Report 120 · Energy Storage
Every home battery is sold on one number, its kilowatt-hours, and that number cannot answer the question. Whether your air conditioner restarts during an outage is decided by a specification most brochures never mention: the motor-start rating. Two popular batteries with similar storage differ by roughly two to one on it, and one of them cannot start a common three-ton condenser on its own.
A friend called me during an outage last summer, standing in his kitchen, annoyed. He had bought a home battery specifically so the air conditioning would keep running, and the battery was reporting most of a full charge. The lights were on. The refrigerator was humming. The house was 84 degrees and climbing, because every time the thermostat called for cooling, something clicked and nothing happened.
He had not run out of energy. He had run out of a different thing entirely, and nobody selling him the system had ever named it.
I have written before about how a gigawatt of storage is half a number at grid scale. This is the residential version of the same confusion, and at home it has a sharper edge, because at home it is decided in the first third of a second.
A battery system has to satisfy three separate constraints before your compressor turns.
Energy, in kilowatt-hours, is how much total work is stored. This is the brochure number. It determines how long things run.
Continuous power, in kilowatts, is how fast the system can deliver that energy on a sustained basis. It determines how much can run at once.
Motor-start capability is the one nobody talks about. It is the very short, very large current surge a system can supply to get a stationary motor spinning. It determines whether the largest thing in your house turns on at all.
You can pass the first two and fail the third, which is exactly what happened in my friend's kitchen. Failing the third looks like nothing happening, which is why it gets misdiagnosed as a broken battery or a broken air conditioner.
A toaster is a resistor. Apply voltage and it draws its rated current immediately.
An air-conditioning compressor is a motor, and a motor at rest is close to a short circuit. There is no rotation yet, so there is no back-EMF opposing the supply, and until the rotor moves the winding draws enormously more current than it will draw a second later. Manufacturers publish that worst case as Locked Rotor Amps, and it appears on the equipment nameplate next to Rated Load Amps.
The numbers are not close to each other. Take Goodman's GSXN4 residential condenser line, whose published product specifications cover 208/230 volt single-phase units from 1.5 to 5 tons:
1.5 ton: 6.1 RLA, 35.1 LRA
2 ton: 8.4 RLA, 41.2 LRA
2.5 ton: 11.6 RLA, 59 LRA
3 ton: 16 RLA, 91.9 LRA
3.5 ton: 17.7 RLA, 110.2 LRA
4 ton: 19.9 RLA, 110 LRA
5 ton: 25.6 RLA, 150 LRA
The three-ton unit runs at 16 amps and starts at 91.9. That is a factor of 5.7. Running, it is a modest load: 16 amps at 230 volts is under 4 kilowatts, comfortably inside any home battery's continuous rating. Starting, it briefly demands more current than the rest of the house combined.
A utility grid does not notice. It is effectively an infinite stiff source at your service entrance, and it swallows the inrush without comment. This is why the problem is invisible until the grid goes away. Your inverter is not an infinite source. It is a few kilowatts of silicon with a hard current limit, and when the demand exceeds what it can source, it protects itself instead of starting your compressor.
Reputable manufacturers publish a motor-start figure. It is usually one line, in the middle of a specification table, and it is the single most decision-relevant number in the document.
Tesla Powerwall 3 stores 13.5 kWh nominal and delivers up to 11.5 kW AC continuous per unit. Its datasheet lists Load Start Capability: 185 LRA. The marketing text says the same thing in plain English, that it "has the ability to start heavy loads rated up to 185 LRA, meaning a single unit can support the power needs of most homes."
Enphase IQ Battery 10C stores 10.0 kWh usable and delivers 7.08 kVA continuous, with peak output current of 56 amps for three seconds and 44.8 amps for ten seconds. Its datasheet lists Power Start capability: Up to 90 A LRA, with a footnote reading, in full, "Power Start capability may vary."
Now put those against the condenser table. On energy, these two products are 35 percent apart, which sounds like a modest difference and is roughly how they are compared in every buying guide. On motor start they are about two to one, and that ratio lands on the wrong side of a very common piece of equipment: a single IQ Battery 10C is rated to 90 LRA, and a three-ton GSXN4 condenser is 91.9 LRA.
I want to be precise about what that does and does not mean, because it would be easy to turn it into a cheap shot at one vendor. It is not one. Both figures are per unit, and both systems are explicitly designed to be stacked: the Powerwall 3 supports up to four units, and Enphase expects multiple batteries in most backup designs, with the peak capability scaling accordingly. A two-battery Enphase system is a different machine from the one on that spec line. The point is not that one product is inadequate. The point is that the number that decides your outcome is not the number you were shown, and that two systems can look 35 percent apart on the brochure while being 100 percent apart on the thing that actually determines whether you get cool air.
While reading the Powerwall 3 datasheet I noticed a second gap worth knowing about, on the other side of the ledger.
That unit discharges at up to 11.5 kW. It charges at up to 5 kW (20.8 amps AC), rising to 8 kW with up to three expansion units attached. Discharge and charge are not symmetric, and the charge side is the quieter constraint.
In a one-evening outage this never comes up. In a multi-day outage it becomes the whole story, because your battery is then a buffer being refilled by solar during daylight and drained all night. What you can put back between sunrise and sunset sets your real sustainable load, not what the battery can dump on demand. A system that can deliver 11.5 kW but only absorb 5 kW is telling you something specific about how it will behave on day three, and it is not on the front page of anything.
This is the same lesson as what a battery's efficiency number leaves out: the headline spec describes the best case in one direction, and the constraint that binds you lives somewhere else in the table.
If the arithmetic does not work, there are three real options and one bad one.
Add capacity. More battery units means more start capability. This is the expensive answer and the one your installer will reach for first, understandably.
Add a soft starter. A soft starter is a small device wired at the compressor that ramps the winding current instead of applying full voltage instantly. Micro-Air, which makes the widely used EasyStart units, describes the mechanism as controlling "the power supplied to the windings of the motor on each AC cycle," gradually increasing it "until the compressor is running," and claims this "reduces the peak current by 50 to 70%." That is a manufacturer's claim about its own product rather than an independent measurement, and I have not seen third-party verification of the range. But the physics is uncontroversial and the device category is standard practice, and if the claim holds even at the low end it moves a 91.9 LRA compressor to roughly 46, which changes the answer completely. This is usually the cheapest path by a wide margin, and it is a job for your HVAC contractor, not your battery installer.
Do not back up the air conditioning. An unglamorous but often correct answer. A critical-loads panel that carries the refrigerator, well pump, medical equipment, network gear and some lighting will run far longer on the same battery than one dragging a compressor through a duty cycle. Decide what you actually need at 2 a.m.
The bad option is the one people default to: buying a bigger number of kilowatt-hours and hoping. Energy was never the binding constraint. You can double the storage and still hear the same click.
Disclosure, since I work in this field rather than only writing about it: I help design the AI battery-cycling systems for a veteran-owned (HUBZone) energy-storage integrator. I do not own that company and earn nothing from this link. Full policy here. I mention it because most of what I know about this failure mode I learned from people fielding the callback, and because the conclusion here, that some readers should back up less than they planned, does not sell anybody more equipment.
Three questions, and they take about five minutes to answer if the person quoting you is competent.
What is the LRA on the nameplate of my condenser? It is physically printed on the outdoor unit. Not an estimate, not a rule of thumb from tonnage, the actual number on the actual equipment.
What is the motor-start rating of the system you are quoting me, in this configuration? In this configuration matters, because per-unit ratings and stacked ratings are different, and the datasheet figure assumes the manufacturer's own conditions.
What else is running when the compressor tries to start? Start capability is not evaluated in isolation. If the well pump and the refrigerator are drawing at the same moment, the compressor is starting into a system that is already partly committed.
If the answer to any of them is a shrug or a reassurance, you have learned something useful about the quote.
All three product figures come from manufacturer documents, which I opened and read, but manufacturer documents are marketing artifacts as well as engineering ones. I have not tested any of these products, and I have not seen independent bench verification of any published motor-start rating.
Datasheets are revised. The Powerwall 3 document I read is the 2025 US datasheet, and the Enphase document is revision DSH-00565-9.0-EN-2026-02-26 for the IQBATTERY-10C-1P-NA. The Goodman figures are from the SS-GSXN4 product specifications; I checked both the 5/22 and the 11/22 revision and the RLA and LRA values are identical. Check the current revision for your own model before relying on any number here.
The Enphase and Goodman comparison is a like-for-like on paper only. A rated start capability of 90 LRA against a nameplate 91.9 LRA is a margin of about 2 percent, which is inside the range where installation specifics, voltage at the panel, ambient temperature, and the manufacturer's own "may vary" footnote all matter. I am using it to show that the two numbers live in the same range, not to predict the outcome of any particular installation. Do not read it as a claim that this pairing will fail; read it as a claim that this is the pairing you have to check.
I did not evaluate every home battery on the market. Two were chosen because they are common and because both publish the relevant number, which is more than can be said for parts of this industry.
The soft-starter percentage is Micro-Air's own claim, clearly labeled as such above. I found no independent test of it and did not cite the third-party figures that circulate for that product category.
Nothing here rests on my own research. My published work is in microwave spectroscopy, not power electronics. The specifications belong to the manufacturers whose documents are cited below.
There is a pattern behind this that shows up every time a technology gets sold to consumers, and it is worth naming.
A complicated product gets compressed into one number, the number is chosen because it is easy to compare, and the failure modes migrate to whatever the number does not describe. Storage capacity is a genuinely useful figure. It just happens to say nothing about the first third of a second, which is the interval that decides whether a house stays cool.
So when you are told a battery will "run your whole home," ask what that sentence is a claim about. Usually it is a claim about energy, sometimes about continuous power, and almost never about the surge. The three are independent, your air conditioner tests the third one first, and it tests it every single time the thermostat clicks.
Scope note: this report explains a general electrical constraint and compares published manufacturer specifications. It is not an endorsement or criticism of any product, not a design specification, and not a substitute for a load calculation performed for your building. Backup power system design, critical-loads panel selection, and any modification to air-conditioning equipment must be performed by licensed professionals in accordance with the National Electrical Code and local requirements. Specifications are drawn from the document revisions named in the sources and are subject to change; verify against the current revision for your own equipment. Disclosure: the author helps design AI battery-cycling systems for a veteran-owned energy-storage integrator, as stated in the body of this report, and does not own that company.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.