In this post I’ll cover my formative career learnings that led to what I’ve dedicated my life’s work to, hopefully for decades to come.
Early Days and Learnings: “Open-Source Companies” are categorically different
Early on in my career about 10 years ago I was working at Talend, an enterprise software vendor. This was fairly early days for Talend when the company was starting to go to market in the US after having set a decent footing in France where they were founded.
Talend was (and is) a special kind of technology company: one that builds its business on a core set of Open-Source Software (OSS) tools/projects and commercializes them by offering paid commercial versions of those technologies as a managed service along with additional proprietary value-added functionality, also delivered on-prem —the end customers for Talend are data engineers / developers working in corporate IT departments at small and large companies.
Interestingly, Talend was quite self-aware of its identity early on (2010~). The term “Open-Core” (OC) was used to describe their product strategy and differentiation. Sales reps and sales engineers alike (I worked in both areas) were educated about the peculiarities of this model and had to educate end users and customers as well. We also had to painfully educate many highly conservative and risk averse procurement / legal departments of IT organizations inside Enterprises about the OSS licensing dynamics inherent in this OC model, in stark contrast to the traditional fully proprietary (Wikipedia currently conflates terminology here, I will address that later) licensing approach of software companies from yesteryear.
I learned a ton at Talend. This is really where I cut my teeth in understanding the Enterprise software market, IT in general and the middleware stack that is so crucial to keeping the lights on at the largest companies in the world.
Experientially, what I learned the most from Talend during a very high growth period (I joined when revenue was in the low single-digit millions and left then it was nearing the mid-double-digit millions) was Talend is quite a different kind of software company. Very different, in fact. This difference is “felt” so strongly once one has worked at one of these companies. Everything about Talend was different! Product development. Sales approach and motions. Engineering management and workflows. Finance operations. Pre-sales execution. Hiring approaches. Culture. Strategy. Channels. Services/Consulting. Marketing…Literally almost ALL the functions! They each looked and felt very different from the inside looking out. From the outside of the company “looking in”, however, it would be very easy to label Talend as just another software company selling a product. Maybe with a clever distribution model to acquire leads more efficiently (this OSS thing). All too many people in the tech industry make that mistake, unless they have seen / been on both sides of the spectrum (closed core vs. open core) from the inside.
From the outside of the company “looking in”, however, it would be very easy to label Talend as just another software company selling a product. Maybe with a clever distribution model to acquire leads more efficiently (this OSS thing)…Too many people in the tech industry and even more broadly make that classification mistake, unless they have worked at both kinds of companies (Closed Core vs. Open Core) and reached concrete conclusions from the inside.
After Talend, I did a short stint at TIBCO Software (basically the opposite kind of company: closed core / “old school” fully proprietary software company) working again in sales, selling their entire product portfolio to a number of strategic accounts in the bay area. While at TIBCO (2012~), the company was still in denial about the profoundly disruptive shift happening at the infra software layer… driven by OSS. Inside TIBCO, I was known to beat the OSS drum quite loudly. An internal Yammer (ala Slack) competitor called tibbr was the mechanism I used to describe as best I could to the entire company why TIBCO should investigate and better understand OSS across many dimensions: competitive differentiation, product development ideas, overall innovation, etc.
After leaving TIBCO before the company was taken private and then ultimately acquired by Vista Equity, I worked at Enstratius Networks — an early small software company that sold a product to IT for the purpose of unifying cloud management APIs (founded in 2008, but I joined a year or so before the days of Docker, Kubernetes, etc). These Cloud APIs that dominated in those days (6+ years ago) were things like Eucalyptus Systems, CloudStack, VMware’s vCloud Automation (formerly DynamicOps) and, of course, OpenStack! Nowadays, the lingua franca of the Cloud is Kubernetes… Directly after Enstratius, I published a deck describing the tools/products landscape perspective at the time in this area of the IT stack.
Skipping lots of history, after Enstratius where I was fortunate to have learned about Docker, Mesos and Kubernetes, I was also lucky enough to be in a place to help personally fund and start the first enterprise startup company focused on commercializing the Kubernetes project shortly after it was released. During this I also started the Kubernetes conference KubeCon, helped ultimately sell and integrate Kismatic into another software company, worked shortly after that as a corporate EIR on the Rook project and then started another company as well.. I then started blogging about a spreadsheet I had created years prior on the side.
Speeding up even further, if you’ve been following my very active twitter stream over the past year or so, it’s easy to publicly tell that I’ve been busy founding (all in parallel) a new kind of investment firm, launching a new conference, a media site and a new kind of startup company accelerator…. And maybe a few more things that have yet-to-be-clarified sufficiently in my head to actually do. All of these efforts relate and compliment each other. They are, however, intended to be loosely vs. tightly coupled.
I believe that encouraging and working to serve this kind of open core approach in technology company building represents the future of the entire tech industry due to many fundamental advantages of this model.
Why am I now working on all of these risky things? Because I believe that encouraging and working to serve this kind of open core approach in technology company building is the future of the entire $10T+ tech industry.
Each of these aforementioned initiatives has one thing in common: intense focus on something I have decided to dedicate my working life to. Something that I have come to believe has become more clearly codified increasingly so over the last decade, from 2010 to now. Something I like to call “Commercial OSS”. COSS, for short.
Why the term COSS?
Because I do not like the term “Open-Source Company”. This term is false. Open-Source actually means something quite specific: when applied to software source code, Open-Source means that literally anyone without discrimination can see, run, modify and/or distribute the code at any time. In this sense, Open-Source is actually about enabling permissionless rights more than it is about anything else. Open-Source is NOT freemium. Freemium is permissioned. Open-Source is permissionless.
Companies are actually inherently not Open-Source. So, calling a given company “Open-Source” is oxymoronic. It is like calling a feather heavy. Companies absolutely cannot be “seen” from the inside, “modified” by anyone or “commercialized” by anyone, at any time. In fact, companies must and do inherently discriminate in order to find product market fit, hire the right people, find the right investors, effectively decide on the right strategy and more.
I believe COSS needs to be understood as its own category.
Companies that fundamentally build on an OSS core technology are different… on nearly every level. As a consequence, I believe the world needs a new kind of first-principles way of understanding and describing and studying these companies. We need new terminology. We need a new kind of lexicon. We need a new kind of thought process. We need to respect and appreciate their fundamentals.
We need a new COSS movement that inspires many to understand why this open core approach is holistically better than the old way of doing things.
Working Towards an Open Future
This vision of encouraging clarity in COSS is all about promoting a future world directly embracing Open-Source across all industries wherein Open-Source eats Capitalism, Culture, Software, Education, Healthcare, Insurance, Banking, SaaS/Cloud, Consumer applications, all internet applications, AI/ML, IoT/Communications, Enterprise IT, Hardware, eCommerce and more. In short, a future where Open-Source Eats Everything (as mentioned, including Software).
Over the past 25 ~ years, particularly in digital (“software”) technology, Open-Source has driven the vast majority of fundamental innovation. What I mean by fundamental innovation refers to the core building blocks that software creators / developers use to construct their applications. Libraries, packages, frameworks, databases, middleware, runtimes… all of these picks and shovels represent 90%+ of lines of code that digital technology (software) is comprised of in the world. From all the apps you click on your smartphone and use everyday to the software that runs planes to the operating system in your self-driving car.
Over the past 25 ~ years, particularly in digital (“software”) technology, Open-Source has driven the vast majority of fundamental innovation.
The < 10% of code that is not Open-Source is hand written business logic. This logic is what you can think of as “differentiated work” that makes applications function for a particular purpose. All of the “undifferentiated heavy lifting” (great, under-appreciated must-read paper published in 2005 (!!) by my friend and board partner Bruce Perens, one of the founders of the Open-Source movement) in digital technology has already been decoupled and removed from the critical path of work that a developer needs to implement by hand… it is simply available to find on the internet and assemble together. This “assembling” of Open-Source code is where most of the work happens.
I believe that Open-Source has also enabled the creation of more economic value than is possible for anyone to really measure.. perhaps into the low-to-mid-single-digit quadrillions of dollars.
Measurement is important here. Economists and the industry overall have clear, agreed-upon ways of measuring value capture: company valuation, revenue generated, profits made, markets that are sized, etc… these are quantitative things. However, we do NOT have ways to measure value creation: how productive a given technology amplifies human abilities, reduces pain, maximizes leverage.. these are qualitative aspects that do, however, translate to real gains.
Without going too deep, this belief is important to note because value is captured and value is created, but those two things are quite different and independent of each other (creating vs. capturing). Most people use the term value when referring to monetary/economic things — but value also critically relates strongly to time, energy, resources, health, efficiency, productivity and more. Not just for companies, but for individuals that actually create the value that companies capture. Customers also capture value.
Indirect and Direct Value Capture
As a primer, it would be useful to read this post to better understand the following:
Starting in the late 1980s, I believe that the OSS value capture pendulum initially started to swing along one end of a sort of economic spectrum: consumer applications, enterprise software, networks and cloud computing are powered (90%+ of code) by OSS… this is the indirect value capture end of the spectrum.
The other end of this spectrum refers to direct value capture. I think the pendulum is swinging in that direction and will continue to for decades to come for many fundamental efficiency reasons.
The value created by OSS in those aforementioned indirect areas (social networks, eCommerce, enterprise software, communications, cloud, etc) has been largely captured, so far, by companies who have consumed OSS, building fully proprietary offerings based on OSS code — those proprietary offerings / products are shipped and sold not by the individuals / creators of the OSS that actually created the value… but by tech companies that charge for a higher level service or product. This dynamic naturally implies that companies are largely responsible for digital technology innovation. I do not believe this to be true. Instead, I actually believe that the indirect value captured by companies has been exploiting the work of the true innovators (OSS creators) and that those companies themselves have profited due to an abstraction advantage in the world due to the lack of general societal and end user consumer understanding and empathy for where the true innovation is actually coming from. Further, I believe that it is fair to say the following, at least in digital technology innovation: Companies are the wrong abstraction for innovation output but the right abstraction for value capture input.
I’d say that today the aggregate value of all the worlds tech companies of all sizes that indirectly capture OSS value to be between $10–15 Trillion: I’ll refer to these companies as “Closed Core” since their core essence/IP/technology is fundamentally locked/proprietary/fully-centralized and compiled out of a patchwork of hundreds or thousands of OSS projects:
- In networks: LinkedIn, Facebook, Twitter.
- In SaaS: Salesforce, Atlassian, Twilio.
- In eCommerce: Amazon, Shopify, eBay.
I also estimate that 95%+ of all technology companies today are Closed Core.
Note that as mentioned previously, I also believe that the value OSS has created ($3 Quadrillion or more) far eclipses the number that has been captured indirectly AND directly ($10–15 Trillion ~)…
Direct OSS Value Capture
Over the coming decades, I believe two things will occur:
1) OSS will continue to drive the vast majority of real fundamental digital technology (the world of bits) value creation and, increasingly, physical technology (the world of atoms) as well (see RISC-V beginning to shake up the $500B+ semi/chip industry).
And, 2) The value capture pendulum will gradually and increasingly swing in the other direction, shifting the world towards direct value capture of Open-Source, thanks to the rise of a new kind of small yet exponentially growing category: Commercial Open-Source Software — “COSS”, as I call it.
What is COSS?
Here’s an elaboration on my views spanning 50 years of history in Permissionless Software Innovation: Free Software, Open-Source and Commercial Open-Source eras since 1983 to 2030~: https://COSS.Media/five-decades-of-permissionless-software-innovation-movements/
Contrasted to Closed Core technology companies, “Open Core” oriented companies (COSS companies) capture value directly: they build on a largely singular Open-Source core which itself justifies the existence of the company around it.
COSS companies, fundamentally and definitionally would have no compelling reason to exist without their Open-Source core also existing in parallel.
COSS companies, in my view, are largely misunderstood by nearly all constituents: founders, VC and the broad ecosystem around them. I believe this is the case because COSS companies have materialized through properties of accidental emergence and evolution since the first COSS company was founded 30 years ~ ago.
COSS companies, without their given congruent and largely singular OSS core, simply would not exist. This specific definitional clarity is agnostic to the business model of the company, the relationship between the company and the project, and any other aspect.
Thus, COSS is an abstract superset category. This is important to understand. This type of company category is quite different on many dimensions as I hinted at early in this post:
Index the Value of COSS as a Category
Today, the COSS category is a small, but growing exponentially if the data over past 7 years of our close tracking is any indication of reality: the value of the total COSS category is now $150 Billion ~. In 2013, it was 10X smaller.
I’ve been tracking the top COSS companies by revenue ($100M/year minimum) closely since 2013 in an actively maintained spreadsheet called the COSSI index.
In this index, we look at a number of dimensions: when was the company founded, what the monetization model is, amount of VC dollars raised, size of company by employee count and much more.
This sheet is references and cited very frequently in the VC industry and across research firms and COSS founders alike.
Future Growth Driving COSS
COSS as a company category would fundamentally not exist without OSS… So, I strongly believe the two primary levers key to increase the rate at which Open-Source truly eats everything are
- the sheer number of developers who are increasingly familiar with all levels of the stack.
- the collective levels of end user empathy as a result of more reliable and trusted OSS technologies being available as the world moves further into a technology-powered digital future.
Today, there are 30M ~ developers on earth. This number is growing 30–50%~ YoY (or faster). By the year 2030, I believe there will be 500M + developers creating technology and software driven primarily by OSS.
With this tipping point somewhere along the spectrum of 300–500M software creators, I predict that COSS will become the default tech company model for creation and innovation up and down the stack (all apps and beyond). COSS will also become better understood across all asset classes, stages of capital, level of founder maturity and more.
What about this Cloud computing thing? Here’s my take:
COSS as the dominant path for digital tech innovation and consequential company creation
With all this background, it probably makes sense now why I’m working on building the following things:
- OSS Capital: the world’s first investment fund focused exclusively on serving and investing in this category
- Open Core Summit: the COSS ecosystem community (online and physical events. globally)
- COSS Media: Blogs, Videos, Research, Podcasts and more
- COSS Accelerator: A new kind of startup accelerator focused on helping OSS creators learn to become COSS founders and launch their new companies
All of these things can benefit many kinds of stakeholders (investors, team members, enterprises, analysts, etc), but the primary constituent in focus is the OSS creator: we want to help OSS creators learn how to become company founders to learn about the path to capturing some of the value they create in an effective way. The difficulty and complexity in founding and scaling a traditional software startup has dropped significantly over the past decade.. interestingly, inversely in tandem with the number of COSS companies that have organically learned how to scale and grow into monster businesses.
All of these efforts are exclusively in service of demystifying this COSS model, category and approach. They aim to execute a kind of mass knowledge transfer for the next generation of startup founders who will begin with an OSS project they create and maintain. Those creators should be the founders of the next great tech companies. That’s the future I want to see in the world.
We need a ton of help building all of these things. We are building them all in the open. By “We”, I mostly mean me… with the help of some close friends, supporters, private investors and people who believe in me. I have been and have been going all in on this thesis for 18+ months. We need a ton of help. So far, the most progress towards this mission has been in all the areas except the Accelerator which we hope to launch soon.
More on the motivating thesis for the COSS Accelerator:
If you are interested in supporting this mission as an investor, founder, creator or someone who sees and stands for the same kind of future — please reach out! Much more information on exactly progress and execution is available and the goal is to build as much of this in the Open-Source way as possible. Running an investment fund with a fully Open-Source philosophy is not allowed by the SEC, yet, but we hope to make progress there eventually as well!