Podcast
Join Viktor, a proud nerd and seasoned entrepreneur, whose academic journey at Santa Clara University in Silicon Valley sparked a career marked by innovation and foresight. From his college days, Viktor embarked on an entrepreneurial path, beginning with YippieMove, a groundbreaking email migration service, and continuing with a series of bootstrapped ventures.
Links
Follow Me
Follow Me
Join Viktor, a proud nerd and seasoned entrepreneur, whose academic journey at Santa Clara University in Silicon Valley sparked a career marked by innovation and foresight. From his college days, Viktor embarked on an entrepreneurial path, beginning with YippieMove, a groundbreaking email migration service, and continuing with a series of bootstrapped ventures.
Coreboot and Linux Hardware: Inside Star Labs with Sean Rhodes
Play On
16 SEP • 2025
0 mins
Share:
How do you design and ship coreboot and Linux hardware that actually works in the real world? In this episode of Nerding Out with Viktor, I sit down with Sean Rhodes, Firmware Engineer at Star Labs, to explore what it takes to build Linux-first laptops and mini PCs backed by open firmware.
Sean reflects on the early days of Star Labs, when a handful of engineers set out to create laptops that ran Linux reliably. What began as a side project grew into a full-fledged company with a reputation for supporting open source systems. Along the way, the team faced high tooling costs, unpredictable supply chains, and the challenge of meeting the needs of a niche but demanding user base.
As the conversation unfolds, Sean details how Star Labs survived the COVID-era supply chain crisis, why they moved away from AMI BIOS in favor of coreboot, and how firmware updates delivered through LVFS became central to their long-term reliability strategy. He also explains why some features, like fingerprint readers, failed to deliver real-world value.
At its core, the Star Labs philosophy is about empowering users with secure, reliable, and open hardware. From haptic trackpads to distro-agnostic support for Fedora, Qubes, and openSUSE, Sean’s perspective makes clear that Linux-first machines can thrive outside the mainstream PC market.
Transcript
Show/Hide Transcript
The code base they use is about 10 years old and it's this sea of hacks and you sort of look at it and it's just like, how on earth can this work?
Right.
But the surprising thing is it does and it works very well.
So you get this 10 year old code base and you can't work with it at all, but it will just work.
Yeah, it's brilliant in that regard.
And then you get coreboot which is, you know, does everything to the specification.
If it doesn't, you know, adhere to the specification, it's wrong.
We're not going to do it.
Welcome back to another episode of Nerding Out With Viktor.
Today I'm joined by Sean Rhodes from Star Labs.
Welcome, Sean.
Hi there.
Good to have you on.
So this is going to be another Linux on a desktop, Linux in the supply chain, Linux in general kind of episode.
So I'm really excited to have you on.
You guys have been in this space for a while now and maybe let's start with the most basic things.
Who's Sean and what?
Star Labs?
Let's start there.
Yeah, so we've been going eight years now and it was literally just a bunch of us wanting something to run Linux reliably, 100% selfish reasons.
And it sort of evolved there, you know, in sort of, you know, Core Boot with all the licensing and the goodness that comes with that.
Yeah, you know, a bunch of other things.
I mean half the things we do are because people we asked ask us to do them.
So we've kind of just gone with the flow.
But yeah, I primarily work on the firmware side.
Yeah.
Corporate stuff.
So.
Yeah, good.
I mean, let's talk about Core Boot.
Right.
Because coreboot, I've had an episode of coreboot, had people on there about that.
For those not familiar with Core, but or people have been watched that episode, it's obviously important to you guys in terms of your blueprint for your of machinery.
What?
Talk to me about what?
Coreboot.
What's coreboot and what should people know about that and why it's important?
Well, I mean, aside from all the things it does actually really well, it fits quite well with Linux, obviously.
Kernels, GPL, 2, Corbo, it was used to be called Linux BIOS.
So yeah, it just fits well with sort of the ethos being open source and then if you take all the licensing and stuff out of the equation, it's good.
Yeah.
I mean the code quality is very high, things are very fast, very efficient, it's very easy to maintain Cross boards.
So when we're doing things here, you know, we're doing an update for something that we made three months ago that's going to apply to something we made eight years ago and it just makes maintenance a lot better.
So.
Yeah, and translates into the end users who've got eight year old machines getting updates which is.
Yeah, yeah, let's talk about that.
Because obviously people, I guess people don't really think about biases too much.
Right, let's, I guess let's preface it.
But Core Boot is a bias.
Right.
So let's start for those not familiar with that.
What in terms of what in terms of the trade offs and like, in terms of like getting this up and running, delivering Core boot, what have you been experienced with that?
Like, because that's obviously it's diving into the deep end of the pool rather than just getting something off the shelf from AMI or something.
What has your experience with that in the supply chain world?
Well, yeah, so I mean AMI love and hate relationship there.
I mean like the code base they use is about 10 years old and it's this sea of hacks and you sort of look at it and it's just like how on earth can this work?
Right.
But the surprising thing is it does and it works very well.
So you get a 10 year old code base and you can't work with it at all, but it will just work.
Yeah, it's brilliant in that regard.
And then you get coreboot which is, you know, does everything to, you know, the specification.
If it isn't, you know, adheres to the specification, it's wrong, we're not going to do it.
It doesn't matter if the device doesn't work, but you know, it's matching the specification type thing.
So when you're going for a new platform, you get coreboot work and you do your, you know, our sort of vendor code side of things and then the common directory usually doesn't work because it hasn't been, you know, that tested in real world things, you know, with devices that don't comply to specifications and all sorts of things.
So I guess the experience with coreboot is the to generally get it working to the specifications is actually quite easy compared to omi and then you have to do all the sorts of tweaking afterwards.
Whereas AMI is the other way around.
Get it working for your stuff is harder but then all this hacky old mess of stuff will just magically work on anything.
Yeah, that's the thing.
It's the Perfect theory versus the pragmatic kind of.
It kind of works but nobody kind of knows why it works.
Yeah, definitely it.
Yeah.
Because if you poke around because I. I got one of you guys boxes here.
This is the bite, I believe.
Yeah, yeah, the bite.
And that's a good example of that is.
Well I used.
For the viewers who haven't seen it, this is the bite by two.
Is it byte one?
I actually don't know.
That looked like a two, the N201.
So that was the fanless one and then we're on a. Yeah, three.
Now I'm scared.
But a good example of coreboot there is something like Secure boot is actually rather challenging to do well with core boot.
Right.
So what is your experience been with that?
Right.
To get.
Could that get that working right now with.
With secure boot and like getting a full featured bios I guess instead of just that super sliver custom use case like Chromebooks.
They're very narrow scope of what they need the BIOS to do.
Right.
Versus off the shelf commodity PCs.
Yeah, definitely.
I think it depends on where you sort of where you're viewing as getting it working.
I mean me and Matt, I think you had on for your Cogbook topic are both very much in the mind of working is when it's been upstreamed and it's been reviewed by multiple people and that's, you know, your perfect working state.
Yeah.
Which when you think secure boot you're getting patches into core boot side and you're also getting patches into EDK2 side.
Whereas EDK2 is less focused on core boots implementation and more on slim bootloaders.
So you've got a whole load of complexity and just upstreaming it in terms of getting it working, it's not horrific.
Generally most of the challenges lie in getting core groups to talk to EDK to.
Right.
But yeah, I think the core boot secure boot patches were sort of faffing around for about a couple of years sort of in various states and then they finally got merged.
I think it's about a year ago.
Okay.
And yeah, so that's.
That's all working nicely now.
And then the next stage is get to enrolling the keys automatically upstream which I mean that's going to be another two years easy because.
Yeah.
I mean, yeah, pulling from other repos and stuff in EDK2 is complicated.
Yeah.
At least.
Interesting.
I'm curious about where your head is at with regards to.
So obviously the bias is still in the whole like EFI kind of world mentality.
Worldview Right.
But then you have companies out there like Oxide who says, you know what, screw bias.
Well, not screw bias, but screw bootloaders in general.
And just like they bypass that entire step.
Like what is your view on like how that adoption is going to drive beyond like the extreme edge case that is I guess Oxide.
I think it will stay as an edge case.
I mean over the years played with different payloads with and you know, if your end goal is to run Linux, you can get core boot to start the kernel straight away.
Yeah.
And there's a whole load of simplification there.
There's obviously an armada of drawbacks.
You know, Windows, forget about it, which is still a chunk of our user base who dual boot or have to use it every now and then.
You know, dual booting, updating is just a nightmare and then you get incompatibilities with firmware matching and things like that.
So it's really cool as something you can do.
But I don't think it's ever gonna evolve out of, you know, someone building for themselves realistically or buying something super specific.
Yeah, it'd be nice if there was.
But yeah, I mean I guess the Chromebook is a good example of something like that.
It's a tailor built machine that is building a tailored os, I mean Chrome os.
Right.
Where they can strip all those components off and just like remove I guess entire attack surface from like the buyers as well.
Right?
Yes, yeah, definitely.
Yeah.
I mean, yeah, I think the depth charge that they use is probably a 50th of the size of UDK2.
There's just a lot less to work with and go wrong.
Right, right.
All right, let's get out of the rabbit hole that is Core boot that I'm happy to explore further.
But I think for the Saturday of our viewers, let's move on out of that particular pocket and talk more about STAR Labs.
Let's talk about your products.
Obviously we showed the byte 2.
Rather talk me through about your experience so far building hardware from like where you started to the byte 2, which I believe is the last product that you have shipped.
Right.
We're Byte three now.
Oh, sorry.
Okay.
Yes.
So, well, way back when we started off doing the clevo thing, same as many other vendors.
And yeah, we, Clevo is kind of a mixed bag.
They're very hard to fault.
But we wanted something that you know, you could sort of like, you know, a little bit more sort of aesthetically pleasing and you know, tailored so we set about building it and I think the issue you hit first with building hardware is the tooling costs being so high, you know, compared to you know, Apple and Dell when you know they've got £300,000, you know, to play in tooling but then they're splitting it over, you know, hundreds of thousands of models and when we're in a niche already and then we keep partnering into a niche because that's what people ask us to do, then.
Yeah.
Becomes quite costly.
Yeah.
So yeah, it's, it's a balance.
Yeah.
Because a lot of people say well you know, one example would be can you put a charging port on either side of your chassis?
And that's one of the things that's tooling costs.
The actual production cost is infinitely small.
But when you're talking most people will predominantly right handed, they plug their charger on the left for the small part who want it on either side.
Is that $75,000 split over however many machines.
Is it worth it reps?
You know, interesting.
Apple do on the left.
Yes, they do.
Yeah.
I, I think that is literally, you know, going for the, you know, the masses.
Everyone's right handed.
Most people are plugging on the left.
I guess that's the default.
Yeah, yeah, some laptops do have it either side.
But yeah, it's one of those things where you're just weighing up pros and cons of, you know, costs and things like that.
So you started, I mean so where do you even start?
Right.
Because you going to build something from scratch today is like I said, tooling cost extremely expensive.
Right.
Like your moqs.
Because I mean most vendors who do, I guess white label ish PCs, like if you look at the PC market there are very few people that make PCs.
Right?
Very, very few.
Most of them are customized nukes or like a nuke board with some custom casing.
Like very few vendors out there would go and say I want to build a board myself.
Like I guess the hyperscalers would do that.
But like for the vast majority, like that's extreme rare.
But that's the approach you guys have decided to go down, right, building, tailoring things from basically zero and build it from that ground up.
And that's playing in the hard line, hard lane for sure.
Yes.
Yeah.
So when we first started it, we're very adamant.
We want to make it, we don't want to share anything.
And then sort of reality dawns, you know, when you get to Something like a battery.
So if you.
It was a, it was a topic on our laptop Mark iii.
So you're probably talking five, six years ago we had a chassis and we could put, I think it was a 52 watt hour battery in it.
But the tooling costs which weren't horrific, I think they're about 40,000.
But then you get all your certifications, certification certifying an individual battery because of explosion, fire, blah, blah.
Yeah, it's insanely high.
So we ended up using a off the shelf 45 battery.
So yeah, that was one of those things where yeah, we could have done a bigger battery if it was Apple or Dell, they probably would have done.
Yeah, we didn't because it becomes inefficient.
But so do you do all the PCB design yourself and then you work with.
I presume, I presume you don't because you're based in the uk.
You don't do manufacturing in the uk.
Right.
You do them in, I presume in China like most other people do.
Right.
Or like.
Yeah.
So yeah, the presence in the UK is finishing.
It's you know, pop a PCB in, burn the firmware, you know, screwdrivers, nothing, you know, no big machinery.
Yeah, most parts are sort of China or Taiwan.
It's one of those really strange things we follow around the globe because you know, you have process cases coming from the States and you have parts coming from Taiwan and Japan and stuff and it comes up to us.
So all the countries kind of.
Yeah.
Don't make too much sense in all that regard.
And that's, and that's what when people say bring home all the manufacturing when the do this, you realize that it's near impossible to do it.
Like even like a company like Apple is struggling greatly to bring manufacturing to the US because the tooling that you need to do a PCB is like there were like a handful of factories on the planet who could do them maybe from like they can.
There aren't that many that can actually do all these like lower level components.
Right, that's right.
And then, I mean less of an issue for Apple but more for us.
We'd be having to pay for that tooling in full.
Whereas you know.
Yeah, put it simply, we're just renting a machine to you know, make things.
So yeah, it's.
Yeah, yeah.
The, the age old.
I mean because the way we do it, we are technically able to say, you know, they're made in the UK because of the way they put Together with different bits and stuff.
But we don't.
Because.
Right.
Seems deceptive because of the amount of stuff that in China, the States and Taiwan and wherever else.
So it's.
Yeah, yeah, one of those.
Yeah, exactly.
In your journey so far, what has been absolutely the most hardest thing to do in terms of building your supply chain from zero and obviously compared to Apple on a shoestring budget.
Right, yeah.
And also playing the hard lane, like not going to know DM and say you put this board in different case.
I think it's been getting the stuff, you know, stuck rolling because you know we sort of build things and you know the MOQs are big some things, you know, buying 5,000 off, sometimes you're buying a thousand.
So you've got to line all these up.
Some, you know, bits are put together in one place and then they move to another place to put in and you've got to ship them over in time.
So it's literally getting a consistent supply stock that is possibly your hardest thing in the world.
And also you factor in that our stuff takes us longer to make.
You know, I'm sure Clevo can churn something out in probably about four weeks.
I don't know what it is.
It's now but you know, with us it's sort of, you know, an eight week minimum.
It's usually closer to 12 weeks.
You know, from once we say right we're on these bits to be together so getting all of that together whilst also, you know, keeping up to date in 12 weeks time, you know, there could be a new processor or whatever else it is.
Aligning all that.
Yeah, consistent stock is, I mean this is almost impossible.
You know, we've been doing this eight years and we've I think probably twice at everything on our store in stock.
Once you, once you get the stock consistent and then you shipping them out and it's all nice and stable, then you get more orders and then that's how route gets.
Yeah, you also have to like, you also have to do some fair bit of forecasting.
Right.
Because you don't know exactly.
I mean particularly if you have a product line like you don't know exactly what's going to sell when and so forth.
Like there are a lot of dynamics that are difficult to predict.
Of course.
Yes, yeah, definitely.
So looking at that experience and like timing obviously has been very difficult.
But right now, so right now you make the borders in the uk, in China or Taiwan, ship them to the uk, assemble things in the uk, you flash all the firmwares in the uk, I presume that.
So that they are fresh up to date and they dispatch from the uk.
Right, yes.
So that's.
I.
Because that's the thing that at least when I started doing supply chain at screening, when we started buying machinery.
Machinery, my buying software from odms in China.
Firmware, sorry, buying device from China is the firmware.
I think a lot of people underestimate the importance of that because when you speak to ODMs in China, they view themselves as hardware vendors and box pushers, not software vendors.
Right.
And I think that mentality is so different.
Right.
I'm curious about what's your experience.
But like bias firmware is one of those things.
They're like, oh yeah, it's a five year old bias, like who cares, it's working.
Yes, finally.
Yeah, I mean we've done some tests.
I mean intel stuff, you roughly follow reference design.
So there'll always be a version of AMI that will boot and it's designed to, you know, not be efficient, but will boot on similar ish hardware.
So some, when we get our stuff, you know, when we're first doing it, you'll chuck this on it just so you can actually check, you know, things are working and efficiency is horrific.
You're talking sort of, you know, if you're like a 25 watt processor laptop, you'll be getting sort of, you know, two hours battery.
Whereas once you get Corvette One and everything actually configured you know, 12 hours battery.
So it's just, it's bad.
But yeah, I mean a lot of the sort of, you know, white labeled restuff, the stuff on that, it's never touched and it's just, it's not optimized because they just use that reference thing change, you know, logo and name and they think chip done.
Yeah.
And I mean, I think at least for me the big wakeup call for this was when logo fail and pixie fail happened.
Right.
And that's when you like that was for the first time In probably a decade I started thinking about bias again because like if you live in the Mac ecosystem, like it's a black box, but it's kind of like it's firmware.
Everything's managed for you.
Right.
So you never think about components.
Right.
Yeah.
Whereas.
And when that happened you're like, oh crap, how do we do this?
And that's when you realize the lack of sophistication most ODMs have when it comes to Getting these out, these firmware updates and bios updates and lifecycle of that was at least for us, that was a big insight into like why we need to move our supply chain away from those ODMs.
Right.
I'm sure it's about your experience there because I would imagine you have had similar conversations with a lot of these ODMs and OEMs and manufacturers overall.
Yes, yeah.
I mean our sort of engagement on that side has sort of died down because yeah, we used to do the.
Probably about four years ago now we used to offer AMI alongside coreboot.
That was when the sort of coreboot common features were lacking a bit with all our AMI licenses.
So we take that reference and actually make it work and provide updates for it.
So yeah, we haven't really touched, should we say their version of it for a very long time.
Right, right.
For numerous reasons.
Fair enough, Fair enough.
All right, interesting.
So talking about laptops and desktop was on laptop and I guess mini PCs, I guess it's like proper word for these devices you're shipping.
How different has that been like in terms of like building a laptop from scratch versus building a mini PC from scratch?
Because I obviously, at least to me a laptop feels like a lot more complicated in terms of like the component that goes in there, obviously the monitor, but far beyond that 100%.
The, the byte was, I mean it was a walk in the park.
We kind of approached it expecting it to be comparable to a laptop.
And then you know, once you do it, even your packaging simpler because you haven't got a screen to protect, you haven't got batteries, you haven't got, you know, certify your packaging and even things like that.
And you know, there's no display calibration, no built in speakers.
So I mean it was so much less work than we anticipated because we thought it's basically another one will, you know, save a bit of time here and there.
But that was yeah, so small.
I think the new Starlight's probably been the most, the trickiest one we've done that sort of, I mean that has been a rabbit hole on sort of doing things because, you know, so we had a, we had our 11.6 inch starlight, which is a classic clamshell and it basically became impossible to make cheaply.
You know, when we first started making them, they were 300.
I think they're 300 pounds, including VAT.
So American US dollars, it's $320, something like that.
Very cost effective and it was growing and then it hit a point where, you know, no one was making screens that small.
You just couldn't get, you know, stuff around the right vultures.
The processors had moved on to a more powerful sleep.
See an N100 generating too much heat to them ashes.
So if we basically continue making these little things which were desirable because they were cheap and small, they would have been 7, 800 and it was just a.
Things would be fun.
That's.
Yeah.
And that's the problem with doing like niche hardware because MOQs rule everything.
Right.
And it's so difficult to do niche hardware and compete with the big guys.
Right.
Because you simply do not have a fighting chance really to compete on price with the big guys.
Right.
Because they are shorting out so many large volume.
So like how you.
Obviously you have your niche, but what is the.
I want to go back to the laptop conversation a second.
But you get kind of queued up that.
Because I'm really curious about like how you position that.
And obviously you have your user base.
It's obviously far more tech savvy and all that.
But.
But how do you.
What's your rebuttal to.
To those.
That pushback?
Because obviously, yes, you could shrink your margins, but you still have to run a business.
Right.
So I'm curious about how.
How that, how that those conversations have gone.
So I think.
Yeah.
I mean if you take that one, the Starlight example, it was.
It's not viable to do anymore.
So we essentially turned it into a tablet.
A slightly bigger tablet.
Yeah.
Because.
Yeah.
I mean, sort of going against that curve.
If we wanted to build a screen because no one was building them anymore.
You're talking, you know, hundreds of thousands of toys.
We've never had a screen made for us.
Yeah.
It's just.
Yeah.
Unless you had, you know, a bunch of people who are so set on having that product which, you know, that's a niche instead of a nation.
It's just.
Yeah.
On any scale to make it viable.
Yeah, exactly.
And in.
In the laptop right now and I guess in.
In the.
In the bite as well.
How much off the shelf components and how much is like custom made?
Custom made?
Like do you make from scratch?
Like beyond software?
Of course, Breakdown Bytes probably got a few more, I guess, open tooling things.
What we've done with quite a lot of our things is we've used.
So rather than making, you know, the top cover of the bike from entire scratch, find it and make the changes we want to make it, you know, nice.
So I think the bite is probably Closest thing to, you know, stock available designs, parts available.
Yeah, I mean, I think, yeah, we have sort of evolved to more of that approach than anything.
It depends on the model.
I mean, so the style, like the new touch one, you know, we built the keyboards ourselves from scratch because everything we found was rubbish.
So.
Yeah, it is.
Yeah.
It's weighing up.
Yeah.
Cost and how good it is.
Staffizer.
We made literally everything on that Starbuck.
Now we've got.
Yeah.
Off the shelf speakers, batteries, off the shelf switches.
We share with someone.
We're not allowed to say who.
We share with them with one of the big boys.
Yeah.
So yeah, it's very case by case.
If you were to embark on the same mission again to do.
To do the laptop again to the Starflight again, what was your biggest learning of like, oh, crap, we spend so much money or so much time on this one thing that if we only knew this, we'd have saved us a world of pain.
Is there any particular thing that springs to mind?
Fingerprint readers?
Yeah, it wasn't crazy high.
I think it was about.
I think it was about say, $19,000.
Put a little fingerprint reader on the trackpad and this was.
It started because people ask us, I want a fingerprint reader.
Sure.
Catch 22 is your trackpad alignment is much harder when you've got a fingerprint in there because you can't just use the cutout in the input cover.
So there were failures during production.
Obviously this is just cost.
It's not actually.
And then you get to the people who've asked for the fingerprint reader, but then realized that in Linux all it's going to do is log you in.
You still got to type your password in to unlock your keychain and therefore it's pretty pointless.
So then people ask us for an option.
Can we disable the fingerprint reader?
So it was like, oh, okay, that was pointless.
I think we kept them in, I think two generations and now they're gone.
And yeah, I highly doubt we'll go back unless something changes on the next slide.
Interesting.
In terms of Linux distros.
What.
Where's your stance on that?
Right, because I know, I mean, system 76, they are a bit more opinionated.
They have their POP os.
Where do you guys stand on that?
How do you see that shaping up on your end?
So we try to stay agnostic.
We don't want to have a favorite or.
Or you know, say this one.
And we don't want to build our own.
We want to stick to our focus, which is the hardware and firmware we've essentially, if there's.
If people ask us to offer one installed, if we can do it and there's enough demand, then we'll do it.
So yeah, cubes popped up because of that.
Fedora will be available soon.
Then there's one other list leap is coming up.
So yeah, really it is.
As long as it people want it and it works and we can install it, we'll put it on just because in terms of, you know, if it'll actually work, that's more down to Linux Excel than the distro.
Yeah.
How it looks doesn't really affect, you know, capabilities of a computer.
And speaking of that, in terms of when, obviously hardware enablement in the Linux kernel, I guess when you pick components like WI Fi or Bluetooth, whatever it may be, do you always go for things that work out of the box or do you go down the path of oh crap, this is not supported, let's patch it up ourselves and contribute upstream.
How do you see that for like enablement?
I guess so we've always done it as we build it for what's working okay.
In the kind of.
We've never tried to add support for it for a new device just because, I mean even also say we did it now, we did a new sound card or whatever it might be, you had support for it, you know, and then, you know, people still want us to, you know, install, say Zorin.
I think that's the oldest thing that we've got in there that's still running 6.8.
So we had support for a driver in 616.
You know, we're not going to be able to offer a Zorin on that for probably another four years.
And then you get to this all versioning things like that, whereas something's already supported, chances are it's been there for 10 years.
So you can use any kernel version.
It's just simpler.
Yeah, no, that, that's fair enough.
That's fair enough.
And I mean it's.
Yeah.
Because you can't really control that ecosystem at all really.
Particularly if you look at upstream.
Right.
So it's.
Yeah, it makes a lot of sense.
We talk about what's.
What makes hard to make a device from scratch.
But I guess let's go back to the horror days of COVID and I'm curious about how you guys lived through the supply chain crisis that impacted everybody in that world.
Right.
In the entire hardware world.
Like how did you get out of that alive?
Like how, how was that I'm curious about your firsthand experience of dealing with those crisis.
Yeah, fun.
So, yeah, so we fortunately right before COVID we had just had, I think it was two big shipments.
So it was kind of good timing for us.
So we had a bit of inventory to keep us going for a bit.
But then after that off that, you know, everyone's prices from all of our suppliers, you know, went up.
There was not the figures of the, of my head but we, yeah, Soundcard, I think it was like 1, $1.90 we used to buy them for and in CO it was 30, $32 or something like that.
So you know, some components were just.
That's hard to pass on through the roof.
So yeah, I think we swallowed some of them and bumped the prices up for others.
And then lead times became really long because production lines couldn't be as efficient and shipping was congested.
And so essentially I think we made it through it fortunate timing because we just had a delivery.
And then also the fact that everyone was in the same position.
No one could, you know, get anything new.
So everyone was very understanding about things taking a lot longer and costing them more.
So I think it was a. Yeah, luck.
Luck, yeah, fair enough.
And our customers and speaking about that a bit like what.
How.
Who's.
Who are your customers today?
Like.
And like is it mostly consumers or they've been picked up by enterprises more?
What, I mean what do you, how do you kind of navigate that and.
What do you focus on?
We don't focus on anything.
So we don't, you know, target one specific group, you know, like Profile X or you know, this type of company.
It's literally as random as anything.
You know, we've got companies that everyone knows the name of to teeny tiny companies who map out manholes and things like that.
To someone who wants to use cubes because they want to be private, to someone who's annoyed with Microsoft.
It literally is your biggest, most diverse customer base you'll ever get.
I mean sometimes we even have people pop up sort of, you know, Linux developers and that, you know, say things that, you know, just you guys over our head because you know, they know more about Linux than we do.
Right.
So yeah, it's insanely diverse.
There definitely isn't, you know, this is our key person is.
And in terms of volume, can you speak a bit about like how many units you've sold or how many you're selling per year or like any order of magnitude numbers we're probably close.
To 10,000 at the moment.
the number in total or per year?
Yeah, since we started.
Don't know.
Yeah, I'd have to check.
But yeah, most of the quantities come from the Starlight is definitely our best seller at the moment.
Okay.
Yeah, that's, that's a lot.
It's quite hard to make enough.
And if you compare you guys with obviously we have, I guess in the last decade, I say we have.
We had system 76 kind of done this guys, you guys have been in this game framework.
Seems to be getting a lot of, obviously, I guess slightly more recent.
How much collaboration is there between you guys in terms of like raising the bar for Linux on the desktop?
Is there any at all or.
Not really.
We don't really engage with many of the others.
I think you cross paths with System 76 occasionally on A, you know, core book patch that's being upstream.
But in terms of anyone else, I haven't really.
We don't really run into them, so to speak.
I don't think, you know, we work on the same things.
I mean, me and Matt mainly work on, you know, upstream core book and that's why they're.
Right, yeah.
Smallish people, smallest group of people who work on it, which is pretty much, you know, me, Matt, a bunch of Google engineers and then, you know, seam into there and a bunch of engineers and then I think most of the others primarily work on downstream forks.
So there isn't that much crossover.
Yeah, it just doesn't really come up.
That's, that's fair enough.
I mean, and do you, what do you.
I mean, obviously coreboot is close to your heart and like, where do you.
Yeah, what.
I mean, do you see a world where it will become more mainstream, where it becomes a viable real alternative to efi, to AMI and the likes of.
And Phoenix and all those guys.
How do you see the maturity of that?
I don't think it would be corrupt because coreboot is quite stuck in its ways.
Not in a bad way, but stuck in its ways for sort of, you know, mainstream vendors to pick it up and it become approachable.
I think an open source alternative is possible because AMI dropped their open source one, I think it was a couple of years ago.
That was a very streamlined EVK2 base.
I bought that.
Or slim bootloader, I guess.
But I think those could appeal to, you know, someone like Dell or something like that because you get the benefits of open source, the shared platform, you know, code management.
But they're not going to be completely learning how to do things again because yeah, I mean going from AMI to core boot they are entirely different.
Whereas AMI to slim bootloader is still really K2.
So it's pretty much the same thing.
Okay, so that's where you would you think, do you see the future of CoreBook more being in custom tailored embedded vendors like yourself?
Is that purely plus?
I guess Google with the Chromebooks, is that this the future you see of it or.
I guess so.
If you say, if you say not going to go mainstream.
Right?
Yeah, I don't think it will for you know, computers in general, mini PCs, laptops, etc.
Not overly okay on the topic, but I seem to remember reading that Tesla used it in their cars.
Oh.
Whether that's right or not, they've got, definitely got a fork of it.
It's popped up.
But I think that sort of implementation I guess could grow because I mean if you think if you want a lightweight fast bootloader with no junk in it, cool boot is, it's the dream.
Yes.
I think you better have a fork of it as well.
So I could see.
Oh yeah, that, that, yeah that doesn't surprise me because I, I know there are people in the corporate world from, that has been in at meta before.
Right.
So that doesn't surprise me at all.
Interesting.
So yeah, I mean I, I, I, I'm really bullish on corporate.
I would, I would love to see our devices running on co at screenly but we're not quite there yet.
Right.
Maybe there's a future scenario with that because I mean I, I like the lockdown approach.
Right.
We're like slim as possible attack vector because I think there are more like we've seen, we have seen more attack vectors against biases and I think that's just going to keep coming.
Right.
So I don't think that's, that's hardly over.
Right.
Yeah.
Cool.
Let's go things back.
I want to go back a little bit to your origin story and kind of talk a little bit more about that.
You mentioned clevo before for people not familiar.
Let's, it's come up before on the show but I don't think we had a proper explanation.
Let's start there.
What is clevo?
And just for the users who are not familiar, I know we've already mentioned it so probably worth it as at least a brief explanation of what that is.
Sure.
I mean my obviously experience is seven years out to date but as far As I know they exactly the same thing.
They don't sell their things directly.
You don't go into a shop and buy a, you know, this is our laptop one whatever.
They are primarily designed to be white labeled.
So seven years or what ago they have a chassis and then they've built it to be configurable.
Configurable to whoever's ordering it.
You know, did you want that 50W hour battery or do you want a 30W hour battery and two SSDs?
And it's all designed so people can order it in their implementation of it.
And theory is people buy these, stick their logo on it, you know, choose option lists, buying a new car, sort of type list of things you can do and then they don't have to, you know, pay for tooling, pay for thousands at a time.
All the other complexities that come with.
And do they blow fuses on the boards and patch them and like.
Or do you have the ability like patch on bias on that, put your own bias on there or the.
Way it was back then is that you did maintain their own internal firmware so you could get updates for I think it was six months or something like that from their AMI firmware.
Okay.
But you couldn't.
You'd have to, if you wanted to do it, you'd have to port it yourself because you have to do your license independently.
Right.
With ami, which you have to pay per board, whereas they're already paying for it.
So it's a bit complicated.
Got it.
Okay, I got two more things I want to cover with regards to like Linux on the desktop, obviously, which is a big topic, but the first one is you mentioned firmware updates.
Right.
how do you go about that?
We've, I've had Richard Hughes on before talking about firmware update.
how do you like, how do you make, what do you make of the firmware update life cycle on Linux today and how.
What do you guys suggest for your customers to do?
Firmware updates.
But particular over there, right.
If you're talking about larger fleet deployments, enterprise deployments, so forth.
Right.
How do you, what do you guys.
Recommend for those that so.
Yeah, I mean we are well worked with, spoken to Richard for a long time.
Yeah.
The preferred method is Firmware Update Manager.
It's built into pretty much every distro now and generally speaking it works.
We've seen whole use of it has tailed off a bit because there's been a lot of issues especially on Arch because they build and push things so fast and especially I think, you know the lockdown side of things and the kernels been dialed up compared to where it was a year ago and lots of things have stopped working.
So yeah, it's.
It's the preferred one.
It's wicked.
You know, you upload something to the LVFs, it'll run tests on it, check, it's fine, it'll get assigned by us, you know, it'll be checked, it hasn't changed whilst you're downloading it.
I mean, there's a long list of benefits and yeah, up until recently it works on every distro.
Yeah, a few hiccups I think, which will probably smooth over in the next six months, a year, hopefully it's back to working everywhere.
And you provide your bias updates in the lbfs?
Yes.
So you can just pull them down from there?
Yes, that's right, yeah.
And that works for coreboot and for your aima, because I believe you still ship with AIMA as well as an option.
No, no.
So that's all gone.
So, yeah, in now we've got a couple updates, your SSD updates.
Because we've been working with Lexar quite recently.
Okay.
With Fison and yeah, Lexar have actually been really wicked about tuning their SSDs to work well with core boot.
A bunch of extra timings and stuff.
So we've got our own firmware for Lexar SSDs, touchscreen updates, keyboard updates, hub updates.
Pretty much everything we do is in there if we've had to update it.
Oh, very nice.
Yeah, Richard is absolutely wicked in that regard because were speaking to him about how to update their keyboards and the new lights and I think almost word for word, his reply was, I accidentally wrote you a plugin.
Try this out.
Here's a firmware file and yeah, it just worked brilliantly.
Yeah, were pushing updates out for that for nine months or so.
Amazing.
That's really cool.
Kind of the last thing I wanted to cover is Linux on desktop 2025.
What are the biggest shortcomings?
Is it still in power management or.
Where do you see the biggest weaknesses in.
In Linux on a desktop today?
Compared to alternative ecosystems?
I guess.
Tricky one too many to be honest.
I don't think it's too bad because, yeah, so we as a company, we've grown massively in the last year and so we put on people.
When we started it was everyone used Linux, so it was never really a question of what operating system are you going to use.
Do you know how to use Linux?
That was just a Given obviously we've had to bring on people who have skills in other areas rather than Linux.
Right.
And each one of them has been thrust a laptop with Linux now and we haven't had any sort of internal support questions if you like or anything else like that.
Yeah, I think you get the occasional one of I can't.
I think Apple music people aren't into the Linux ecosystem like to use that and things like that.
So I don't think there's any one like major drawback.
I think it really is quite solid now.
Okay.
Obviously that's me saying it.
He's used Linux for a long time.
So yeah, I'm going to be not the most neutral in that regard.
But yeah, because historically power management has always been like heels heel.
Right.
Like I just even to this day like people like they don't hibernate.
They just like when I'm done my laptop gonna shut down before I put my bag.
All right.
Which is I left the Linux world as my daily driver many years ago in favor of the Apple ecosystem for various reasons.
But like that was one thing that I remember vividly the first time he's like wait, it just hibernates and it doesn't drain the battery when it's hibernator.
That's, that's one of the things that are still to be fair shocks me and people will sort of, you know, sending a message and stuff like that.
It's like do you.
Does suspend you work when you use Ubuntu on your laptop and you're just like yes.
You know, it's almost shocking that people think that it doesn't work with a laptop that's designed to work for Linux.
I mean if you can suspend it's.
Something's broken.
Yeah and I think, yeah, I mean a lot of that comes from people buying stuff that wasn't designed to.
It probably hasn't been updated in years and yeah, it doesn't work.
So yeah, it's yeah, I think yeah low expectation bar from quite a lot of people.
So there's nothing that you still struggle with that you would like to see greatly improved?
Is it just relative parity across compared to everything else or there must be some areas that you feel like this area really would need some love.
Right.
I mean, yeah, I think I'll be biased by recent experience but I think sort of more sort of continuity and testing and things like that.
So we've had an absolute wave of bugs on the outside recently and you know, part of that is because of the best thing about archers.
The worst thing is because if it's newer, you know, they'll include it, you know, just as a, a rule.
And so you get a lot of bugs slipping through.
So I think, yeah, just, I think a little bit more continuity and testing would probably do it well from bugs and things like that.
But, you know, it still is using arch as the example.
The best and worst thing, if you find an issue on it, they'll fix it in a couple of days.
Yeah.
So it's definitely not the end of the world.
Interesting.
And on that note, on a desktop, how many proprietary blobs do you need to run your devices today?
I would imagine not everything is open source that runs there.
Right.
So talk to me about that.
How does that look like a landscape today?
So everything intel based, we'll keep it simple.
You've got a descriptor, which is a blob, but it's so tiny and coreboot can modify them that no one's ever bothered to change them.
So you've got the descriptor, you've got the me, you'll likely always have the me.
And then you have fsp, which people like us and everyone who works on corebook has source code for, so they've never bothered to try and attack.
So in terms of the ones firmware wise, people aren't too happy about is the me.
And yeah, OS wise, you got wireless card, which.
Wireless cards have had blobs for donkeys.
Yeah, I mean, some people cling to.
There was a brand of wireless cars that used to put blobs on the chip and everyone thought was brilliant because you could use a distro that didn't provide blobs.
Right.
Negating the fact that it could never be updated.
Pretty big oversight.
Athros.
Athros.
I want an Athros card because I don't want my.
I want a blob for eos.
And it's like you could technically do it, but then you just got an outdated blob that's burnt into your card, so it's kind of defeats the point.
So, yeah, wireless card, I mean, I guess they're the ones that people don't want, but are they out of requirement.
And I guess Bluetooth included in that?
The wireless card?
Yes.
Yeah, I guess that's a.
That's one.
That's a dual function card, I guess.
Yeah, yeah.
I mean, to be fair, they are two separate blogs that the distro throw either.
But yeah.
Amazing.
So, Sean, what lays ahead for STAR Labs in the next 12 months.
What are you guys focusing on?
What should people be on the lookout for?
It almost make it sound a bit boring.
It is more of the same.
It is just evolving down the, you know, features that people have asked us for.
The biggest one is people asking us to make the haptic trackpad available on a smaller laptop because we've done a fighter.
And people go, this is wicked because the only of Apple have done this, but I don't want to spend three grand on a laptop.
And that was theory of the fighter, that we make something that was.
Had everything on it, didn't look at the costs.
So, yeah, we've got that coming to a.
To a cheaper model, which is very cool.
And yeah, it really is everything.
Sort of just filling out requests, more kill switches, camera filters, things like that, sort of going more into the.
It's almost going more into the niche which is now becoming more mainstream because people are becoming sort of privacy away 100.
Interesting.
Good.
Well, is there anything you want to wrap up and talk about before we hit the report?
Anything people should be aware of otherwise, beyond the roadmap, things that people should be.
Well, learn more about you guys or anything else you want to do a shout out about?
Not off the top of my head.
We've got.
We've got a few bits working on firmware side of things which will be very.
To come out.
So just insane levels, customizability.
Can't talk today.
It's been a long week.
Yeah, a bunch of things on that side which just.
This is just pooling together everyone's request.
So your firware menus get incredibly long, so you can do literally exactly what you want.
You're talking some very strange features that, you know, only one person in the planet has requested, but we've included it.
So, yeah, it's a bunch of firmware stuff, new hardware choices, more toys elsewhere.
Amazing.
And if you want to learn more about STAR Labs.
Where do they go?
Where do they go?
Yes.
Starlabs systems.
Amazing.
Perfect.
Sean, this has been very fun.
Thank you so much for coming on the show.
Have a good one.
Thank you very much.
Cheers.
Found an error or typo? File PR against this file or the transcript.