Our most recent VC guest at Data Driven NYC, Mike Volpi of Index, has had a pretty amazing last couple of years, with three of his venture investments going public: Zuora, Sonos and Elastic.
Before becoming a VC, Mike ran Cisco’s routing business where he managed a P&L in excess of $10 billion in revenues, and acquired over 70 companies (note: probably a pretty good way to make a lot of friends in Silicon Valley).
A partner at Index Ventures in San Francisco, Mike invests primarily in infrastructure, open-source and artificial intelligence companies, so he was a perfect guest to have at the event. In particular, he invested in two prior presenting companies: Confluent and Cockroach Labs (in which FirstMark is also an investor).
We had a really interesting conversation about open source, AI and venture capital. Here’s the video below, and l have jotted down a few notes as well, below the fold.
Notes from the chat:
On Mike’s consumer investments:
- Mike has a great track record in enterprise-focused investment, but perhaps surprisingly considering most VCs are either B2B or B2C focused, he has made some great consumer investments, including Sonos and Blue Bottle Coffee
- The theory is that you learn to empathize with the consumer if you invest in a consumer company
- Consumer companies also can be great conduit into interesting enterprise investment opportunities: Mike’s investment in Soundcloud (consumer) led him to discover Elastic (enterprise)
- But sometimes it is just about about the opportunity to invest in great entrepreneurs, sometimes doing random stuff, but VC is a lot about random stuff. The CEO of Blue Bottle Coffee
On building open source startups:
- Investing in OSS was originally very contrarian
- First principle is that it is very developer-centric, so you have to have the fundamental belief that modern software is sold through developers who discover the product and evangelize it
- If you believe that, then the next question becomes, how do you build a business on top of that, by charging *some* of your users?
- You have to be happy with the fact that maybe 80% of your users that don’t pay you
- But In OSS models, you spend less on marketing and sales and you give away that piece of software so that the developer becomes your sales person
- The first generation of OSS was all open, with a support business model. Key companies included MySQL, Red Hat
- The second generation was the open core business model.
- The idea is to build a suite of commercial products, which typically are the features or attributes of the software that would be used by users in production (observability
- Hortonworks, while a good investment, is a great counter-example – tough company to build because they didn’t have proprietary software, it was all open
- The third generation is based on the idea that you run a paying cloud service, on top of free open source software: MongoDB, Elastic, Confluent, Cockroach all do that.
- What about the threat of big cloud providers (like Amazon) taking the open code and creating their own cloud service on top of it? Startups are quickly shifting their license model to make it much harder
- Original model was based on Apache license – very open, anyone can use it
- Gen 2 is an Apache 2.0 license where you can have some proprietary elements
- New licence models, like Cockroach’s BSL, effectively says that big cloud providers cannot use the software to compete with the original open source provider
- In addition, if you’re concerned about Amazon, the other cloud providers, particularly Azure and GCP, are more than happy to have the original version of the software and paying you for the privilege of running the cloud version.
- The role of community has also completely changed:
- Community use to write all the coe
- Nowadays, 95% of the code are written by employees of the sponsoring company
- The community’s job has really morphed into product management, by providing feedback on what’s wrong or where priorities should be
On multi-product companies:
- Several visible companies in the current enterprise tech landscape are multi-product companies: Elastic, GitLab, HashiCorp, does Mike think it is a trend?
- On the whole, Mike is still in the camp of single product companies
- Even Elastic was mostly ElasticSearch for a while
- In the early days, creating repeatability is hugely important and requires focus
On trends in data infrastructure and ML/AI:
- We are still so incredibly early in the major theme of businesses extracting value from their data
- Big Data was chapter 1… about collecting and storing data… data science and machine learning are the continuum of that, all about deriving insights, in an increasingly automated fashion
- Over the next decade, this is the single most important theme for an enterprise investor… we are just getting warmed up
- Scale.ai is an example of a recent infrastructure ML/AI investment Mike did. Companies spend a lot of money on data annotation and labeling for purposes of feeding large amounts of relevant data to make their ML models work.
- On the one hand, they are a labor market of people around the world
- On the other hand, they have software tools to help with “pre-labeling”
- What are some ML/AI applications Focus on tabular data with schema fraud detection, inventory forecasting
On entrepreneurs and venture capital:
- What kind of entrepreneur does Mike like:
- Mike likes entrepreneurs that have felt the pain of the problem
- For machine learning, Mike prefers younger entrepreneurs, because so much of the progress has happened in the last 10 years
- Entrepreneurs that have used the technology at scale (Facebook, Twitter, Pinterest, etc) have an edge
- But ultimately it’s also about entrepreneurial
- What should entrepreneurs look for, if they are lucky enough to be able to choose among competing VC firms?
- Check references to verify that the VC can actually be helpful – references from companies that worked out, and from companies that did not.
- Branding does matter, to a degree… it opens doors in the beginning (customers,
- VCs are not your surrogate founders… their role is to help in critical moments, not do your job for you.