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.
Inside System76: Building Open Hardware and a Linux-First Future
Play On
17 JUN • 2025
1 hour 9 mins
Share:
In this episode, I’m joined by Carl Richell, founder and CEO of System76, for an in-depth look at what it takes to build an open source hardware company from the ground up. Carl’s journey, from launching Ubuntu laptops before Linux was mainstream to building open firmware, a desktop environment, and even their own factory, is one of the most hands-on stories in open source computing.
We begin with the early days of System76, where Carl shares what it was like to ship Linux hardware when driver support was shaky and installing an OS meant burning CDs. His focus on delivering a seamless out-of-box experience and supporting users directly helped System76 stand out in a time when Linux on laptops was far from a given.
The conversation really picks up as we talk about System76’s move into manufacturing. Carl explains how and why they decided to build their own factory, and what they’ve learned from designing everything from the chassis to the firmware. We get into supply chain complexity, the challenges of BIOS-level access, and the value of having full visibility into what goes into your hardware.
On the software side, we talk about the evolution from Ubuntu to POP!_OS, and how feedback from users shaped its development. Carl also shares the thinking behind Cosmic, their new desktop environment written in Rust, which is built to be modular and deeply customizable. His approach to user experience, design flexibility, and upstream contributions reflects a strong balance of engineering discipline and community focus.
What really stands out is how System76 blends idealism with pragmatism. Carl talks about what it means to stay open in a world full of proprietary barriers, how they navigate licensing, and why they’re committed to making computers more auditable, secure, and user-owned.
If you’re curious about how a small team can take on hardware, firmware, and software while staying true to open source values, this episode is for you. Carl brings a rare mix of vision, clarity, and hands-on experience that makes this a compelling listen for anyone building in public.
Transcript
Show/Hide Transcript
Technology, it's the most computer's, the most versatile tool we've ever made.
So it amplifies the ability to improve all these other things in the world.
And because of that I believe it must be open source.
And so taking that ethos and trying to identify where I can contribute to help the Linux desktop become more popular and more usable and accessible to people is the reason behind why we, you know, started sim 6.
Welcome back to another episode of Nerding out with Victor.
Today I'm super excited to be joined by Carl Richell, who's the founder of System 76.
Hey Carl, how you doing?
I'm doing great, thank you for having me on.
So the reason why we really our paths crossed was about six months ago by now part of the recording back in the Hague at a Ubuntu conference and we had boots next to each other and we just started chatting really.
And I've known you guys for a long time.
You guys are kind of OGs of the Linux desktop world and Linux, well, I guess desktop and laptop world.
So I have a lot of respect for what you guys have done over the years.
So it's super exciting for myself as well to have you guys on and have you on and maybe for those who are not familiar with System 76, walk me back to the early days of when you started.
Like why was the first like go walk by early 2000s I believe when you guys started like walking back to then, why did you start the company?
Yeah, it's funny, I get asked this question often.
Sometimes I think, you know, I don't remember what I had for lunch yesterday, let alone 20 years ago.
But you know, a lot of the movies are also still really vivid because it's such a special thing when you're building things.
And so it was 2005 when were founded, so this is our 20th year in business when we started Linux, in my view, the Linux desktop in particular had reached a level where it was of the same quality and similar capabilities of its proprietary competitors.
I think Mac was OSX at that time, so of macOS.
And yeah, I don't remember what Windows version were on but the point being Linux was already a fantastic desktop operating system.
But the way that you had to consume Linux was find, you know, get an ISO.
At that time it was CDs that you would burn the ISO to install it on your hardware and then find out what works and what doesn't work.
Live disks weren't really around at that point.
That came a little bit later as well.
So you really kind of had to commit and jump, jump in if you really wanted to see what this was about.
And, and I, you know, I felt that was, you know, a pretty heavy ask of, of most people.
And so, and from other OEMs Linux was on the server, but nobody was shipping Linux.
Linux hardware really.
And the best way to deliver, I think a great experience is to take the complexity of, you know, downloading this ISO, burning it onto a CD and then finding out what works and doesn't work and reducing that to placing an order for the hardware that you'd like and everything just works.
Yeah, you don't have to worry about that.
And if you do have a problem, you have someone to call and talk through whether it's just using Linux or whether there's some kind of issue with hotkey or suspend or you're having a problem with audio.
That's the kind of word by itself.
Yes.
Yeah.
Audio is almost always something that has to be developed for a piece of hardware.
Yeah.
Because the codecs or how they're implemented or wired changes regularly.
So it's typical that we have to do that.
So a lot of the work was making suspend work reliably making all the hotkeys work.
At that time we had dial up modems still.
Oh yeah, I remember, yeah.
So making dial up modems working.
Things have improved out of the box for Linux in a lot of ways.
Yes.
Wireless almost never worked before.
Only certain wireless modules worked, other modules wouldn't work.
Cisco ones were pretty reliable, but they were very expensive.
Right.
Intel was, intel has in my view always been quite supportive of Linux and ensuring the chipsets work on Linux.
And when they entered the wireless module arena that also eased things quite a bit because they were ensuring that the modules worked on Linux.
But before that we're doing all sorts of kind of crazy things to wrap Windows drivers and enable works and sure, it works.
That's a blast from the past.
I remember that.
Yeah, there's a lot of things that we take for granted today that just weren't the case in 2005.
And that also means that was the experience for a lot of users when they would install Linux and they say, well, this thing doesn't really.
And mind you, laptops are coming out at the same time.
So before that it was desktops and desktops were typically quite a bit easier to support, but laptops have more functionality with, you know, with their ACPI requirements with hotkeys, with wireless and Other things that just are less common on desktop.
And so, and so users were, you know, I think understandably trying Linux and saying that there are things that just didn't work for them.
Even though the desktop experience is really nice, it was this hardware enablement part that just didn't smooth out the experience.
So that's why we started Sim 76.
And even before that, my own personal ethos was driven by have foundational ideas with technology.
And that in that austerity itself is generally a technological problem.
Whether we have enough healthy food or clean water or enough energy, these are all technological issues for us to solve.
And if there are technological issues, then the best thing that amplifies that is a computer.
And what amplifies computer is it being open source, because that means more people have access to source code and understanding how software works and using that software and building upon it.
So my belief in open source goes quite deep to how it can improve humanity itself.
Because Technology Computer is the most versatile tool we've ever made.
So it amplifies the ability to improve all these other things in the world.
And because of that I believe it must be open source.
And so taking that ethos and trying to identify where I can contribute to help the Linux desktop become more popular and more usable and accessible to people is the reason behind why we started Sims 26.
We'll dive into more recent developments, I guess in systems with POPOs and COSMIC and all those things in a second.
But what did you ship?
The first devices, what did they ship with?
Was that Ubuntu, Debian, what they actually ship with, or any all of above?
The first version was Ubuntu Breezy Badger 510.
So there was October 2005.
I do like their numbering because I can even remember, you know, the month that it came out in 2005 and the year it came out.
So I think the first reason first release was 504 and that was Warty Warthog and.
And then Breezy Badger was the second release and that was the Linux distribution that we shipped with.
So it was very early days for Ubuntu and it was somewhat of an odd decision to make at the time.
I like to say it was intuition.
Maybe it was luck.
It was much more likely luck.
But I had been running RHEL and Open and Susei and I really like this small distribution called Yopper and what they were working on at the time not around anymore, but thought it was a cool distro and the first time I tried Ubuntu, I installed it, you know, from my burned disc and I couldn't su to root what's, you know, what's up with this?
And I, I spent like five minutes on it and I was like, okay, I'm moving on.
And then like a month or two later there was this Ubuntu guide that came out and it was maybe 40 or 50 pages.
And I printed it out and I put it in a binder.
These all feel really like archaic things to do, but.
And then I reinstalled it and I walked through it step by step.
I started understanding the design decisions behind it, why Ubuntu had built it the way that they did.
And then it all kind of clicked that.
I think they're onto something here.
And that's.
That eventually led to Ubuntu being the distro of choice.
We specifically decided not to have multiple distributions available because I think the decision to buy a Linux computer in and of itself for a lot of people that are using Windows, we're not here.
Yes, we care very much about supporting the Linux community and the users of Linux, but what we're really here for is to expand the that Linux market to all the people that this is brand new to right them coming to a website and they're okay, now I need to pick which computer makes sense for me.
Okay, let's try to make that as easy as possible then.
Okay, now I have to choose which Linux distribution and I don't know the differences or what all these things mean.
Yeah, so I think it was important to decide and use, you know, one Linux distribution and you know, make that a great experience.
It also means our focus could be very tightly aligned with fixing any issues with hardware with one distribution and trying to support them.
I think quality of product would have been lower.
So for all those reasons, we chose Ubuntu and went with one distribution that we would ship.
And I think it was roughly around that time both Dell and back then IBM launched.
They had, both them had laptops.
I believe, at least with Ubuntu as well, it was roughly, not maybe it was a few years later when they shipped that.
Right.
Was a.
There was a wave and then they killed it and then it came back and they killed it.
Right.
Yeah, I remember this distinctly because I think it was a couple of years after we started, maybe three.
Three years.
It was a while after we started and Dell threw up this website that said, what do people want?
And they were kind of crowdsourcing.
What, what?
Well, you know, what the people, you know what their potential customers wanted.
And the top one was Linux on laptops or Linux on their products.
And to their credit they responded to that.
Interestingly, Dell, Michael Dell actually called me.
Oh really?
Yes, that was a treat.
Because I was always a fan of Michael and his work and what he had done with his company and what he had built.
And so they were trying to navigate, okay, this is what our users are, you know, this is what people are saying they want.
You know, system 26 is doing this, you know, let's talk to them.
And so I got to talk to Michael.
It was funny, I was on a.
I don't know, it was like this.
I put in my own VoIP phone system.
It was called Asterisk.
Oh yes, yes.
So I had done PBX programming before and so I put asterisks in and that's what it was doing, like our support routing and automated attendance and all those kind of things.
But you know, it's all over the Internet and my Internet, while I was on the phone with Michael Dell, my Internet starts like crapping out.
Oh man.
But the first thing is chopping up and he says, I'll just call you back and calls me back.
And it was smooth after that.
But yeah, that was a little bit embarrassing.
That's funny.
We had a good conversation.
I told him, you know, the first 30 days, because I first again, our goal is Linux desktop success.
And so to my mind, Dell entering the market meant that would help, right?
With that success, it would legitimize Linux on the desktop because a Tier 1 OEM is doing it.
I didn't think of it as in some kind of competitive way.
So I just answered all questions that he had honestly and openly and we talked about it because I wanted to help if I could any way that meant Linux could be more successful.
And I think they shipped with Red Hat.
I believe they shipped with Ubuntu.
Yeah, they were shipping Red Hat on servers and servers but on the laptop Ubuntu had at that point it established itself in the mind share of the Linux community.
That that was really the, of the distribution to target.
We were right.
Okay, again, lucky.
Or had some intuition, whichever it was when we started.
But by the time Dell was looking at it was pretty clear choice.
Interesting.
Yeah, yeah, so they kind of, I'm not sure like if they are still shipping with that as an option because I, back then it was like the Windows Tax was a big topic of discussion in The.
In the Linux world.
Right?
Yeah, Yeah.
I don't know.
It's.
I know it was a Sputnik project, and I'm pretty sure it's still going up.
It's interesting you say that, because it's been a while since I've heard about it or thought about it.
So.
Yeah, I think the Lenovo X1s you can get a special version of with Ubuntu, but not like the wide range of all of the versions, which is kind of.
Yeah, I've had some conversation with people at Canonical about that as well, but.
And it's.
Yeah, it usually comes down to the whole driver issue that you're describing.
Right.
Like making sure that it works across all of them.
Right.
And that's not a trivial part, a trivial thing to get to.
All right, so you launched that, and then you decided to launch on Ubuntu, you launch your first few devices, and at that point in time, how much was custom, unlike with regards to your devices or the system 76 devices?
And how much.
How much was just a.
I guess a branded OEM device?
How did you start from there?
Yeah, so the.
On the software side there weren't.
We were taking vanilla Ubuntu and just doing hardware enablement.
Right.
So, you know, identifying.
One of the interesting things about QA and QC is that your.
Your checklist for quality only grows, never becomes shorter.
So we had a checklist that we go through and identify all the holes and support, and then we do the work on the hardware enablement.
So you ended up writing drivers or fixing drivers upstream in the kernel, or how do you do the enablement on that side?
Right.
Yeah.
Some things like touchpad might not have been working correctly.
And.
And so we would push patches upstream and then we would deliver a DKMS package in the interim until the upstream packages were released, and then went through the entire flow of Ubuntu's process and out to customers.
So we have this gap in time between when support is coming directly from Sims 96 and then when it comes from the kernel.
Right.
Okay, interesting.
And then you evolve from there to.
To do a lot more customizations these days, like how much is entirely custom today and how much is bored, like, and I guess, how many components are you actually doing yourself today?
I guess you're not doing your CPUs, obviously you're not doing your GPUs, but in terms of how much is custom today, how far have you gone, the manufacturing journey of that?
Right.
Well, in 2018, we built a factory in Denmark and that was to start taking more control over the design of the products and start taking more control of how they're delivered and how we can iterate on improvements.
Coming from largely a software background, we think of things as continuous integration and continuous improvements.
And when we're working with hardware, particularly case vendors, they're kind of designed in a way that they're supposed to fit lots of different needs, but that doesn't mean they fit the needs when you're talking to your customers about what they want.
Right.
So that's.
So that was our first foray into manufacturing ourselves instead of using external manufacturing and our own design as well.
We started with desktop because it's just an easier form factor.
And long term the objective is to, well, we work with ODMs for laptop manufacturing and our mini PC manufacturing, but the long term objective is to take that engineering and design expertise that we've grown and start moving that into laptop products and into other products as well.
Our launch keyboard was like an interim project to start getting there because we wanted to learn precision milling and we wanted to learn high speed electrical design and things that went in that.
That's why our keyboard has a high speed USB hub in it because it's rather complex to do but also really nice feature for customers.
And we get that to get to learn the institutional knowledge of how to achieve that in delivering a product.
Plus the precision milling around USB C ports, USB A ports and fitting that all together.
So all of it's been a journey to kind of grow our expertise as we bring more and more in house.
How ambitious can you be there?
Because they're obviously you can't fabricate a lot of chips.
Right.
They, they.
That's more or less impossible to do outside of a few manufacturers in like Taiwan and China.
Right.
So, so like efficient either.
Sorry, it's not efficient either.
No.
Right, yeah.
Your, your volume is not quite up there with Intel's volume or with the volume of of these like Broadcom or all these guys.
Right?
Oh yeah, Nvidia.
It doesn't make sense for Nvidia or for Dell or HP for a lot of these.
That's kind of changing somewhat.
You see it in the cloud, Amazon and Google making specialized chips.
Yeah, the hyperscalers are doing a lot of customization stuff.
Right.
So yes, I think that says a.
Lot about their margins, certainly possibly more than other things and how heavy the comp.
The competition is.
Yeah, but I mean sharing supply chains I think really important for efficiency and competitive and keeping prices Competitive for customers.
So I think there's decisions to be made.
I would love to have an SMT line to manufacture motherboards.
I think that would be awesome.
Yeah.
But is it efficient?
Does it make sense?
There are, you know, should we reproduce that or.
I think a lot of the secret sauce is how you take those components and integrate them into a final product.
More so than the individual components themselves?
In a lot of cases.
Yeah.
I mean, obviously the landscape in the last 12 months when it comes to supply chain and supply chain security as well for hardware and chips like that is a debate.
We've seen a lot more in recent time.
And being able to at the very least control the firmware of a lot of these chips is massive.
And making sure that's done in a friendly environment.
Right.
Because look at the security landscape, for instance, around like hikvision is one of those, like CCTV camera companies.
Right.
They're banned by the UK government, for instance, and the US government, I think, because don't really trust it.
Right.
Or Huawei.
Right.
So like that's bringing back, obviously bringing back manufacturing, but like bringing back assembly and understanding of what goes in there, I think.
Is something happening more and more.
Is that reflecting what you're seeing as well?
I think that's a great point.
Having an.
When you're building technology, having an understanding of that technology from the firmware level all the way up, I think is really valuable not only for the experience you can provide for your customers and the support, but also the security for your customers in that when they have a question about something very technical, lower level, you actually have either the source code or the ability to respond to that in a way that builds confidence with those customers.
It's complex and goes right to your point about supply chain and supply chain challenges.
Because even with firmware, when intel introduced bootguard, that became very complex because there's commands you run at the end of firmware provisioning in the manufacturing facility that.
That burns the fuses.
And that means you can't rewrite the firmware, we can't hack on the firmware.
And so we have to have special SKUs just to be able to do things we're doing with open firmware.
Fortunately, in a lot of places we have the scale and relationships that we can do that, but it's not in all places.
And so that's also a growing project where we can expand our use of open firmware to more and more products that we're building.
Yeah.
And that was.
You set me up for a great segue here because like the reason why we at screenly reach out to you guys and wanted to work with you guys because we wanted to have a partner we can work with that can actually understand all the layers, right?
And for me, the reason why this really became a more urgent priority was in the rise of like logo fail and pixie fail.
That really highlighted the importance of understanding what actually goes into the blobs that sits on your device.
And I work with ODMs, OEMs in China and very few of them are technical.
They are box pushers, right?
They were just like, yes, here's a box.
If you ask them anything about it, they will not be able to tell you.
Like there are obviously more savvy ones that will do that, but a lot of them that are within reach of anybody who's not doing like insane moqs will not be able to do that.
Right?
So for me, so just with our new screenly player, Max MK2, which is powered by System 76, that's the one thing that I'm super excited about.
Being able to have a partner that really understands all the way layer down.
So like from the bias all the way up, how things work.
So open firmware, open hardware, those are two things that really resonate well with me in the ethos of there.
So the play we have, we are not quite on core boot in this iteration, but the goal is the next version of this Max player will have core boot, which I am super excited about having.
That's kind of what we started.
But talk to me a bit more about your vision on open firmware, because we can dive into open hardware in a second.
But open firmware, let's start with that because I think that's something that we've only in the last year or two start to see more and more conversations around.
Right?
Well, this kind of goes all the way back to 2005, because I remember distinctly sitting in my basement at this table with a laptop and just in awe of how well Linux is working on this laptop and booting it.
But then you boot in the bios and I remember thinking, but this, here's the bios and that's that last, the last bit that yeah, if we're ever going to cross, get to cross that threshold.
But sure enough, years later with our, it actually kind of started.
One of the problems with jumping into open firmware is that I wanted to stay consistent with being able to offer open firmware.
So when we're going from one intel chipset to the next intel chipset.
It appeared to happen to me within Kubrick community, was that it would be three, six months, maybe longer before that new hardware was supported that would work as an oem because you need to be much more rapid with your releases.
And we're competing with Windows.
We don't want to have, we don't want to be on old generations.
Plus the hardware supply chain doesn't flow that way either.
Yeah, actually hard to buy old CPUs that integrated more.
So intel contacted us once, said, you know, we'd like to work with you on open firmware.
That three engineers come out and we sat in a room.
It was like blizzarding outside.
And, and like eight hours go by.
They have really long.
I love this story because it's so funny to me, this is like super long presentation.
We're going through the slide deck and everything and they talking, we're chatting.
And then the next morning they came and said, so will you work with us on this?
And just said, well yeah, you had us at open source.
That was all.
This is exactly what we want to do.
We're going to be able to do.
We wanted zero day open firmware.
We've had some challenges recently that we haven't had in the past with that Intel's latest Horse X chipset platform was such a dramatic change.
And because of supply chain things where were moving supply from China to Taiwan, all the timing just became complex.
And so now we have some products that are shipping with closed firmware while we port the open firmware because the timing just didn't work out.
But we dedicated to, we're dedicated to open firmware and ensuring we can continue that on products.
And today in the bo.
In.
In the device you're selling, how many proprietary blobs are in there still?
Like, are.
Are most open firmware or like I would imagine GPU drivers and whatnot may not be.
But like how much of the blobs are actually closed?
That's a, that's a great question.
I don't have a exact answer because a lot of the functionality for firmware, a lot of the stuff that you actually feel is in the embedded controller.
So how rapidly your laptop charges, what your CPU wattage is set to, what's it set to when you unplug for more efficiency.
Same with GPU boosts.
And why does that set out?
How much the CPU and GPU can share all those things, your hotkeys, backlight, all that comes from the Embedded controller.
So you know, I feel like that's actually the most, one of the most important parts to be open.
The BIOS or the firmware itself outside of it is really doing hardware initialization to bring up the memory and CPU and everything else.
I know that there's a FSP component that comes from intel that's integrated into the core boot, but exactly how many like blobs and how much exposure there is in the fsp, I'm not entirely sure but I know it's a long leap from that first laptop in 2005 that I thought would never get to open forever.
So I take the approach in a.
We, we work within the confines that we have and we chip away over time.
Yeah.
So it's, we're not going to get my dream is to completely liberate the computer.
But that's a huge ask.
But I'm also here, I have the endurance.
I'm here to, I'm here over time to just chip away at it.
Yeah, how, I mean how you started by yourself, tinkering your basement.
How big is system 76 today in terms of headcount and 50 employees?
Okay, so that's how.
What's the breakdown between engineering support, sales, whatever other rules that you may and manufacturing as well?
I guess.
Yeah, I think that's one of the cool things in this like a 50 person company.
We do everything from firmware development to deciding what aluminum alloy to use for a part.
Yeah.
As the expertise between all of those, how paint adheres to metal and how to write a desktop environment with auto tiling and session management and audio and Bluetooth, all those other things, it's all condensed into a 50 person company.
So it's impressive thinking.
These are all rough numbers.
Yeah.
Engineering is about 10, the design is 2 and marketing's like 4 and sales is like 4 and manufacturing is like 10 and operations is some, you know, filled in within that group.
So yeah, it's everything we have.
And with care department, it's a five team department.
So yeah, it's care, marketing, sales, manufacturing.
There's the production team which does production, is taking all the metal in and producing a product that the build team then completes.
So the build team takes that chassis and then builds it to order for the customer, installing motherboard memory and so on and so forth.
And then you know, operations is the, you know, the stuff with HR and.
Yeah, yeah.
Healthcare and our systems for imaging and how those are working and how we Take orders and how an order that we take ends up in the ERP and the accounting and all the, all that fun stuff that's, you know, part of operating a company.
Yeah.
And they're all based in Colorado.
Everybody's on site or how's your, how is your company structured?
I think it's about 35 or so are in Denver, 35 to 40.
And we have about 10 remote folks from Germany and Poland to California and Tennessee and so yeah, quite a few folks that are also remote.
Interesting.
Oh yeah.
Cool.
It gives a bit of insight into how you guys are working and what I'm really curious about because I have this vision for at least for the screen, the players, for the long run.
I, I really want to embrace open hardware and open like firmware.
Right.
I from, for me that comes in part from a philosophical perspective, but in, but perhaps more importantly from the commercial side of it.
If it's not open, I can't trust it.
So it more comes back from a security perspective.
Right.
Like then it can be audited.
Not necessarily they will be audited, but at least it can be audited.
And I think that's an important distinction.
So what's your pipe dream there?
Like is it for.
Can we.
Is there a world where the future generations of systems of six devices are fully open hardware, open firmware, open source through and through?
Is that, is that the pipe dream or is that a bit too ambitious?
No, it is but.
And I think you have to start with what your long term and your most ambitious goal is.
So if we think about hardware, everything that systems success is open source.
So our all hardware design, all electrical design, all firmware and software, everything we do is open source.
So when we started it was, you know, just drivers and hardware enablement.
And now that's driven into like the electrical design for the launch keyboard.
If you want to understand how high speed routing can be done on a pcb, you, everyone in the world has a great example of how to do it well and can take that and learn from it and adapt it.
Interesting thing about that is electrons really don't like sharp corners.
And then when you're doing something as high speed as that, some people are looking at our traces and saying, oh, they're just not pretty.
But you know, that's actually there's something more to learn from that.
When you're looking at another keyboard, pcb, it's because they're not doing high speed.
Right.
So over time we've been able to introduce more things that are open source.
And I don't really see an, I don't see a necessary requirement to limit ourselves in that.
There are practical things that just are the case and RISC CPUs are extremely exciting.
But I don't see us.
But we're not at the sites where we're going to be designing and integrating risk CPUs quite yet.
We would use a partner for that.
And the thing about risk CPUs is although installation, construction, set and things are all open source, there's a lot of proprietary things that slip in for competitive reasons even within that ecosystem.
So.
Yeah, yeah, so I think it's, I think we just chip away and day after day and year after year we do more and more.
That's open source and over the long haul, I mean I think every day we can look at something so we've built something that's more open than it was before.
And if we can say that again next year where it's more open than it was before, then we've made progress.
Yeah, I mean the ability like to do these things are required.
I mean to do they requires a lot of skill.
Like I had even Upton on a while back and he was talking about, I was asking him kind of similar question about the pie by being able to do risk 5 PI and basically like even with their volumes, like it's almost.
Is very difficult.
Right.
And obviously they are doing a large volume, larger volumes.
Right.
Than any PC manufacturer would do.
Well, not any, but most boutique manufacturers would do.
Right.
So.
And I think he, his biggest problem was around the GPUs, if not mistaken.
Like doing an open source GPU like would be almost impossible.
Right.
Even though that would be the dream.
Yeah.
And we should dream for a more open future.
But we work within the, you know, the world that we live today and we just, yeah, we just chip away at it.
I don't know what I'm going to achieve in my lifetime, but I know it'll be more open by the end.
I just, I, you know, we just.
The important part is that it sticks.
Yeah.
Things go backwards too.
But for it to stick, I think that means it has to have utility beyond just the fact that it's open.
And I think open in general often means it has more utility, but it means taking a very liberal perspective of what open source means.
It's, you know, sometimes there's the, sometimes there's complaints about taking open source software and someone else using it to do something and some ideas come out of that like nih, like not in house or ideas like Upstream first or things like that.
I think generally to make open source more useful and sticky and have brighter future, we have to take a very liberal view that is the reason that we're doing this is for people to take it and even if they're adapting it, as long as it's, you know, all boats are rising in the same way thanks to the way the GPL is structured.
And so their changes are also GPL means that it's good for open source.
Even if it feels like I made this thing and now it's being, you know, changed and user, that's we should encourage it.
Yeah, I mean, I think that's.
I was going to ask you about Chrome os, right.
Because I think Google has probably done more for Linux open source than most other in the sense of like Chrome OS actually in terms of deployment fleets.
Right.
Because Chrome OS is if you measure by desktop like it's probably by far the biggest Linux distribution.
Right.
And much to your point above, pragmatism that by itself has helped with core boot, has helped with probably a lot of driver issues.
Like how.
How do you see that symbiosis?
Like is that something that you've seen on your end in terms of like has the help the tide rise or how like how have you seen that kind of changing the landscape?
Yeah, I really, I think, I don't have direct evidence, but I think you must be right that with Chromebooks running essentially a Linux kernel, that means the upstream ODMs are thinking more about making sure Linux works, gaining experience in Linux and contributing to the kernel so that logically follows that path.
I don't know if Chrome OS is a Linux distribution.
Well, you could argue that it is.
You can argue it isn't.
Right.
It depends on how pragmatic you are about and purist you are about the argument.
I guess I think it's almost like what I'm thinking about with the liberal view that you should be able to build other things with the software we built.
That's what makes it valuable.
That's what Chrome has utilized.
But is that what Chrome OS is?
Can we take Chrome OS and build Linux distributions or do we not do that because we feel it's limiting by its own nature?
I don't know if you can build your own Chrome os.
I actually, I'm not sure.
I don't know how much is open source.
That's a very good question.
I actually don't know Android is the same thing.
I was thinking about this.
My drive's like an hour and a half from work to my house, and so I get lots of time to think about stuff and driving.
Thinking time is always good.
But I was thinking about this.
With Android, we think about.
I remember when Mark Shuttle with the first Ubuntu bug was Windows has a majority of market share.
Right.
And I remember when he closed it, because Android became the most popular operating system and it's a Linux operating system.
So that bug was no longer the case.
And technically it wasn't as.
It's true that Windows wasn't.
But does that mean that Linux was?
Because especially with the case of Android where its functionality is kind of nerfed intentionally without these other services, you certainly do things with it, but it's built in a way that doesn't make it as.
It can't be as useful.
We can't.
The app ecosystem is by no means open in a way that you can build a product off of it and build your own Android.
You could argue that it is because there are alternative app stores.
But then are they really like.
But they're equally.
I don't know, it's.
Yeah, yeah, right, yeah.
And it's not really to knock Google or their approach.
it's just that when I think about a Linux distribution and I think about Arch or Ubuntu or Debian or all of these, they're, I mean, they're very easy to build things with and on.
And I think that's what makes them, that's what makes them powerful.
If you're building, if you're building something with Android, you're, you're locked into a different ecosystem that isn't as open.
So I guess that's the same thing.
I, I take a very liberal view about what open source means and some.
And sometimes I think of it like there's the GPL in license and there's the GPL in spirit.
Right.
And when I think about the GPL and spirit, whatever the intention, I think what the Solomon's intentions really were, I think about it in spirit.
It's more like Debian and less like Andrew.
For, for what it's worth, that's how, how it feels to me.
Right, so is that.
It sounds like that's a principle.
You really take it to heart in terms of like philosophy, in terms of operating the company as well.
Yeah, yeah.
It's.
We do everything that's open source first.
There's always these little business logic things that just don't make sense being.
Yeah.
In the open, like.
But even a lot of those, unless there's a really good reason, they're just in our repo and other people can look at them and see them.
If we go down the path, let's assume for a second that open hardware really thrives and everything is open and the next.
And then like it's fully open source.
So anybody can manufacture a System 76 device.
Right.
What, how do you see that world right where you.
Where if you have complete open specs for everything and anybody can just go and say build me this in China, like how do you see.
Like, do you have a moral objection against that?
Or like from both from a commercial standpoint and from a, I guess, quality control standpoint as well.
How do you see that?
Like if we go full circle.
Not full circle, but if we go extrapolate further, like where it's fully open source, like istec, here are my PCB files, M Schematics, whatever it is, send them off to China and build myself a system laptop.
Right?
Yeah.
How.
How do you see that philosophically and I guess as a business owner as well.
I think that would be.
That's the point.
So we did this launch keyboard.
We have fully open schematics and anyone can take those schematics and build the pcb.
And I, and I don't know that anyone has done that.
I think maybe what we've done is maybe a little crazy because the features that we built into it, but it's entirely open and the point is to enable people to do that.
It wouldn't be no different for a laptop if.
And we've done quite a bit of work on laptop motherboards as well, but those are more complex in certain areas.
There's the licensing makes it very hard to do an open pcb.
Yeah.
So we've looked at ways to build laptop PCB without violating upstream licenses, but make it as open as possible.
Possible.
So there are complications there.
But to your question, let's say we release an open source motherboard and we do want to do this.
We want to build a modular open source motherboard that.
Whereby you can, you know, attach different kind of IO to the motherboard to build any product that you want.
Basically what we're doing with Cosmic.
But do it in hardware, we will do that.
So.
And then, you know, additional.
And then we start having that manufactured and then it has a utility that other people can more efficiently just because.
Just a matter of being able to take something that's built and build upon it instead of reinventing that wheel for every laptop.
Well, that means our efficiencies grow.
That means would it make sense for them to go to another PCB manufacturer and perhaps, or perhaps we can just do larger MOQs and then, oh, okay, our numbers are getting bigger together.
We're building different products, we're adding values.
Value at different levels.
Maybe somebody's doing a Windows laptop with it, we're doing a Linux laptop with it.
Or maybe they're building an entirely different type of device.
Maybe they're doing, you know, screens and, you know, those type of.
Maybe they're doing that.
But now this ecosystem is grown where we have open hardware that different people can take individual components of that hardware and build products.
And now we can scale that up.
And now maybe we start getting to the scale where, okay, maybe we can do ARM CPU design because we're doing this much.
Maybe we can do an ARM board and we can do an ARM CPU design because we're doing this much volume now maybe we can start working on risk because now we have the volume that we can make the chip that could go on the board.
All these small players together could achieve more through this openness.
And even if big players would adopt a platform like that, it just means more scale for everyone and more ideas and more improvements.
And I think it's not different than software and that, yeah, there's much more capital involved, but it has the same potential to lift all boats because we can share in platforms.
Yeah.
So you're not worried about Noem in China basically taking the schematics undercut you on their MO.
Undercut you because they do a higher MOQs.
No, no.
Okay.
Yeah, because I would, that would be like a, if somebody can compete with your product at a lower price, that would obviously be a bad business risk, right?
Yeah, Well, I don't know.
I don't know.
That is, if so, you know, you could always be undercut.
But I think when we think about products and we think about brands, the way that, I think the right way to protect what your brand is through copyright.
So, or, and trademarks.
And I think these are important because a trademark or copyright isn't about not making something available to people.
It's about the reputation of the person that's making something.
So we preserve our reputation through our system 76 and that logo and that brand.
Right.
What you can expect when you get a SIM six products versus another company is what defines who we are.
Right.
And so even if the guts are the same because were sharing in this platform or we've open sourced this platform, you know, that doesn't mean that they, the product is the same, the software or the firmware on top of it is the same or the customer support experience or that.
Right.
Or when we're helping our, when we're holding our customer's hand through a process to build for a project.
All of those things are what make us who we are.
The hardware and the software I think is one component of, you know, an important component of it, but not a full design or holistic.
So you want to see that commoditized and then kind of like the brand becomes kind of like.
Well, it ties it all together really as an experience.
And in many ways I think you see, when something is, has utility and other people have, can gain utility from.
Oh, I just remembered this.
It's funny, it's an interesting conversation because someone tried to sell us our open firmware.
They sent us emails saying, hey, I can make open firmware for you.
And you know, here it is, there's like a small company outfit and we're like, wait a sec, that's our firmware.
So somebody's out there trying to sell other people our open firmware.
Right.
And perfectly fine.
I don't care.
I don't care at all.
It's okay that people.
I think it's fine to sell open source software too.
It's much more important that it's open source.
But the, but yeah, this is part of, it's, it takes a lot of bravery to give things away.
It takes a lot of generosity to give things away.
But we're not going to make the impact on the world that we can.
If we're not very liberal with the way that we approach this, we can't be angry that other people are using our embedded firmware.
That's the whole point, you know.
Yeah.
Grant it, you know, go forth and make, yeah, we're all part of, I appreciate people that just make stuff.
Right.
Even if it's on top of other, you know, other people's stuff.
That's, that's great.
Add value to the world.
Yeah, fair enough.
No, I mean that's, it's genuinely like obviously a, A, a solid approach.
It's just, it just takes a lot of Bravery, like you said, because you can easily be.
Be attacked from angles that you couldn't otherwise be attacked, I guess, in terms of like.
But it's true.
Yeah.
All right, let's take a look into your other portfolio.
So we kind of alluded to popos.
Is that how you pronounce it, popos or.
It is, yeah.
Why, like why?
You started with Ubuntu.
Now you're like, let's make our own OS or fork our nos.
I guess.
Yeah.
So it started.
So, yeah, we had spin shipping.
I'm trying to remember years now.
I think it was 20.
We're shipping Ubuntu from 2005 to 2016.
And at that time Ubuntu was building their Unity desktop, but they're from scratch desktop.
So before then they were building a new UX that they felt would better for their users, and they're building it using GNOME technologies.
There were some challenges with that because it was hard to package Unity for other distributions because they had to patch all these GNOME libraries and so it would conflict with running just vanilla GNOME on other desktops.
So Unity really was stuck in the.
Or stayed in the Ubuntu distribution and ecosystem.
But the project build the Unity DE its own desktop environment.
Eventually they started with mobile.
And I, I can.
I completely understand why, because mobile is kind of taking over.
I'm.
I'm sure we'll probably get into Cosmic, but we took a different approach with Cosmic.
Some things I just, I had the opportunity to learn from other folks and this is, I think, one of those.
Right.
So.
So I.
I think being a student of is always a student's really important.
So.
So when Canonical decided to no longer develop the DE and go back to gnome, it was a shift that was an opportunity because there was a window opening where our customers are going to move something different one way or another.
So we had, over the years built up this understanding and knowledge of our customers and gaps that they had with their experience with Ubuntu and with gnome.
And so we decided that was the right time to build POP OS and take our understanding of what those gaps were and fill them in.
Interesting.
Some were really simple, especially at the beginning.
One, we wanted to establish a brand and an identity and a visual look.
And so there was a lot of work with that, but there were also a lot of small things.
One was that our.
Well, maybe they're not small.
Our customers really wanted full disk encryption out of the box.
Yes.
To do.
I'm sorry, TPM backed.
We don't do TPM backed for a reason because you can't move the drive.
And I think that's important characteristic being able to move the drive.
And so we don't do it with tpm.
We use, I think it's lux L U K s.
But to do that as an oem, if you want to shift full disk encryption, you can't encrypt the disk in your insole and in your image that you use to duplicate onto other products because then you have the key.
So we built a new installer for popos as well, whereby the installation is done during the installation, during the first time setup.
But it's so fast customers don't even recognize that's what's happening.
It's actually doing an install and encrypting the drive at their, you know, at their first setup.
So that was one thing.
Smaller things are like do not disturb notifications were kept noisy and bothersome.
And so we added a extension for that and little things that we'd shave off over time.
And then I think it was 2004 is when we added tiling auto tiling to Popwise.
So the core answer is that were able to deliver solutions that our customers needed that upstream wasn't in agreement with.
But POP OS is a fork of Ubuntu.
How would you position.
Uses Ubuntu's repositories?
Okay.
So.
And so the GNOME experience in popos is completely different than the GNOME experience in Ubuntu.
Right.
So it's lots of forks.
Okay.
I guess.
Really?
Yeah.
Like we did this tree of forks now.
So Ubuntu started off forking Debian, which started.
Yeah.
So it's built on Ubuntu.
So when you're using POP 2404, 2204, but soon it'd be 2404 using the same binaries.
There are differences in what we deliver because of hardware enablement.
We ship a newer kernel, newer mesa, newer Nvidia drivers, things like that.
That's just the nature of being an OEM and hardware a little bit sooner than happens in Ubuntu.
But the whole purpose autotiling is a great example of this.
Our tech support team, we're a small company and I'm not very far from the tech support team and they're getting calls constantly like second third release of POPOs, how do I install i3 on POP OS?
Or how do I install and set up sway on Popos?
It was mostly i3 at that time and all of our developers were using i3, even though they're building POPOs and they're working on an extension that we're delivering customer, they were using i3.
And then I went to a Google Summit or open firmware summit at the Google campus in Mountain View and everyone was, that I was talking to was using i3 and it was obvious we're not delivering what our customers want out of the box.
Yeah, they're calling us and they're telling us.
So what PopWest enabled us to do was to act on that need.
We.
It's not really.
I think different communities have different and different customer bases have different needs and that we should embrace that rather than this, you know, one operating system to rule them all.
I think distributions are fantastic.
You should have lots of different distributions.
Yeah, because you can solve different problems for different people and we don't all need to eat the same cornflakes.
We can, we can have different things.
It's good.
We had certain problems that like the GNOME community didn't feel like was a general computing issue, perhaps in auto tiling.
And so we wanted to develop.
So we built an extension in GNOME to do that.
And so that further differentiated POPOs from what it was before and all with the intent just delivering to our customers the things that they were telling us they need it.
And that eventually led to Cosmic.
It did.
It's a great segue.
So that is what led to Cosmic.
And there was, you know, a resistance to some of the changes that we wanted to make to build the product for our customers.
And so we decided that probably the best route for us long term would be to build, go ahead and build our own desktop environment.
We looked at a lot of different words.
Yeah, we don't lack ambition, that's for certain.
So we, so we looked at different routes of doing so.
We could continue with extensions to gnome, but we would never really be able to deliver the product we would want to.
And you're always kind of adapting it to what's happening, what's changing upstream.
And so you're spending a lot of time just staying at par instead of going beyond with new features.
So, so we started working on.
We decided that path.
We started investing in IT and bringing in the engineering team for the different components and parts that would be necessary.
We wanted to take a very different approach.
And our approach is that Cosmic is going to be a flagship desktop for POP os.
But our experience with wanting to build a product using a Linux desktop environment was that it was very difficult to do.
Not just because of sometimes resistance from Upstream but also because of how they're architected.
So Cosmic means it stands for Computer Operating System Main Interface Components.
Nice.
And that's precisely what it is.
Everything is modular, everything is composable.
So with the Cosmic desktop, if you look at the repositories, everything is separated into a separate repository.
Everything is a different piece of software.
And you can combine that to create any user experience that you want.
So the intent is very similar to our intent long term with hardware platforms, is that anyone can take these components and the way they're designed and build unique experiences with them.
And it's intended to be done that way.
With Cosmic, we're delivering a full desktop experience.
But Cosmic can, it can be a photo frame, a gaming console, it can be customized in a way.
We're taking a lot of heavy lifting to be able to build this kind of experience and making that easy.
And then you're taking these components and arranging them and writing some custom components to build your dream out of it.
And it's all written in rust.
It's all written in rust to make it sexy, to appeal to all the cool kids out there.
But I guess you view Cosmic then more as a develop environment, I guess building blocks more so than a desktop environment, right?
Well, it's.
Right, it's a platform for building user experiences, right?
Doesn't matter what that device or that screen, the screen is going on might be.
You can build user experiences with it.
And for popos's purpose, it had to be really broad because we're building this for a full desktop experience.
But you could take a much smaller set of components or a larger set of components, or precisely what we're doing.
And just if you're a Linux distribution, for instance, you can rebrand it for your distribution, change the way the panels work, may modify applets that are in panels because that's, you know, better for your users, which you've identified as their needs, and build the perfect experience for your users all the way from the smallest device.
And you think about Linux itself.
What made, why is Linux so successful?
The kernel, that is, it's.
It's because it's so useful for a lot of people to build things with.
We have all the components from audio subsystems with storage and memory and hardware drivers, and we have all the initialization and services of systemd and we have all these components.
But what we didn't have was just that slice on top to build the user experience that they're seeing and interacting with on the screen.
And you Built it natively in Wayland.
I believe it is Wayland native.
Yeah.
Yeah.
So you kind of leapfrog the whole problem that much of the desktop world is suffering from in the Linux world.
Right.
Yeah, it has its own challenges.
Yeah.
But, yeah, we lived at.
Certainly at that point when we started making decision, reached the point where we didn't feel like our users would be losing a significant amount by not.
We do have X Wayland for games applications.
It's just a support for a lot of things that are not Wayland native, but not a Wayland session.
And so there were things that were outside of our control.
The Nvidia driver wasn't great at that when we started.
Now, for three years in.
And now I can say confidently that, okay, the Nvidia experience on Wayland is now up to par with what it was on X11.
So some things that were outside of our control fortunately filled in the gaps during our development too.
So I think Wayland is.
I think it's.
There, there are some gaps, but I think there it's.
It's smaller and becoming maybe small nuisance things than just deal breakers.
Yeah.
Yeah.
Interesting.
Yeah.
So I think when.
So when will we see Cosmic as a default desktop environment for POP OS?
It'll be in the 2404 release.
We're working on the beta now.
Okay.
We just got below where we did seven alphas.
And one of the things about a desktop environment and computers is that they're used in so many different ways that a long alpha period, I think was really healthy for the project because I think there's probably tens of thousands of users on Cosmic now.
So all of that feedback is helping with all those tiny little.
This weird toolkit that the menus don't show up, or this other exotic piece of hardware where the compositor needs some, you know, special logic perhaps for that.
All those are, you know, things that.
That this cycle is, you know, helps solidify.
And the ex.
That's all the experience that Gnome had before, Katie had before.
So now we're on the beta.
I think there's.
We just got under a hundred bugs, which.
The team size, it's not very big.
So hopefully within a couple months.
And I'm remiss, I keep missing my target.
So I remember this.
Do you think we'll go full circle and see Cosmic in Ubuntu as well?
That would be a dream, wouldn't it?
I, you know, I don't.
I.
I would love to see Cosmic be so good that it's Irresistible for folks to ship to their users.
So yeah, that's our dream.
Amazing.
All right, I'm going to run through a few more questions before we wrap up today, but one of the questions that I'm curious about is what was your most proud moment so far in the journey of System 76?
Either like a device shipped that you were really proud of or like landing a customer that you're like, oh wow, this is really cool.
Oh, you know, it's probably open firmware.
There are so many things that I'm proud of, what our team has done.
I'm building a factory.
I can't pick one.
We, went from not knowing you.
We never built a random factory before and we just, we did it all organically internally.
So that was.
That's really special.
And then also cracking open firmware.
So I have two.
There you go.
Try to limit to that.
But yeah, it's been a journey of really good successes, I think.
Interesting.
And what kind of.
Outside of your tinkerer and your kind of regular Linux desktop user, what was the use case that you've seen?
They were like, oh wow, that's really cool.
Probably like in some kind of non local environment, I guess.
Non desktop environment.
Right.
We have, well, we have desktops.
We have servers and desktops running in Apache helicopters.
That's pretty cool.
Which I never had expected something like that.
But yeah, apparently something like a four or five star General was in a helicopter while they were going over the tech they installed in it.
And so it was System 76 hardware.
So I thought that was.
That was pretty awesome there.
And also the GPL Lab, Jet Propulsion Lab at NASA uses a lot of hardware and that's pretty special what they're doing with it.
I think it was because the Cassini project, when we sent a satellite to Saturn where Sims 86 hardware was used.
I'm not on the satellite itself, but in designing and developing it.
So that was awesome.
That's fine.
We have actually.
So we have actually a shared customer though that's actually one of the screen.
The customers as well.
So it's.
Yeah.
So they have very.
So maybe we could get some System 76 new hardware to them as well going up, going forward.
So that'll probably be a good synergy.
Really cool.
Good.
Carl, this been a pleasure.
I think we've covered a lot of ground and I really appreciate you coming on the show.
Anything you want to do a shout out to before we wrap up things people should keep an eye out on or things you want to just spread the word about.
In general.
I think your audience probably is very similar to our audience.
I think let's all just keep fighting the good fight, open source, the future.
It's all because of all the work we're doing each day.
Amazing.
Thank you so much, Carl.
I'm very much appreciated.
Thank you so much.
Have a good one.
Cheers.
Bye.
Found an error or typo? File PR against this file or the transcript.