Ahead of the State of COSS 2026 launch, my co-author Sam Boysel, data scientist at the Linux Foundation, and I joined Andrea Griffiths and GPS on The GitHub Podcast (S02E07) to walk through what the research shows, for founders as much as for investors.
We discuss venture money in commercial open source, why stars don’t predict funding while criticality tracks valuation, how founders should translate open source traction into investor metrics, what the 2026 survey says about layering business models and about AI and how to charge without alienating your community. You can also listen to the episode in full or read the transcript below, edited for length and clarity.
Timestamps
0:00 – Intros: Andrea Griffiths & GPS (GitHub), Matt (Serena) & Sam Boysel (Linux Foundation)
3:51 – Three years of the COSS report
4:50 – Where the $26.4B comes from
6:18 – The third stakeholder: the community
8:15 – Bigger, faster rounds and better exits
9:30 – Criticality, not popularity
13:52 – Build your own project, or commercialize an existing one?
15:01 – Stars are a vanity metric
18:39 – Translating open source metrics for investors
20:55 – Pitch the value proposition, not the KPIs
23:09 – Is the open source vs. money clash fading?
25:23 – 2026 findings: adapt, layer, build enterprise sales
28:33 – Stop agonizing, and founders are bullish on AI
30:25 – From research repo to company: the pyannoteAI story
35:08 – Find the features that gain value at scale
37:22 – Charging without alienating your community
Transcript
Intros: Andrea Griffiths & GPS (GitHub), Matt (Serena) & Sam Boysel (Linux Foundation) (0:00)
Andrea: This is The GitHub Podcast, a show about the topics, trends, stories and culture in and around the open source developer community on GitHub. I’m Andrea Griffiths, and I’m joined by my co-worker GPS.
GPS: Happy to be here for another episode.
Andrea: Many of you have built an open source project to solve your own problem, shared it because you love open source, and then watched people start using it. So you start wondering: could this sustain a business? Who pays, and for what? And how do you make it work while keeping the community’s trust? That’s today’s topic. I’m joined by Matthieu Lavergne, who goes by Matt, and Sam Boysel, to talk about the State of Commercial Open Source 2026 study: what it tells us about building a business around open source, and which signals to look at beyond stars. Matt, let’s start with you.
Matt: I’m an investor based in Paris and a partner at Serena, a VC firm. I invest in early-stage companies, focusing on technical founders who build technical products bought by technical personas: CIOs, CTOs, data scientists and so on.
I’ve been a consumer of open source for as long as I can remember. My dad put me in front of a terminal when I was eight, and I still use open source every day. When I joined the investment world a few years ago, I realized investors were not as excited about open source as I was. I kept hearing that open source is bad for business and very hard to monetize. It just felt wrong. So I wanted to take a data-backed approach to the question: does open source make a good business?
Andrea: Thank you for joining us all the way from Paris. Sam?
Sam: I’m a data scientist at the Linux Foundation and a co-author of the study, both the 2025 edition we did with Serena and the 2026 iteration. Like Matt, I’ve been a passionate open source user since middle school; an openSUSE for Dummies CD-ROM was my entry point into Linux. My background is in academia and economic research, and I found a way to marry that training with my passion. Now I study open source and community dynamics for a living.
Three years of the COSS report (3:51)
GPS: How long has this study been running?
Matt: It dates back to 2024, when we did an MVP: we combined data from GitHub and PitchBook and started publishing benchmarks on open source companies. The feedback was very strong, because there is so little research on open source and business together. There’s plenty of content on open source as a category, but commercial open source was poorly documented.
Thanks to that feedback, we doubled down in 2025 with an extensive study of the entire history of VC and open source. The response was just as good, so we’re continuing this year.
Where the $26.4B comes from (4:50)
Andrea: One number that comes up in last year’s report is $26.4 billion. Where does it come from, and what does it measure?
Matt: It’s the cash venture capitalists invested into commercial open source companies in 2024. That year was special: a few mega-rounds biased the data. In steady state, VCs invest close to $9 billion a year into commercial open source, across roughly 250 deals.
The broader question behind the number is whether open source is a VC-compatible category. These numbers, and the ones in the report on the equity journey of these companies, show there is deep investor interest. There’s no excuse for good founders with good open source technology not to build a business on top of their project.
The third stakeholder: the community (6:18)
Sam: On top of the relationship between founder and investor, the study adds a third stakeholder: the community. We fortify Serena’s analysis with community data, from GitHub activity metrics to composite measures like the OpenSSF criticality score. The goal is to capture the community’s investment and attention in commercial open source projects, both to correlate commercial value with community metrics and to see how money entering the ecosystem changes contribution dynamics.
I came into this project quite skeptical. I’d heard the anecdotes about money entering open source and bad behavior following. What convinced me was large-sample evidence that adds the community perspective to the outcome metrics: the incentives of the founder, the investor and the community align. There’s a triple coincidence of desires that leads to outcomes that benefit everybody.
So the $26 billion is also money flowing toward the sustained maintenance of high-quality, critical, permissively licensed technology.
Bigger, faster rounds and better exits (8:15)
Matt: The $26 billion is only the beginning of the story. Over its life cycle, a company first needs to raise money. Then it needs to refinance: hit milestones, raise more, grow. Across the data, commercial open source companies have an easier equity journey than their closed-source competitors in infrastructure software. They raise bigger, faster and with better probability.
The third part is exit, because at the end of the day you need to return money to your investors and to yourself as a founder. It’s very liquid: more than 50% of companies that enter the VC funnel find liquidity at some point. And their exit value is significantly higher than their closest competitors’. When they exit, they return far more money to shareholders than closed-source equivalents.
Criticality, not popularity (9:30)
Andrea: That’s something I’d never heard before. There’s a lot of pressure on open source companies deciding what stays free and what they charge for, so it’s great to see the data shows the outcome is more money. Sam, on the community side: what do you look at? Downloads, stars, activity? What positions a project as a potential commercial success?
Sam: We looked for durable signals that users are spending time and attention on a piece of software because it’s meaningful, useful and, most importantly, critical. We start from the OpenSSF criticality score, a composite of several metrics: distinct contributors, contribution frequency, project age, update frequency, how many organizations contribute, and downstream usage such as dependent packages and projects. These go beyond popularity metrics like stars. The Linux kernel would obviously rank among the most critical projects in the world.
What we find is that value and criticality coincide. There’s a strong correlation, especially in later rounds, between a COSS company’s criticality score and its valuation. These companies aren’t just building something to sell to a handful of paying customers. They commercialize a subset of features, but they also sustain an important part of the open source stack.
That link matters, because most Postgres users care very little about the finances of the companies commercializing Postgres. They care whether Postgres is healthy. Is maintenance sustained? Am I dependent on a single company? Has that company built a bigger tent and brought new blood into the project?
Build your own project, or commercialize an existing one? (13:52)
GPS: Are these projects mostly started from scratch with commercial intent, or did people stumble into it?
Sam: It’s a mix. The majority of COSS companies create their own projects, but about a quarter build on an existing project they didn’t create, sometimes a large, popular project that originated decades ago in academia. There’s no single origin story. Creating your own is the dominant pattern, but not the only pathway.
Matt: And conversely, many companies start before they have a project. At seed stage, probably 50% of companies haven’t released a proper open source project yet. It goes both ways.
Stars are a vanity metric (15:01)
Andrea: Is there a set of metrics you particularly value, the ones that show adoption can turn into revenue?
Matt: It’s a question I get a lot. One of our most surprising findings was about stars: we looked at how star count relates to valuation or the ability to raise, and found no relationship. Stars are a vanity metric and shouldn’t be the North Star for funding.
That doesn’t mean community doesn’t matter; as Sam said, some metrics are very telling at later stages. But at the very early stage, we look at more qualitative aspects. Stars are one piece of information: they signal developer attention and some kind of traction. But they have to be read comparatively. Some categories are star-intensive, with dozens of similar projects. Taken at face value, 40K stars in a month looks huge and fundable, until you see that similar projects got the same traction. So I benchmark stars against comparable projects.
Then I look at who the users are. Are they in rich economies where IT buyers sit, the US or Europe? Can they translate into enterprise buyers? What kind of organization do they come from: indie developers, universities, large corporations? The latter is far more monetizable.
Finally, I look at the category, and whether there’s a case for the enterprise, not just the developer. Monetizing developer convenience alone has historically been very hard. Software development frameworks are great and used by every developer, but they don’t convert into invoices for organizations. In data or cybersecurity, by contrast, the appeal is to the developer and to the organization: governance, security, all the features that serve the company rather than the individual. That’s when a project is most likely to get financed and become a profitable business.
Translating open source metrics for investors (18:39)
Sam: That’s the investor side, screening for COSS companies. On the other side of the market, founders have a role in making these metrics palatable. Investors don’t necessarily care about open source; they want metrics they understand. Across the 2025 and 2026 studies, we find commercial open source is really efficient: onboarding new users is cheap, and the path from open source user to paying customer can be short.
So founders should map the open source user journey to investor metrics: CAC payback period, conversion rate from open source user to paying customer, time from first use to first contract. There’s no universal metric or standard yet, and not everyone captures this data. Downstream usage can be hard to measure, but not impossible: open source intelligence platforms like Scarf help founders identify who is using their product.
When these metrics are measured, they’re stunning for COSS: very efficient conversion rates and payback periods. Our message to founders is to make it easier for the investor. Track what you can, and translate open source metrics into investor language.
Pitch the value proposition, not the KPIs (20:55)
Andrea: Lots of stars don’t necessarily mean willingness to pay. Just because I use a project doesn’t mean I’ll champion a contract at my company. So what should maintainers measure, and how should they show it? Matt, as an investor, what would make you look at a project as a potential investment?
Matt: The answer is probably not a benchmark or a set of KPIs. If you want to build a company, you need to sell good software, open source or not. You need to pitch the value of what you’re building to your users, mostly organizations with business problems, and show how your software helps them achieve better results.
The bias of open source founders is that their starting point is the project. It’s sometimes hard to step back and say: now I’m a CEO, and I need to pitch a value proposition before any KPI or feature. That’s my advice: figure out the value prop and how to pitch it.
Open source then makes it tangible from day one. Unlike closed-source businesses, you already have proof of traction: people around the world using your software. You’re not just pitching a vision; you’re pitching a vision with tangible proof points.
Is the open source vs. money clash fading? (23:09)
GPS: There’s a popular trope that open source and money always clash, whether in the community’s view or among maintainers who insist on no money whatsoever. Do you see that changing?
Sam: We hope we’ve provided large-sample evidence to dispel that notion. I came in with exactly that suspicion: is the project going to change, is something going to get rug-pulled? What we’ve shown is that those things happen and they’re non-trivial, but they’re exceptions, not the norm.
The norm is healthy. The COSS company maintains the project and develops features under a permissive license, and commercializes on top. By and large, these are benevolent relationships between commercial interests and the community.
Matt: And the relationship is evolving. Nadia Eghbal, a former GitHub employee, makes a great point on this in Working in Public. Open source started with highly convicted, almost religious people around software, and it has shifted toward something more mainstream, almost influencer-like, where publishing code is more about getting adoption than about a strong ethos. In that respect, the relationship between making money and building open source software is healthier and more efficient than it used to be.
Andrea: Nadia is a former co-worker of mine; my first time at GitHub, I was lucky to be on her team. The book is Working in Public: The Making and Maintenance of Open Source Software. Highly recommended.
2026 findings: adapt, layer, build enterprise sales (25:23)
Andrea: Let’s talk about this year’s findings. Was there anything that surprised you? We’ll eventually have to talk about AI too.
Sam: The 2025 report was large-scale observational evidence, very outcome-oriented: we proved that open source is more efficient. This year we wanted to be more process-oriented and answer the prospective founder’s questions, to flesh out a commercial open source playbook: how to approach customers, how to structure projects, how to go to market and monetize.
The theme that stood out to me is adaptability. There’s no one-size-fits-all. COSS companies adapt as they scale. They don’t swap strategies out wholesale; they recombine existing ones at different levels and change their go-to-market motion as they grow.
And one truth came up constantly: the need to build enterprise sales capacity. Across the board, it’s a top concern, a top strategic priority and a top scaling obstacle. Investing in enterprise sales is the closest thing to a universal truth for COSS.
Stop agonizing, and founders are bullish on AI (28:33)
Matt: What struck me is that the founders we meet agonize, and I think that’s the right word, over certain choices. Which license? Which business model? Am I a PLG company or a sales-led one? It turns out you don’t have to choose. You layer them. On average, the companies we surveyed run two different business models, and the go-to-market motions that work best combine PLG and sales. That was mind-blowing to me. It’s much more flexible than you think, and you don’t need to agonize.
The second thing was AI. I was pretty sure open source founders were pessimistic about it, because I read Hacker News every day and it’s flooded with doom predictions about open source and developers disappearing. But the majority of founders we surveyed were very optimistic about AI, and the share who said they were pessimistic was very small. I wasn’t expecting that.
From research repo to company: the pyannoteAI story (30:25)
Andrea: Maybe the most pessimistic folks are the ones posting on Hacker News. And it’s a relief for founders to hear they don’t have to lock themselves into one license or one model. Can you share an example of a company that turned community traction into durable revenue?
Matt: Here’s a typical go-to-market motion for a company capitalizing on an already successful project. It’s in AI, so it’s relevant to people building right now. I’m a board member at pyannoteAI. They build models for voice AI, specifically speaker diarization: the input is an audio file, and the model separates the speakers so you can process it much more accurately downstream, with speech-to-text or voice generation. They’re the very low layer of the audio processing pipeline.
One of the co-founders, Hervé, is a great researcher. For every paper he wrote, he published the code on GitHub, which wasn’t common when he started about ten years ago. He grew a community around what became the most popular speaker diarization repo out there.
Then they turned it into a company. First, Hervé met a business person, Vincent: a deep technical founder paired with someone who wanted to build a business from the research. Second, they worked hard to understand the value the project could create for organizations. To use the open source model, you had to share some basic information. Are you an organization, a researcher, a solo developer? Why would you use it? Would you be willing to pay for it?
So when they started the company, they already had a massive survey of their users’ intent. The results were very good. They raised from us and other investors, and they’re now building a very profitable business around speaker diarization. It’s a great illustration of the journey from pure research and a repo to a growing business.
Andrea: So surveying their user base was the starting point that told them it would be a viable business.
Matt: And only an open source community can give you that. Without a prior project, you have no usage and no information. It’s far less tangible than having a large community telling you whether you can build a business.
Find the features that gain value at scale (35:08)
GPS: So if someone has a popular project and wants to find a commercial opportunity, is the best approach to research the community, find what they want, and start from there?
Sam: Identifying features that gain value at scale is a common pattern. It’s hard to prescribe one model, but if you’re a DevTools product, our report finds DevTools overwhelmingly prefer open core, so maybe that’s your path. The idea is to find features that are valuable to people who won’t DIY the open source solution themselves.
From the user’s side: I come from an impoverished academic background where you use the cheapest, most widely available tool, and as a data scientist I use DuckDB. As I scale, with a bigger team and more resources, I look at MotherDuck, a company that builds a commercial product on top of DuckDB and contributes back to the project. They solve pain points I don’t have as an individual but do have when working with others: MCP, security and collaboration features. That’s something I’d be willing to pay for.
Charging without alienating your community (37:22)
Andrea: Many maintainers worry what going commercial will do to their contributors and users. Am I going to upset people if I bring a commercial component into this project? How have you seen companies do this well?
Matt: What communities don’t like is big changes, and inconsistency with the philosophy of what you’re building. Project stewards need a very clear guiding principle on what is free and what is paid, and they need to stick to it. Changing those principles is where you risk antagonizing the community.
Last year I was with Samar from Temporal at re:Invent. They’re building a great company that’s doing extremely well, and one of Temporal’s guiding principles is that everything for the developer is free and that will never change, while everything for the organization must be paid for. That’s how they build a good business.
As long as you have a clear moral contract with your community and respect it, it’s fine to do business. People understand you need money to keep improving what you build. What they don’t accept is losing access to a product they use, or having to pay for something that used to be free. It’s hard, because you have to set those principles very early in the company’s journey.
Sam: And the community always has the defensive fork. If you make a license change that alienates part of your community, there may be enough momentum to fork the project. The overwhelming majority of COSS companies derive much of their value from the community’s interest and participation, from where the attention is focused. Alienate that and you risk ruining your project.
In the new report, we see a concerning trend toward source-available or non-permissive licenses as companies scale. That’s something we want to highlight and keep an eye on.
Early on, take the risk (41:21)
Matt: To step back: the common view is to make sure you don’t alienate your community. But at the very early stage, we see more companies fail because they don’t dare make the changes needed to turn a project into a business than companies that alienate their community and lose value because of it.
So early on, take risks with your community and your product, because you’re trying to find a path to a profitable business. Once you’re more mature, it’s a different challenge: you need to be very strong on your guiding principles and stay faithful to your community.
Where to find the report (42:10)
Andrea: I’m looking forward to this year’s report. Where can people get it, take part in the survey, or support the initiative?
Matt: Last year’s report is available on both the Linux Foundation and Serena websites. This year’s report will be released on 10 November, on stage at Web Summit, and then online. Our contacts are in the report, and we’re happy to hear from founders and anyone deeply interested in the space. We’ll keep running research, and AI is something we want to investigate more. Reach out on LinkedIn or by email; we’re always on the hunt for great founders building technical products.
Sam: We welcome critical analysis of the research, and ideas or concepts we didn’t explore well enough. Please take part in our future surveys, not just on commercial open source: the Linux Foundation runs a broad range of research, and first-hand experience from the community really informs it.
GPS: I feel hopeful. I once spoke with Francesco, one of the founders of CUA, and he said open source will always be how they operate, because their moat is the community, the people. That’s the message I’m getting from both of you.
Andrea: We’re leaving on an optimistic note. Go read the report and see how you can contribute. Where can people find you?
Matt: By email or LinkedIn, and I publish all my open source research at cossreport.com.
Sam: Email, LinkedIn and GitHub.
Andrea: Thank you both so much, and thank you, GPS.

