Jordan Tigani — a founding engineer at Google BigQuery, Google’s fully managed data analysis platform — was working as the chief product officer at SingleStore when he noticed that the vast majority of database workloads were small (less than 10GB in size) and low-bandwidth. While vendors were building for massive data sets, the term “big data” was becoming a misnomer thanks to recent advances in hardware, the way Tigani saw it.
Around the same time, Tigani got in touch with Hannes Mühleisen, the co-creator of the lightweight database platform DuckDB, to toss ideas for a paid service back and forth. Seeking to launch a product for developers with light database requirements, Tigani — with Mühleisen’s blessing — began building a DuckDB-based cloud service. The service became the cheekily named MotherDuck, a startup independent of the original DuckDB that’s focused on commercializing open source DuckDB packages.
“Users want easy and fast answers to their questions — they don’t want to wait for the cloud,” Tigani told TechCrunch via email. “The fact is that a modern laptop is faster than a modern data warehouse. Cloud data vendors are focused on the performance of 100TB queries, which is not only irrelevant for the vast majority of users, but also distracts from vendors’ ability to deliver a great user experience.”
It’s a classic playbook — take an open source tool and build a service on top of it. But while it might not be original, Tigani’s plan has already paid dividends. MotherDuck today announced that it raised $47.5 million across seed and Series A rounds, valuing the company at $175 million post-money.
Redpoint led the seed while Andreessen Horowitz (a16z) led the Series A — other investors include Madrona, Amplify Partners and Altimeter. Tigani says that MotherDuck wasn’t planning to raise the Series A so soon after the seed, but did so at the urging of LPs — and for the opportunity to work with a16z.
“With this funding, MotherDuck is able to build out their world-class engineering team and add a go-to-market function to provide a cloud analytics platform for organizations that want to use DuckDB in an evolved way,” Tigani said. “At the same time, it allows DuckDB to continue to be a vehicle for academic research.”
Tigani claims that MotherDuck’s service — powered by DuckDB, which HackerNoon once described as “mutant offspring of SQLite and Redshift” — allows practitioners to start answering questions from data faster than most existing tooling. It uses local computing resources in concert with the cloud, driving data analytics and other data-heavy workloads.
That’s in contrast to typical data warehouse systems that offer reporting and tools almost exclusively for enterprise-scale analytics.
As Madrona’s Jon Turow explains in a forthcoming blog post (TechCrunch got a sneak peak), MotherDuck uses a “hybrid execution” technique to query a data set that’s spread across multiple places. Some of the data might be on a developers laptop, some in the cloud instance and the rest in a different cloud, but MotherDuck makes it possible for a dev to query the combination of these sources. “The platform intelligently decides where to operate upon each bit of data to minimize the costs of compute and data transfer,” Turow writes.
The data warehouse concept has existed since the ’80s, but it’s risen to prominence in recent years as companies shift their workloads to the cloud. There’s startups like Firebolt and Hydra, which aim to become the open cloud data warehouse of choice for large companies. Panoply, another player in the data warehouse space, has taken a different approach, developing tools that make it easier for businesses to analyze their data with standard database queries.
While Tigani sees MotherDuck as a competitor in the data analytics market alongside data warehouse vendors, he positions the platform as the technological superior alternative.
“The high efficiency of DuckDB will allow MotherDuck to be cost-competitive, while also being more performant for most data workloads,” Tigani asserted. “Advances in CPU, memory, disk performance and networks are making existing architectures obsolete. Large distributed analytics clusters are no longer necessary due to these advances. Single-node DuckDB can often be much faster, cheaper and simpler than these distributed systems.”
The DuckDB team is involved to a degree with MotherDuck, which in turn is a member of the DuckDB Foundation, the nonprofit that holds much of DuckDB’s IP. DuckDB’s own commercial arm, DuckDB Labs, is a shareholder in the company and contributed code to the cloud platform. Tigani assures me that DuckDB will continue to be freely available under a permissive MIT license and that the original DuckDB team will build, maintain and promote the core DuckDB codebase going forward.
Fueled by the fresh capital, MotherDuck plans to expand its small workforce from 13 people to 18 by the end of the year. When asked, he declined to answer questions about the size of the startup’s customer base or revenue, saying it’s too early.