Working with world-class entrepreneurs is what I love most about the venture business, as anyone who has read any of my past blog posts has probably figured out. They inspire me. They deliver the returns to our Limited Partners that keep us in business. And in ways big and small, they change the world around us. At NEA, we’ve been fortunate to partner with some of our most exceptional entrepreneurs to build not just one company, but multiple companies. Ion Stoica is one of those entrepreneurs.
Ion’s first company, Conviva, set out to re-imagine how Internet video works and was one of my earliest investments at NEA. I didn’t hesitate to back his next company, Databricks, which has become the de facto processing platform for big data. Ion wasn’t done yet, and neither were we—last year we invested in Anyscale, the company powering open-source distributed computing platform Ray, which he founded alongside two of his PhD graduate students at Berkeley’s RISELab, Robert Nishihara and Philipp Moritz. Today, I’m thrilled to announce that NEA is leading a $40M Series B round in Anyscale.
The need for distributed computing has become increasingly acute. Widespread adoption of artificial intelligence and machine learning in application development translates to massive computing workloads—more than a single machine can handle, and far more than Moore’s Law ever anticipated. Anyscale is putting the power of Ray in developer’s hands with an end-to-end computing platform that makes building and managing a scaled application across clouds as easy as developing an app on a single computer.
Although the potential for a product like this has existed since the advent of cloud computing, ‘Platform-as-a-Service’ offerings (the term we bounced around back then) never really took off as the cloud vendors rapidly developed their own solutions, which largely targeted ops instead of developers. In today’s world of empowered developers, serverless computing, and the ever-expanding volume of data accompanying AI/ML proliferation, the need for distributed computing across multiple clouds has both intensified and expanded.
With Anyscale, developers can easily build and serve their applications on any cloud or multiple clouds. This is increasingly critical as more than 80% of public cloud users work with two or more vendors. Although serverless computing has made this possible for many stateless applications (i.e. not requiring data storage), Ray solves a key deficiency of serverless computing by working on all clouds as well as on-premise clusters, and supporting stateful applications—a foundational requirement for practical use of the technology.
Anyscale’s path so far bears all the hallmarks of a truly remarkable company in the making. Ion’s co-founders are incredibly talented (so much so that I already find myself wondering what they might do next). I think they’ve got the timing right on a rapidly growing and potentially massive market. They also have great momentum when it comes to their open-source offering, with more than 13k stars on Github and over 360 contributors. This is a great signal that the community not only appreciates and values Ray, but is also committing substantial time and talent to making distributed computing simple and pervasive.
The promise of Anyscale reminds me of what made VMware so special when I worked there in the early 00s. The VMware workstation could be used by developers for coding and testing, while the virtual machine could be deployed to production seamlessly with the VMware server. A big part of the VMware success story centers around the developer-focused approach of their workstation, which is echoed in Anyscale’s solution enabling a developer to easily build, deploy and scale applications across clouds.
This is a technology that’s time has come. It’s also the next chapter in a partnership that has spanned almost my entire investing career, and I couldn’t be more excited to support Ion, Robert, Philipp and their team on this journey.