Mike Volpi, a partner with Index Ventures, had just secured a billion-dollar payday with Confluent’s IPO ($1.3 billion, to be more exact).
So when he told TechCrunch that venture capitalists weren’t too “hyped” about investing in open source in 2015, and that this changed only recently, he can be forgiven for a bit of revisionist history. I get it.
But VCs have been pouring money into open source since Red Hat lit the fuse back in 1999 when it went public, with no deceleration since then. What has changed is the kind of open source investment that succeeds.
Open source used to succeed by copycatting proprietary software. Now it’s winning by creating whole new categories. (And yes, Volpi has been successful by investing in both.)
More and faster
I worked for my first open source start-up (Lineo) in 2000. We raised roughly $60 million on the promise of embedding Linux in a host of new devices, from microcontrollers to personal digital assistants, the precursor to smartphones.
We were part of a host of open source companies raising money then, with more than $700 million invested in the build-up to Red Hat’s IPO in companies like TurboLinux, LinuxCare, etc. Lineo flamed out in 2002, sold for scrap, as did a few of these others, but open source soldiered on, with VCs anxious to keep swinging for the billion-dollar fences.
By 2005 The New York Times was writing about “open wallets for open source,” citing investments in JBoss, SugarCRM, MySQL, and more. They even quoted me, arguing that “you can guarantee that a year from now, there’ll be dozens of companies using the SugarCRM model” of open source monetisation, which relied on a mix of an open source core and proprietary add-ons/extensions.
I’d like to say I was prescient, but it was simple math: Open source was quickly becoming dominant with developers, and VCs recognised this. Some VC firms, like Volpi’s Index Ventures and Peter Fenton’s Benchmark, were much quicker to recognise the trend and invested heavily.
Volpi is right to display some pride of ownership in the category, having invested in Confluent, Elastic, Cockroach Labs, Hortonworks, Kong, and Starburst. His firm could add MySQL and others to that roster.
Even slower-moving VCs caught on, to the tune of billions in investments and tens of billions in returns.
In 2018 alone, IBM’s acquisition of Red Hat generated $55 billion in open source returns, but Joseph Jacks, general partner at OSS Capital, keeps a tally of all returns and, although not all of the companies he credits as open source fit that description, there’s still enough there to put the final sum well over $100 billion.
Back to that TechCrunch interview. It’s demonstrably, dramatically untrue that open source didn’t really see much VC action until 2015. Somewhat hilariously, Andreessen Horowitz investor and former XenSource CEO Peter Levine wrote in 2014 that there would never be another Red Hat, even as multiple “Red Hats” were being funded.
It’s also untrue that the open source “market has become increasingly picked over as more investors pile into backing the model.” It’s true there are more VCs funding open source, but that’s simply because there’s a lot more open source to fund.
And that’s where the real story is.
Building the future on open source
Many of those early open source investments were companies like Alfresco, one of my start-ups, which aimed to offer an open source alternative in an established software category. In the case of Alfresco, it was document management. We billed ourselves as the “open source Documentum".
This made marketing easy: Someone else had already spent the money to create the category, and all you had to do was commoditise it through open source. We had open source app servers (JBoss), open source customer relationship management (SugarCRM), open source business process management (BonitaSoft), and more.
But then a funny thing happened on the way to the market. Open source got innovative.
By 2013 open source luminary Mike Olson could declare with a straight face that “No dominant platform-level software infrastructure has emerged in the last 10 years in closed source, proprietary form.” He said that “there’s been a stunning and irreversible trend in enterprise infrastructure” toward open source. He was right. But this actually made picking winners harder for VCs, not easier.
Take Confluent, the primary sponsor of the now popular Apache Kafka project. CEO Jay Kreps, in a revealing Twitter thread, reminds us that this wasn’t always the case. “We wrote the initial Kafka code base at LinkedIn in 2009-2010. In 2011 we released the initial Kafka code as open source to...resounding silence. No one cared!”
Put another way, very few were thinking seriously about streaming data back then. Kafka is a publish-subscribe messaging system used for processing real-time streaming and batch-based data processing.
Around the time of Olson’s statement, I spent a year working for Nodeable, which tried to build a business on Apache Storm, giving developers a way to tap into stream-based processing of data. It failed. Lots of start-ups fail.
Perhaps we did a poor job marketing the company, and maybe our product wasn’t great (both likely true), but mostly we struggled to convince developers that they needed to embrace streaming data.
No one needs convincing today. Streaming data is becoming the norm, and Kafka sits at the heart of it. Back to Kreps: “We had a lot of big ideas: building a data architecture around events, moving from batch to real-time stream processing, doing this around a kind of commit log that brought together real-time change and data storage. We knew it could be a big deal!”
Kreps continues, “But how do you get people to think in a different way about data? Maybe you need a company to help drive it, we thought. But there’s no point starting a company if your product is all about a paradigm that no one wants. How to test that?”
In December 2013, Kreps wrote a blog post about how “event streams could act as a central nervous system for companies and the use of data could transition into real time with stream processing.”
“If the blog is popular,’ he remembers, “maybe there could be a company there. If not, well, maybe this won’t be able to get going. The blog post was really popular! We decided to make the jump and start a company. Confluent was born in September 2014.”
This brings us back to investing in open source. There was once a thought that investing in open source was relatively cheap because you just piggybacked on a project that had taken off with developers.
That still happens, but when open source companies are creating categories, not merely aping them, much more capital may be required. This is certainly the message that MongoDB CEO Dev Ittycheria has given, citing the more than $700 million the company has invested in the product since the project’s inception. There was no document database product for MongoDB to mimic. It has had to spend.
Ditto Confluent, which raised $455.9 million before its IPO. Or Neo4j, an open source graph database company, which has raised over $500 million, including a recent $325 million Series E. And so on.
So no, there never has been a slowing in open source investment since Red Hat lit the fuse in 1999. If anything, it keeps accelerating. And because open source companies must spend heavily to create new markets rather than copying old markets, we can expect the funding to get even bigger, faster. There’s a lot of money in free software, it seems.