Huawei Set To Challenge Tesla Powerwall with New, 20 kW Home Battery (electrek.co) 53
"Huawei believes its new Luterra S2 is more than just a home backup battery," reports Electrek:
This high-tech Huawei home battery pack hopes to hold back Tesla's latest Powerwall with silicon carbide technology, AI-powered energy management baked in, and enough modular storage to keep even large homes bright and warm when the rest of the block goes dark... The Chinese tech firm is calling it an "all-in-one residential energy system" that combines a hybrid inverter, modular battery storage, and AI-powered energy management in a single, simple, and stackable package that's available in single-phase versions that put out between 3 and 10 kW and three-phase versions that can push up to 20 kW. The Luterra S2 is the first system of its kind to use Huawei's proprietary full-silicon carbide (SiC) architecture, which promises improved efficiency, power density, and ease of use. It's so easy, in fact, that Huawei claims it can be installed and commissioned in just minutes.
Huawei's basic Luterra S2 configuration starts with "just" 5 kWh of battery energy storage and 3 kW of output. That's enough to keep essential household appliances, lighting, and electronics running during a power outage, but you'll want to keep going if you're preparing for more than a day or two without power. If that sounds like you, Luterra's got you covered. The modular system can be expanded with modular 7 kWh LFP battery packs and can keep scaling up to a massive 336 kWh of total storage across interconnected systems, with inverter output peaking at an impressive 60 kW using three 20 kW units in parallel. Keeping all that power cool and quiet is a Huawei-developed passive cooling system that does without external fans, keeping operating noise levels below a claimed 29 dB (about the same as my Mac mini).
Huawei's basic Luterra S2 configuration starts with "just" 5 kWh of battery energy storage and 3 kW of output. That's enough to keep essential household appliances, lighting, and electronics running during a power outage, but you'll want to keep going if you're preparing for more than a day or two without power. If that sounds like you, Luterra's got you covered. The modular system can be expanded with modular 7 kWh LFP battery packs and can keep scaling up to a massive 336 kWh of total storage across interconnected systems, with inverter output peaking at an impressive 60 kW using three 20 kW units in parallel. Keeping all that power cool and quiet is a Huawei-developed passive cooling system that does without external fans, keeping operating noise levels below a claimed 29 dB (about the same as my Mac mini).
Re-use case (Score:2, Insightful)
While the idea remains cool and useful for remote deployments, or deployments with very spiky electricity prices (i.e. too much intermittents in the grid), it's really odd to me that all these automotive giants haven't started adopting power wall-like solutions for automotive battery reuse.
Worn out 15 year old battery that's at something like 60% capacity? Just take it as is, add appropriate inverters and use as residential backup battery at scale. Should cost pennies compared to new batteries and offers a
Re:Re-use case (Score:4, Interesting)
.. it's really odd to me that all these automotive giants haven't started adopting power wall-like solutions for automotive battery reuse..
Already happening:
https://www.canarymedia.com/ar... [canarymedia.com]
and
https://pv-magazine-usa.com/20... [pv-magazine-usa.com]
As examples.
Re: (Score:1)
That's grid level, and there these are basically irrelevant outside edge cases like "we're on a smallish energy island far away from main grid lines, so this is cheaper than building our own peaker or pulling cables over long distance". There's ironically nowhere near enough of automotive batteries to matter for that use case at more common scales of grid level deployments, and won't be for a long while. Maybe in a decade when 2020 levels of EV sales of batteries start to get retired. Until then, there's un
Re: (Score:2)
Re: (Score:1)
Markets in heavily bureaucratized areas do not form on their own because bureaucratic inertia acts as a massive brake on any such forces.
AI required to do power management (Score:4, Funny)
This reminds me of Douglas Adams's AI elevators that just sulk in the basement because of their menial existence. These AI batteries could very well also become depressed and simply refuse to take a charge.
Re: (Score:2)
There's a difference between "required" and "may be a good idea". AI is still a relatively good tool to adapt and adjust parameters for real time multi-variable information, especially when a model is well trained to a specific task.
I know that in 2026 everyone thinks AI = Talking to a hallucinating LLM, but the term covers far more than that.
Re: (Score:2)
I know that in 2026 everyone thinks AI = Talking to a hallucinating LLM, but the term covers far more than that.
They used the term specifically to invoke that idea, though. Otherwise they wouldn't mention it.
Re: (Score:2)
Re:No price? (Score:4, Informative)
520 euros with free shipping for a 5kWh LiFePo Chinese battery in a box with smart controller and all bells and whistles. Tested a set of 4 (so similar capacity), working fine. The MPPT solar controller is another 160 euros, can drive 16 of those. Solar panels configured to deliver 100-120V and about 20A. Don't remember how much, prolly around eur 700. Wiring is trivial, but thick, so another 200-250 eur. for wires and another 100 for fuses, boxes, etc.
So, a total of about 3-3.2k eur for my smallest country home.
I seriously doubt they Huawei can match the price or demonstrate a much better quality.
Re: (Score:1)
Re: (Score:2)
Yeah, a neighbor installed a powerwall 2 last year. He paid four times more than I did - and that was only the battery. Mine doesn't seem to perform any worse. If Huawei sets the price to half of the tesla, that'll still be two times more expensive, and I don't think there is much value added for the difference.
Why AI? (Score:5, Insightful)
The algorithm required to handle a home-scale UPS, even if you do more than just keep it properly charged while utility power is available and occasionally run a diagnostic test or two, is so basic and long-established that is seems like adding 'AI' anything is a massive waste of time and potentially introduces new failure modes.
Re: (Score:2)
Slow learner. I simply can't get it through my head that these techniques work well enough for unethical people to keep employing them.
It's just so stupid that I'm continually incredulous the behavior persists.
Re: (Score:2)
There is value in a battery to "know" when to discharge the battery. You might get multiple cycles each day by being aggressive. So you need some ruleset. Why not "learn" what days the customer works from home vs. commutes to work?
Re: (Score:2)
The algorithm required to handle a home-scale UPS, even if you do more than just keep it properly charged while utility power is available and occasionally run a diagnostic test or two, is so basic and long-established that is seems like adding 'AI' anything is a massive waste of time and potentially introduces new failure modes.
My guess it is a gimmick for scheduling to optimize cost / ROI based on PV / weather / consumption / TOU / export..etc.
PV and TOU add a lot of complexity because there are all sorts of competing goals and prediction algorithms informing them. If you don't use your batteries you lose out on PV production during the day. If you use them too much you pay more in grid energy than you earned by exporting. If you use the grid at the wrong time it costs more than it would otherwise. If you export during period
Re: (Score:2)
Because static algorithms don't handle dynamic changes in parameters very well, e.g. compensating for suddenly your parents coming over an additional power draw, or being on vacation without manually changing the parameters to indicate as such.
You ask "why AI"? The answer is because marketing means we now need to advertise that the underlying feature has a trained model. In reality systems operating with similar optimisation / management tasks have been using "AI" for decades even if it didn't advertise its
Re: (Score:2)
Sure, the algorithms seem simple, but power is not.
You have a home battery, and you know how much roughly you use every day, and such.
But then you got things like Time of Use and vi
Re: (Score:2)
Trying to predict needs for a single household based on weather reports is folly. They're insufficiently high-resolution even when they're reasonably accurate. That's something that should happen at the VPP level, not the household level. The system should be maintaining a reasonable safety factor instead. Where ML can be useful at home is knowing what time you tend to take showers, and how much energy it's going to take to heat the house to where you want it.
Re: (Score:2)
The "AI" is probably just a simple algorithm that looks at when you typically use power and then avoids fully charging the battery until just before then. Phones have been doing it for many years now - charge to 80% overnight, look at when the user set their alarm for, and aim to be at 100% about 5 minutes before then.
Look Into House Batteries if You Haven't (Score:4, Informative)
A small generator (capable of charging the battery to full in just a few hours) is still useful for prolonged outages, preferably dual-fuel since you can store unused propane for a decade, unlike gasoline.
The peace of mind was worth setting up a second system for a close relative who lives a couple hours away.
The Luterra S2 is the first system of its kind to use Huawei's proprietary full-silicon carbide (SiC) architecture
Huh? What did they mispell?
BTW, no one tell Jeremy Clarkson about this; he'd have to leave the room [youtube.com].
Re: (Score:2)
The Luterra S2 is the first system of its kind to use Huawei's proprietary full-silicon carbide (SiC) architecture
Huh? What did they mispell?
Huh? I don't know what they misspelled but you misspelled misspell.
Re: Look Into House Batteries if You Haven't (Score:1)
Re: (Score:2)
It's not entirely marketing buzz, though it's comparable to Intel bragging about EUV... Yes, it's actually really cool, but almost nobody really cares in the least as long as it fits favorably on the price/performance curve.
Inverter (Score:2)
They're referring only to the charger / inverter circuitry, not the battery itself. SiC based substrates perform very well in power control situations.
Do they expect SiC to last any longer? In my research, I remember people noting that inverters can fail.
Re: (Score:2)
Everything can fail, it depends on how its designed. Failure isn't a given. There's countless battery/solar installations around the world with inverters which have been happily humming away for over a decade without issue.
That said in theory SiC MOSFETs are more robust due to higher thermal conductivity allowing them to dissipate energy through the case. They have junction temperatures upwards of 175C which is incredibly high. Typical Silicon caps out around 150C for practical purposes.
Re: (Score:2)
A small generator (capable of charging the battery to full in just a few hours) is still useful for prolonged outages, preferably dual-fuel since you can store unused propane for a decade, unlike gasoline.
That sounds very 2001. My neighbour simply ran an extension cord from their BYD Atto 3 into the house and called it a day. The 60kWh battery could keep them running most of the week.
Huh? What did they mispell?
Nothing? Are you talking about SiC? That's the latest hotness as far as high power efficient MOSFET switching goes. You know how all those small battery chargers are advertising GaN FETs? Yeah well this is similar in terms of advertising, except GaN is most suited for low power high efficiency, and SiC is more suited to high pow
Re: (Score:2)
It's not the most interesting battery tech at the moment, but it's good that another player is entering the market.
Recently an electric passenger aircraft made its first piloted flight (previous flights has been automated) with a 500Wh/kg battery, and even denser ones are headed towards mass production. It's looking very practical for regional transport services.
Re: (Score:2)
Because lady brains just aren't up to the task of knowing anything about electricity... /s
Re: (Score:2)
Not available in US (Score:3, Informative)
Re: (Score:2)
50/60Hz is a simple matter of changing a value or, if you're weird, a constant.
Meeting code requirements might come down to simple changes in software, depending on what they have now.
Re: (Score:2)
Re: (Score:2)
20 kW (Score:2)
Re: (Score:2)
They messed up and accidentally used correct units (KWh) once in TFS.
Was hoping for something different (Score:2)
I hoped it was a 20kWh module size; that is big enough for your average household to have full 24-hour backup with a single box, and three boxes would give large households ~40 hours of backup. That seems to be about what you need for a reliable and distributed, renewable-focused generation, with typical discharge levels around 50%.
We have 40kWh of battery in 3 modules, and the parasitic losses for each module really take a bite. Reducing module count would help there.
Re: (Score:2)
We have 40kWh of battery in 3 modules
Try using less power.
Re: (Score:2)
Easier said than done. The only conditioned space is the master bedroom, all lighting is LED, we have a heat pump water heater. That does cover our average daily charging for the car. We live in the "country" though, so refrigerators and chest freezer dominate the rest of the load. Replacement of the main fridge should save a bit, but it still has plenty of life left. ...then there is the IT load of ~300W continuous.
Why silicon carbide? (Score:2)
Re: (Score:2)
Low RDS(on). Means lower losses, less waste heat.
How reliable? (Score:1)
Re: (Score:1)
Re: (Score:2)
People are desperate to post their fanfiction with AO3 down huh?
Re: (Score:2)
Safe to say that it wouldn't.
Interesting (Score:3)
It will be interesting to see the cost. Not yet sold on SiC awesome tech but still a little beta/finicky.
What I would like to see on the market is a relatively cheap MPPT able to directly DC charge EVs without the string of DC-DC-AC-DC losses. Not too big of a deal for a 600V string of PV so I can't imagine it being that difficult given voltages are already in the range of what EVs will natively accept.
Obligatory cheap shot (Score:2)
Can they make batteries that last longer than their brake pedals?
https://carnewschina.com/2026/... [carnewschina.com]
Software (Score:4, Insightful)
No doubt these battery controllers use proprietary software with Chinese remote control. No thanks!
Give me one with open-source firmware, then I'll be interested.