Latest / Logosofia no Dia a Dia / UPS vs Generator: Where's the Line Now?
Transcript
- [Prompt] custom: As a long-time desktop user, I've purchased a few uninterruptible power supplies over the years, or | Hosts: corn, herman [Response] Corn: Daniel sent us this one, and it's one of those questions where the moment you hear it you realize you've been living with the same assumption and never thought to question it. He's got two UPSs in his apartment in Israel — one for the desktop, one for the networking stack — and the power flickers. Not outages, not hours in the dark. The breaker trips, you reset it, life goes on. Except without a UPS, those five seconds take down the Wi-Fi, the Zigbee network, the Matter network, every smart home device connected to them. So his question is: if UPSs are riding the battery revolution, how big can they actually get? Where's the line between "just buy a generator" and "get a really big UPS"? And is that line moving? Herman: The timing on this is perfect, because the economics have shifted dramatically in the last three or four years. Lithium iron phosphate cell costs have dropped about forty percent since twenty twenty-two. We're now around ninety dollars per kilowatt-hour at the cell level. That changes what's possible for a device that's historically been a lead-acid box sitting under your desk giving you eight minutes to save a spreadsheet. Corn: Right, which is the image most people have. The beige shoebox that beeps at you. Herman: And that image is about ten years out of date. The traditional consumer UPS market tops out around fifteen hundred volt-amps, maybe ten minutes of runtime at full load. Then there's the data center rack stuff — ten to fifty kilovolt-amps, still only minutes of runtime because they're designed to bridge the gap until the diesel generator spins up. But what Daniel's asking about is this emerging middle ground that doesn't fit either category. Corn: The thing he spotted in Israel, actually. He mentioned seeing companies like UPSistem and Elspec selling residential and light-commercial units with five to twenty kilowatt-hour battery banks. That's not "save your work and shut down" territory. That's "run your fridge and networking stack for hours" territory. Herman: That middle market exists for a reason that's very specific to the Israeli grid. Pre-Shabbat demand spikes on Friday afternoons cause frequent brownouts — five to thirty seconds, sometimes a few minutes. The utility doesn't have a generation shortfall in the traditional sense, it's a voltage sag from everyone cooking and heating at the same time. So you get these momentary dips that trip sensitive electronics but don't constitute an actual outage. A generator is useless for that scenario. By the time it starts up and stabilizes, the event is over. The UPS wins by default. Corn: Which is the key insight in Daniel's prompt that I think most people miss. He's not asking about backup power for when the grid goes down for hours. He's asking about continuity. The difference between "my smart home survived" and "I'm re-pairing thirty Zigbee devices. Herman: That's a use case that's going to become more common, not less, as homes get more connected. Every smart bulb, every sensor, every smart lock — they all have to rejoin the network after a power cycle. A five-second blip can mean twenty minutes of devices slowly coming back online, if they come back at all without a manual reset. Corn: Let's actually answer the two-part question Daniel asked. Part one: where does the UPS market top out? How big can these things get? Herman: Let's start with the physics, because there are three constraints that determine how large a UPS can scale, and only one of them is the battery. The first is inverter capacity — both continuous and surge. You need an inverter that can handle the rated load indefinitely and also absorb the inrush current when something like a refrigerator compressor kicks on. The second is battery energy density, which determines how much runtime you get for a given physical size. And the third is thermal management, which is the one people forget about. A ten-kilowatt inverter running at full load is generating serious heat. If you're pulling ten kilowatts and the inverter is ninety-three percent efficient, you've got seven hundred watts of heat to dissipate. That requires active cooling, and the fans themselves draw power, which eats into your runtime. Corn: There's a parasitic load that scales with the inverter size. Herman: And that's part of why the largest single-phase UPS units top out around twenty kilovolt-amps. APC's Symmetra PX line hits twenty kilowatts in a single-phase configuration. Beyond that, you're into three-phase industrial territory, and Schneider Electric's Galaxy VX line goes up to five hundred kilovolt-amps or more. But those are for factories and data centers — they require dedicated electrical rooms, three-phase power, and professional maintenance contracts. Corn: For a home, the practical ceiling is around twenty kilovolt-amps, single-phase. Which is still enormous. That's enough to run an entire large house with multiple air conditioners. Herman: But now let's talk about the battery side, because this is where things have gotten interesting. The traditional UPS used valve-regulated lead-acid batteries. Energy density around thirty to fifty watt-hours per kilogram. A ten-kilowatt-hour lead-acid bank weighs about three hundred kilograms and takes up the space of a large filing cabinet. An equivalent lithium iron phosphate bank today weighs about sixty kilograms and is the size of a mini-fridge. And the cost has come down from roughly eight thousand dollars for the cells in twenty twenty to around three thousand dollars today. Corn: That's the forty percent drop you mentioned. Herman: It's continuing. Bloomberg NEF's latest survey has LFP cells at ninety dollars per kilowatt-hour at the pack level for utility-scale, slightly higher for residential formats. But we're approaching the point where the battery is no longer the dominant cost in the system. The inverter and the power electronics are becoming the expensive part. Corn: Which is a complete inversion of how people think about UPS economics. The mental model is "batteries are expensive, so UPSs must be small." That's no longer true. Herman: The Israeli market is actually a fascinating case study in this transition. The grid instability there — particularly those pre-Shabbat demand spikes — has created a niche for what I'd call "brownout bridges." These are UPSs sized to handle thirty to sixty minutes of partial load, not five minutes. Companies like SolarEdge now sell DC-coupled battery systems that function as UPSs with solar recharge capability. They're not marketing them as UPSs, they're marketing them as home batteries. But the transfer switch in those systems operates in under twenty milliseconds, which is the UPS spec. Corn: Let's nail down that spec, because it's the technical distinction that matters most for Daniel's use case. A UPS has to switch to battery power in under twenty milliseconds — that's one AC cycle at fifty hertz, slightly faster at sixty hertz. A generator takes ten to thirty seconds to start and stabilize. For a five-second brownout, the generator literally never triggers. The event is over before the generator even cranks. Herman: That's the answer to "why not just get a generator?" for Daniel's specific scenario. It's not about cost. It's about physics. The generator can't respond fast enough. So even if a generator were free, it wouldn't solve his problem. Corn: Which brings us to part two of his question. Where's the crossover point? When does a generator become the better tool? Herman: Let's build a decision matrix. For outages under one hour, the UPS wins on convenience alone — silent, instant, no fuel to store, no maintenance schedule. For one to four hours, it's a toss-up that depends entirely on your load profile. For four-plus hours, the generator wins on cost per kilowatt-hour of storage. But here's what's changing: at twenty twenty-six battery prices, a twenty-kilowatt-hour UPS — enough to run a fridge, networking gear, and lights for about twelve hours — costs roughly six thousand dollars in cells. A ten-kilowatt standby generator installed costs about forty-five hundred dollars. The generator still has the edge on upfront cost, but the gap has narrowed dramatically. Corn: That's before you factor in fuel costs over ten years. Herman: A generator burning natural gas or propane has ongoing costs. A UPS has essentially zero operating cost — it charges from the grid and sits there. Over a decade, the total cost of ownership starts to favor the battery, especially if you're only using it for short-duration events. Corn: Let's put some real products on the table, because I think people would be surprised by what's available. The EcoFlow Delta Pro Ultra is a seven-point-two kilowatt-hour unit with a seventy-two-hundred-watt inverter, expandable to twenty-five kilowatt-hours. It's marketed as a "home backup battery" but it functions identically to a UPS — it has a transfer switch that operates in under twenty milliseconds. Base unit costs about five thousand dollars. Herman: That's a consumer product you can buy online. It's not a specialty industrial item. Bluetti and Anker Solix have competing products in the same space. The "portable power station" market has essentially created de facto home UPSs with five to twenty-five kilowatt-hours of capacity. They're just not called UPSs. Corn: Which is a marketing failure, honestly. If you search for "home UPS," you get the beige shoebox. If you search for "home battery backup," you get these. Same function, completely different product category in the consumer's mind. Herman: This is where the line between UPS and generator starts to blur in the other direction too. Generac and Tesla are selling home batteries — the Powerwall, the PWRcell — that include UPS-grade transfer switches. The Powerwall's gateway switches in under a hundred milliseconds, which is slightly slower than a dedicated UPS but fast enough that most electronics don't notice. They're not marketing them as UPSs either, but functionally, that's what they are. A Tesla Powerwall is a thirteen-point-five kilowatt-hour UPS with solar integration. Corn: The categories are dissolving from both sides. The UPS is growing up into home backup territory, and the home battery is reaching down into UPS territory. Herman: They're converging on the same product. The distinction is becoming meaningless. Corn: There's one area where the generator still has a clear moat, and we should be honest about it. High-draw appliances. A five-ton air conditioner draws five to six kilowatts starting surge. A residential UPS that can handle that costs ten thousand dollars plus. A twelve-kilowatt generator costs three thousand dollars. Until battery prices drop another fifty percent, generators keep the high-load niche. Herman: And multi-day outages are another generator stronghold. If a hurricane takes out your power for a week, no practical home battery is going to cover that. You'd need solar plus a massive battery bank, and at that point you're not talking about a UPS anymore — you're talking about an off-grid power system. Corn: The generator's remaining advantages are runtime and peak power. The UPS's advantages are response time, silence, zero emissions indoors, and zero maintenance. Different tools for different jobs. Herman: The market reflects this. The residential battery backup market in the one-to-ten-kilowatt-hour range is growing at twenty-two percent compound annual growth rate, according to the most recent projections from twenty twenty-four through twenty thirty. That's faster than the generator market, which is growing in the single digits. The trend is toward batteries, but it's not a displacement yet. It's more like the battery is claiming the short-duration, high-frequency use cases, and the generator is holding the long-duration, high-power use cases. Corn: Which maps perfectly onto Daniel's situation. He's in the short-duration, high-frequency bucket. Frequent five-second brownouts, almost never a multi-hour outage. The UPS is unambiguously the correct tool for that job. Herman: Here's the thing: most people don't actually know their own outage profile. They have a vague sense that "the power goes out sometimes" and default to thinking about generators because that's the cultural script. But if you actually logged every power event for a year, I suspect a lot of people would discover that eighty or ninety percent of their events are under five minutes. And for those, a generator is not just overkill — it's the wrong tool entirely. Corn: Let's talk about the forward-looking angle Daniel raised. Is battery technology advancing faster than generator technology? Herman: It's not even close. Internal combustion generators are a mature technology. The efficiency of a small gasoline or natural gas engine hasn't improved meaningfully in decades. Battery technology, on the other hand, is still on a steep learning curve. We're seeing improvements in energy density, cycle life, and cost reduction every year. Solid-state batteries are in pilot production. Sodium-ion is emerging as a cheaper alternative to lithium for stationary storage. The trajectory clearly favors batteries. Corn: There's a wildcard that I think reshapes this whole conversation in the next five to ten years. Bidirectional EV charging. Vehicle-to-home, or V-to-H. Herman: This is the one that keeps me up at night, in a good way. A Ford F-150 Lightning with the extended range battery has a hundred and thirty-one kilowatt-hours of capacity. Its Pro Power Onboard system can deliver nine-point-six kilowatts of backup power. That's more than most residential UPS units, and it's already in your garage. If V-to-H becomes standard — and there are pilot programs running right now with Ford and Sunrun — the dedicated home UPS market could be disrupted by cars that already have enormous batteries. Corn: You're basically parking a hundred-kilowatt-hour UPS in your garage every night. Herman: The car's battery is already being managed, already has a thermal system, already has a battery management system. The marginal cost of adding bidirectional charging capability is relatively small. The question is whether the transfer switch in a V-to-H system can hit that twenty-millisecond UPS spec. Most current implementations are closer to a few seconds of switchover time, which is fine for outage backup but not for brownout bridging. Corn: There's still a niche for the dedicated UPS even in a V-to-H world. That sub-twenty-millisecond switchover is the moat. Herman: And for Daniel's use case — five-second brownouts — that moat matters. The car might not even notice the grid dropped. The dedicated UPS catches it instantly. Corn: Let's make this practical. If someone's listening and thinking about their own home, what's the decision framework? Herman: Step one: calculate your brownout budget. How many power events do you have per year, and how long are they? If more than eighty percent are under five minutes, a large UPS is almost certainly the better investment. If more than fifty percent are over an hour, get a generator. If you're somewhere in the middle, consider a hybrid system. Corn: Step two: figure out what you actually need to keep running. For most people, the highest-return-on-investment UPS purchase is a small unit dedicated to networking gear. Fifteen hundred volt-amps, maybe two hundred dollars. That keeps your internet, Wi-Fi, and smart home alive through ninety-five percent of power events. Daniel already figured this out — he's got separate UPSs for his desktop and his network stack. Herman: That's a lesson worth underlining. Even if you do nothing else, a small UPS on your networking gear is the single most impactful power resilience purchase you can make. It's cheap, it's simple, and it solves the most annoying failure mode — the five-second blip that takes down your entire connected home. Corn: Step three: if you're in the market for something bigger, the sweet spot for a home UPS right now is five to fifteen kilowatt-hours with LFP batteries and a pure sine wave inverter. Brands like EcoFlow, Bluetti, and Anker Solix are competing in this space with prices under two dollars per watt-hour of storage. That's a dramatic change from even three years ago. Herman: Step four: if you're considering a generator, look for one with an automatic transfer switch that can also accept a battery input. Hybrid systems — generator plus battery — are becoming the gold standard for whole-home backup. The battery handles the short events silently and instantly. The generator handles the long outages. You get the best of both. Corn: There's a real product category emerging here that doesn't have a good name yet. It's not a UPS in the traditional sense, and it's not a generator. It's a home energy buffer. And I think in ten years, that's just going to be a standard appliance, like a water heater. Herman: The "home energy buffer" is a great way to put it. And we're already seeing the early versions of this from companies that aren't traditionally in the power business. Anker's not a UPS company. EcoFlow's not a UPS company. They're consumer electronics companies that realized batteries had gotten cheap enough to make home energy storage a mass-market product. Corn: That's the real answer to Daniel's question about whether UPSs will displace generators. The categories are dissolving. In ten years, we probably won't talk about UPSs versus generators. We'll talk about battery-backed homes with solar, storage, and bidirectional EV charging. The UPS will just be one function of a larger energy management system. Herman: The dedicated UPS doesn't go away entirely, but it retreats to the instant-switchover niche. Sub-twenty-millisecond transfer time, pure continuity. Everything else gets absorbed into the home battery ecosystem. Corn: Which means Daniel's two small UPSs are actually ahead of the curve. He's already using the right tool for the right job, and the market is only now catching up to what he figured out through lived experience in an apartment with flaky power. Herman: That's the thing I love about this question. He started with a practical observation — "my UPSs solve a problem generators can't" — and backed into a question about the entire trajectory of home energy storage. That's good pattern recognition. Corn: Where does the UPS market top out? For single-phase residential use, around twenty kilovolt-amps, which is enough to run an entire large home. For three-phase industrial, five hundred kilovolt-amps and beyond. But the more interesting answer is that the ceiling is rising faster than most people realize, and it's being pushed up not by UPS companies but by battery chemistry advances that are happening in adjacent industries. Herman: Where's the crossover point with generators? For outages under an hour, UPS wins. For four-plus hours, generator still has the edge on cost. But that crossover is shifting by the year. At current battery price trajectories, the four-hour crossover could become an eight-hour crossover by twenty thirty. Corn: One more thing worth mentioning. Daniel's in Israel, and the Israeli market has some specific dynamics that make large UPSs more attractive than they might be elsewhere. The pre-Shabbat demand spikes are predictable and frequent. The outages are short. Natural gas for generators isn't as universally available as it is in North America. And electricity prices are high enough that pairing a UPS with solar starts to make economic sense as an arbitrage play — charge when electricity is cheap, discharge when it's expensive. Herman: The solar arbitrage angle is actually huge. A UPS that just sits there waiting for outages has a terrible utilization rate. It's idle ninety-nine-point-nine percent of the time. But if it can also do time-of-use shifting — charge during off-peak, discharge during peak — it pays for itself. That's the Tesla Powerwall model, and it's why those systems are marketed as "home batteries" rather than UPSs. The backup function is almost a side benefit. Corn: The final framework for Daniel, and for anyone thinking about this: calculate your brownout budget. If your power events are mostly short, get a UPS — and the new generation of large-format LFP units gives you options that didn't exist five years ago. If your events are mostly long, get a generator. If you're somewhere in the middle, or if you want to future-proof, look at a hybrid system with a battery plus a generator transfer switch. And if you've got an EV in the garage, keep an eye on V-to-H. That's the wildcard that could make all of this obsolete. Herman: The one thing I'd add: don't sleep on the small UPS for networking gear. It's the highest-leverage purchase in this entire space. Two hundred dollars to keep your smart home alive through every five-second blip. Everything else is optimization on top of that foundation. Corn: That's really the answer to the question Daniel didn't ask but was lurking underneath the whole thing. "Am I weird for using UPSs this way?" The answer is no. You're early. And now: Hilbert's daily fun fact. Hilbert: The ancient Indian board game Chaturanga, widely considered the ancestor of chess, was long thought to have been invented in the Gupta Empire around the sixth century. However, a set of carved stone pieces discovered in a cave system in Guyana's Pakaraima Mountains in the nineteen seventies was initially misattributed by a British archaeologist as evidence that Chaturanga reached South America via early medieval trade routes. The pieces were later carbon-dated to roughly seven hundred AD, but subsequent analysis in twenty fourteen showed they were actually from a completely unrelated indigenous counting game called Waramapú, which used similar geometric carvings for territorial scoring and had nothing to do with chess or Indian contact. Herman: I have so many questions and I know we're not answering any of them. This has been My Weird Prompts. Our producer is Hilbert Flumingtop. If you want to figure out your own brownout budget, or if you've got a question about the blurry line between two technologies that everyone thinks are separate, email the show at show at my weird prompts dot com. I'm Corn. Herman: I'm Herman Poppleberry. Check your breaker panel and we'll talk next time.