciplines: acquisition, activation, and retention. Most programs fund the first, assume the second, and ignore the third. All three should be measured in dispatchable capacity, the kilowatts a program can actually call on, rather than enrolled nameplate, the flattering number Woolley has called an illusion.

Acquisition: a proposition and channel problem, not a market problem

The obvious reading of the California result is that customers do not want this. It is a fair first reading, but what was tested was just one channel carrying just one offer, and that always hits a ceiling. Two things break through it, which ev.energy calls experience (the proposition) and equity (the channel).

EV and DER owners are not one segment. A household with rooftop solar wants to charge on its own generation during the day, not be nudged to midnight. A household with a Tesla and a Rivian needs a program that works with both. Offering one product to all of them is like making only brown shoes and wondering why the people who wanted black ones never buy.

A well-run program with a single price-based offer can reach around 30% of eligible customers [9]; Con Edison's SmartCharge New York has enrolled more than 31,000 EVs, around 30% of the EV market in Con Edison's territory [8]. Each additional offer, such as whole-home orchestration, away-from-home charging, fleet accounts, or co-marketing with automakers and installers, adds roughly five percentage points. Stack three or four and 50% is reachable, with world-class probably topping out at 60% to 70% [9]. SEPA found regulators typically project around 80% enrollment while most programs reach under a third of target [6]. That gap closes with offers that solve diverse problems, not bigger checks to the same participants.

The channel means meeting people where they already are. Southern California Edison has run street teams across Los Angeles, Clean Power Alliance enrolled drivers on the spot at the LA Auto Show in three languages, and Con Edison and Orange & Rockland opened enrollment through the Tesla app [8]. None of it is free: realistic acquisition cost is close to $100 per EV driver across most of the US and up to $150 in California [9]. Against up to $575 a year in avoided cost [1], that pays back in months. A program budgeted at $25 per customer is not funded to scale.

Underneath every channel sits device coverage. Through vehicle integrations alone, ev.energy's Eve™ platform has modeled coverage of around 90% of US EV drivers, and approximately 95% through vehicle or charger combined [10]. Combining telematics, chargers, and metering data opens several routes into almost every household. Programs built on a narrow set of integrations exclude customers before marketing even starts.

Activation: enrollment is not capacity, and capacity is a ladder

A customer who signs up and never finishes verification is not a dispatchable asset, and this is the industry's largest self-inflicted loss. The worst-designed journeys, asking for account numbers from a paper bill, meter numbers, and exact service addresses, turn fewer than one in ten sign-ups into an active device. Programs that verify against the utility's own data activate the overwhelming majority [9].

The second half of activation gets far less attention. Enrollment counts customers but says nothing about what each one delivers, which is a capacity ladder: dispatchable kilowatts that depend on how a household participates. In ev.energy's illustrative operating ranges [9], manual response to time-of-use pricing delivers about 1 kW of peak reduction, automatically managed charging around 1.5 kW, and orchestration across charger, solar, battery, and thermostat 2 to 3 kW. Vehicle-to-home delivers 3 to 5 kW, and full vehicle-to-grid export delivers 5 to 7 kW.

Same enrollment number, five times the capacity between the bottom rung and the top. A 7 kW charger on a driveway yields perhaps 1 to 1.25 kW a program can count on [3], so where a customer sits on the ladder matters more than their hardware rating. The upper rungs are already real: Clean Power Alliance's program delivers 94% of its energy off-peak and 21% in daytime solar hours, and ANWB in the Netherlands has diverted more than 295 MWh of excess rooftop solar into EVs, with 75% of users reporting high satisfaction [7]. Results vary by geography and program design.

Retention: the discipline nobody budgets for

Retention is the most overlooked line in VPP business cases. In many ev.energy programs, net growth comes from existing users through referrals, added devices, and timely nudges when life events put participation at risk. That only works if users trust the experience. Across ev.energy's active portfolio, more than 96% of enrolled devices remain active [10].

Only around 1% to 2% of customers actively unenroll [9]. Devices are another story: leases end, cars are sold, chargers wear out, and integrations expire, so a mature program should plan for more than 30% device turnover a year. But a household that swaps one EV for another has an open seat, and whether it gets filled is a design decision. The California study found active unenrollment at just five households across the whole experiment [5]. People rarely quit these programs. Instead, their connection breaks and nobody notices.

That makes most lost capacity a maintenance problem with known answers. Around 80% of customers out of contact for 30 to 90 days come back with the right prompt, roughly 30% can still be won back after 90 days, and around 25% of explicit unenrollments return with well-designed win-back communications [9]. Even at 99% monthly retention, more than a tenth of capacity is lost each year, so the garden needs constant watering. A program that funds acquisition but not retention pays full price twice for capacity it already had. Until regulators address the bias toward capital over operating spend, the practical answer is to price retention into the capital cost of the capacity, not a marketing line that gets cut in year two.

Tend your VPP properly, and it grows itself

When all three disciplines work, the base starts adding capacity nobody bought, in three directions at once.

Wider, through referrals. Across ev.energy's programs, referrals already outpace churn: in one program, around 6% against around 4%. A referral only becomes capacity once the new household activates, and that is the number Woolley urges every program manager to measure. Until recently, ev.energy did not measure it either.

Denser, through device attachment. The Eve platform has connected more than 410,000 devices across more than 300,000 users [10]. Each extra device adds dispatchable kilowatts at no acquisition cost, plus another anchor. A car changes hands every few years; a rooftop array or a wall-mounted battery does not.

Deeper, through the ladder. A household that moves from time-of-use response at about 1 kW to full orchestration at 2 to 3 kW roughly triples its contribution, with no new customer and no new hardware.

Every other asset on a utility's balance sheet does the opposite. A peaker's nameplate is the best it will ever be on commissioning day. Flexibility being cheaper is the easy part of the argument. The harder and better claim is that a well-run virtual power plant is the only capacity a utility can own that appreciates.

The green thumb wins

Nameplate megawatts are a claim. Dispatchable megawatts sustained across years are a practice, and ev.energy is running that practice at scale today with Eve Programs.

The winner in this market will not be whoever writes the cleverest dispatch algorithm. It will be whoever has the best green thumb to tend a VPP garden of dispatchable capacity, while everyone else is still admiring their nameplate.

Is your VPP being procured like a power plant, or cultivated like a garden? Book a demo

Sources

1. The Brattle Group and ev.energy, The Utility Playbook for Turning EV Grid Risk into a $30 Billion Opportunity (The Value of Managed Charging), 2025: $145 to $575 of avoided cost per actively managed EV per year; virtual power plants deliver reliable power at costs up to 60% lower than traditional generators; an annual cost-avoidance opportunity of around $30bn by 2035, modeled.

2. Lawrence Berkeley National Laboratory with GridTracker, Queued Up: 2026 Edition, Characteristics of Power Plants Seeking Transmission Interconnection as of the End of 2025, published May 2026: approximately 8,200 projects actively seeking interconnection, representing 1,312 GW of generation and around 749 GW of storage; median duration from interconnection request to commercial operation of over five years for projects built in 2025; of capacity that submitted requests between 2000 and 2020, 13% had reached commercial operation by the end of 2025 and 75% had been withdrawn.

3. Nick Woolley, The GW Illusion: Not All Flexibility is Created Equal: 44% residential feeder peak reduction from coordinated EV, battery and solar management; EV diversity ratio of roughly 1:7, or 1 to 1.25 kW dispatchable per enrolled vehicle. Modeled figure, assumptions as published in the original article.

4. ev.energy analysis: approximately $1bn of transmission or generation investment deferred per 1,000 MW of peak reduction; $5m to $20m per constrained distribution circuit for a five to ten year deferral. ev.energy modeling rather than third-party published figures.

5. Fiona Burlig, James Bushnell and David Rapson, If You Build It, They May Not Come: Willingness to Participate in Managed EV Charging, National Bureau of Economic Research working paper, April 2026 (also Energy Institute at Haas Working Paper 362), conducted with Peninsula Clean Energy: 12,174 eligible EV-owning households randomized; 1.0% enrollment with no incentive, 4.6% at $40 a month, approximately 15% of the average monthly electricity bill; intent-to-treat effects on load indistinguishable from zero; households could collect the incentive without activating the feature; active disenrollment essentially non-existent, at five households.

6. Smart Electric Power Alliance, Managed Charging Programs: Maximizing Customer Satisfaction and Grid Benefits: regulators project that around 80% of eligible customers will enroll in a managed charging program, while most programs assessed reached less than a third of targeted enrollment; SEPA puts the full potential level of customer participation at 70% to 80% enrollment and participation, achievable with the right technology and partnerships.

7. ev.energy, Clean Power Alliance EV SmartCharge program results, published November 2025: 94% of energy delivered off-peak and 21% during daytime solar hours, figures as at November 2025. ev.energy and ANWB, Netherlands: over 295 MWh of excess rooftop solar diverted into EVs and 75% of users reporting high satisfaction with the feature, program-level figures as published.

8. ev.energy, SmartCharge New York program results (Con Edison and Orange & Rockland): 31,000+ EVs enrolled; around 30% EV market share in Con Edison's service territory. ev.energy, The Blueprint for Mass-Market EV Load Management Participation. Southern California Edison street team recruitment and Clean Power Alliance's LA Auto Show Test Track sponsorship with multilingual on-site enrollment, both as published by the programs. Con Edison and Orange & Rockland enrollment via the Tesla app, as announced by the programs.

9. ev.energy program experience and operating model analysis: channel yield and enrollment ceilings, acquisition cost ranges, activation rates by verification design, capability by program type in kW per household, and churn and recovery ranges. These describe how programs of this kind behave in general, drawn from illustrative ranges for a typical V1G portfolio rather than the results of any single program.

10. ev.energy platform metrics, May 2026: modeled US coverage of around 90% of EV drivers via vehicle integrations and approximately 95% via vehicle or charger combined, based on 2025 US EV and home-charger market shares (modeled coverage, not enrolled users); over 410,000 acquired devices and over 300,000 acquired users globally; deactivated device rate of 3.9% across the active portfolio, February to April 2026.