There is a version of the storage story that has hardened into consensus: batteries got cheap, batteries got fast to deploy, and the grid’s next chapter belongs to distributed batteries aggregated into virtual power plants. Most of that is right. The part that is incomplete is the quiet assumption that a behind-the-meter storage asset has to be electrochemical.
The case has no more effective advocate than Jigar Shah, the SunEdison co-founder and former head of the Department of Energy’s Loan Programs Office, whose newsletter essay How Home Batteries Unlock the Grid’s Next Chapter is the sharpest recent version of it. His broader argument is one we would sign without edits: storage should stop being framed as a cost the grid tolerates in order to make wind and solar work, and start being understood as the thing that lets every other asset — fuel plants, transmission, distribution — run closer to what it is actually capable of. Storage is not a renewables accessory. It is grid infrastructure that happens to live in a basement.
Our only quarrel is with the inventory. Walk into the mechanical room of almost any apartment building, hotel, school, or hospital in the country and you will find storage already installed. It holds heat instead of electrons, nobody on the utility side has it on a list, and it is almost never dispatched on purpose. The tanks are there for hydraulic and comfort reasons — to stop a compressor short-cycling, to cover a morning shower peak, to keep a chiller off during the worst hour of the afternoon. Every one of those tanks is also a battery in the only sense the grid actually cares about: it decouples when energy is bought from when energy is used.
In short: any tank holding heated or chilled water is an energy storage device. Sized and controlled deliberately, it moves real electrical load off the peak at a capital cost per stored kWh well below electrochemical storage — in buildings whose largest electrified loads are thermal to begin with. What it cannot do is export power or keep the lights on. That is why the honest framing is complementary, not competitive.
In this article
The federal case for distributed flexibility is not speculative anymore. The U.S. Department of Energy’s Pathways to Commercial Liftoff analysis of virtual power plants concludes that tripling current virtual power plant capacity to 80–160 GW by 2030 could cover 10–20% of national peak load and save on the order of $10 billion a year in grid costs, largely by avoiding generation build-out and deferring distribution investment. DOE’s framing of the problem is the part worth sitting with: the country needs to serve roughly 200 GW of additional peak demand by 2030 while interconnection queues run multi-year, which means the capacity that can be deployed this year is structurally more valuable than capacity that clears a queue in 2031.
Buried in DOE’s own accounting of where that capacity comes from is a line that rarely gets quoted. Alongside the EV chargers and stationary batteries, the department expects the grid to add roughly 5 to 6 GW per year of flexible demand from smart thermostats, smart water heaters, and non-residential distributed resources between 2025 and 2030. Read that again with an engineer’s eye: almost all of that is thermal capacity. It gets filed under “flexible demand” rather than “storage,” which is an accounting convention, not a physical distinction. A water heater that pre-heats at noon and coasts through the evening peak has stored energy. The tank is the storage medium.
There is also a permitting asymmetry that engineers notice before economists do. A utility-scale battery needs an interconnection study, a queue position, and a site. A tank needs a mechanical room and a doorway.
Utilities have known this for forty years. Electric thermal storage programs, in which a utility interrupts or schedules water heater loads to manage peak, are documented in Pacific Northwest National Laboratory’s analysis of large-capacity water heaters in electric thermal storage programs and predate most of the modern DER vocabulary. What has changed is not the physics but the control layer: metering, two-way communication, and aggregation software that can bid a fleet of tanks the way a plant operator bids a unit. Brattle’s work for the rural electric cooperatives made the framing plain years ago — electric water heaters are thermal batteries that are already installed and mostly sitting idle.
The load is not marginal, either. In large apartment buildings, EIA’s Residential Energy Consumption Survey finds water heating frequently exceeds space heating as an end use — the apartments insulate each other, but everyone still showers. In an electrified multifamily building, domestic hot water is often the single largest controllable electrical load in the building, and it is the one load that comes with a storage vessel already attached.
An argument that thermal storage is undervalued is only worth making if it is honest about the ceiling. Heat is a one-way medium. You can put electricity in and take heat out; you cannot take electricity back out. That single fact draws the line between the two technologies, and it decides which one belongs in which building.
This is where most comparisons go wrong, including some made by people selling tanks. A kWh of stored heat is not interchangeable with a kWh of stored electricity. When a heat pump charges the tank, the electrical energy shifted is the thermal energy divided by the coefficient of performance.
Take the 700-gallon module: 108 kWh of stored thermal capacity at a 35°C temperature delta, listing at $5,798, or about $54 per stored thermal kWh. Charge it with a heat pump water heater running at a seasonal COP of 3.0 and the tank represents roughly 36 kWh of shifted electrical consumption — about $161 per electrical kWh of shift capability. That number is higher than the headline, and quoting the headline without it is the kind of thing technical buyers catch immediately.
Two things keep the economics attractive anyway. First, the comparison is not storage against storage; it is incremental storage against the alternatives for serving a load that must be served regardless. The building is buying hot water either way. The marginal question is whether the extra tank volume costs less than the demand charges, peak energy, heat pump capacity, and service upgrade it avoids. Second, thermal storage does not fade. There is no cycle-count budget to spend.
Grid-scale language obscures how a building actually gets paid. In practice there are five channels, and they are not equally sized.
FERC Order 2222 requires wholesale market operators to let aggregations of distributed resources compete alongside conventional generation. The Commission’s own explainer lays out the region-by-region schedule, and the schedule is the story: this is arriving in the back half of the decade, unevenly.
There is a second gap underneath the timelines. As of early 2026, no state has fully built the coordination rules that govern how aggregators and distribution utilities exchange data and manage non-performance, which is the unglamorous plumbing that makes any of this dispatchable in practice. The honest conclusion for a building owner is that wholesale participation should be treated as option value, not as underwriting. Size the storage on the tariff you are on today — time-of-use energy, demand charges, and whatever your utility currently pays for load flexibility — and treat market access as upside that arrives on someone else’s schedule. Utilities and program administrators evaluating thermal storage as a program measure can start with our overview for utilities and program teams.
The practical version of this argument is short. Any building doing an electrified hot water or hydronic retrofit in the next few years is going to install a tank anyway. The decision worth making deliberately is how big, and against what.
It is storage. A tank decouples the time energy is purchased from the time it is used, which is the functional definition of storage. Utilities have run water heater load control and electric thermal storage programs on exactly this basis for decades. What it is not is a generation asset — it reduces and shifts consumption but cannot export power to the grid.
They solve different problems. Thermal storage is far cheaper per stored kWh, has no capacity fade, and sites in an ordinary mechanical room, but it can only shift thermal loads and cannot back up lighting, plug loads, or life safety systems. Batteries can serve any load and export power, at a substantially higher cost per kWh and with fire code and siting requirements. Buildings with large electrified hot water or chilled water loads usually get more flexibility per dollar from thermal storage; buildings that need resilience need a battery.
Divide the stored thermal energy by the coefficient of performance of the heat source. A 108 kWh thermal module charged by a heat pump at a COP of 3.0 shifts roughly 36 kWh of electricity. Quoting thermal kWh as if they were electrical kWh overstates the shift by the COP multiple.
Through utility channels, yes — time-of-use energy savings, demand charge reduction, and demand response or load flexibility program payments where the utility offers them. Wholesale market revenue through distributed aggregation is arriving unevenly under FERC Order 2222, with capacity participation targeted for 2026 in ISO-NE, 2027 in PJM, and 2030 in SPP. Underwrite on the tariff you are on now and treat market access as upside.
The opposite. Storage and generation capacity trade against each other: more storage lets a smaller heat pump meet the same peak draw, lowering connected electrical load. In electrically constrained retrofits this tradeoff is often the reason the project is feasible at all.
For daily cycling, yes. Modular tanks lose roughly 2.4°F to 8°F over 24 hours depending on size, with larger vessels losing proportionally less. That supports charging off-peak and discharging through an evening peak. It does not support multi-day storage, and any design that assumes it will is going to disappoint.
The distributed storage build-out is real and the numbers behind it are large. What has been miscounted is the inventory. Before a single new battery is installed, American buildings already contain an enormous fleet of storage vessels that collectively represent gigawatts of shiftable load — sized for comfort, controlled by a thermostat and a timer, and invisible to the people modeling the grid’s next decade. Turning that fleet into a resource does not require a breakthrough. It requires sizing tanks against the tariff instead of against the minimum, and giving them a control layer that can take a signal.
Shah is right that storage is grid infrastructure rather than a renewables expense. The correction is only that the infrastructure is already partly built, and it is sitting next to the boiler. Batteries will keep doing the things only batteries can do. The point is that the list of things they are needed for is shorter than the current conversation assumes, and the cheapest kWh of flexibility in most buildings is already in the mechanical room.
The fastest way to find out what your building’s thermal load is worth as flexibility is a short engineering conversation about the plant, the tariff, and the service capacity.
Tell our team about your building, operating goals, and energy needs.
