The Iceberg of Overnight Success
🤫 “All overnight success takes 10 years”- Jeff Bezos
A few weeks back, I wrote about the overnight successes we’re seeing in AI. I referenced a secular trend towards faster growth. We’ve all seen the famous chart for time to $100M ARR, in which Cursor topped Wiz, Deel, Ramp, Slack, etc., etc.
BTW, Cursor is now apparently at $300M ARR 😵💫. We’ll come back to that.
What got me thinking about the overnight success Bezos quote was a conversation I had with someone about Wiz. I learned the founding team, led by Assaf Rappaport, had built a security startup before, Adallom, founded in 2012, which they sold to Microsoft in 2015. Rappaport then spent over 4 years as general manager of Microsoft’s Cloud Security Group. It was the culmination of this experience, having built a security startup, then working with some of the biggest enterprises in the world while at Microsoft, that meant Wiz could be so well positioned to close huge deals so quickly after starting. When did Wiz hit $100m in revenue? August 2022, 10.5 years after Adallom was founded.
This pattern exists all over the place when you start to look for it. I’ve mentioned previously the $ 230 m+ Warpstream acquisition by Confluent last year, just 1 year after the company was founded. Well, the CEO, Richie Artoul, was working on Husky, an object-storage-backed analytical database, at Datadog from 2020 to 2023. Before that, he led the development of M3DB, Uber’s distributed time-series open-source database starting in 2015, 9 years before Warpstream’s acquisition.
We’ve talked a lot about Iceberg in this newsletter. Tabular was acquired for over $1B less than 3 years after its founding, but the founders created Iceberg back in 2016. Ryan Blue was working on data formats before Iceberg, starting in 2013. Again, 10+ years.
Elsewhere in data, we see this again and again.
Snowflake was a rocketship to $100m ARR but founders Benoît Dageville and Thierry Cruanes spent over a decade at Oracle working on database infrastructure prior to starting the company.
Confluent once topped the Bessmer time to $100M AR chart when they hit the number in 2019, but the origins of the underlying Kafka open source project span back to 2010 at LinkedIn, with pain points around real-time data preceding that.
Databricks also hit $100M ARR in 2019. Care to guess when the Databricks founders created the Spark open-source project at Berkeley? 2009…
I wrote previously about modern data stack companies hitting $100M ARR. Tristan Handy founded dbt (née Fishtown Analytics) in 2016, but I suspect his time at RJmetrics, going back to 2013, can be counted towards the origins of dbt, ~10 years in the making. I estimated Fivetran hit $100m in 2022, 10 years after its founding in 2010.
While AI seems different, and Cursor is perhaps an exception that proves the rule, OpenAI itself took roughly 7 years from its 2015 founding to hit $100M ARR in 2022.
I’ll risk overusing the word iceberg, but for every overnight success story, there seems to be 80% of the hidden effort hiding under the water.
If you’re out there building, keep that in mind.
Warmly,
Paul Dudley
P.S. I may have written this article as founder therapy. We’re a fast-growing, nearly 3-year-old company, and the old adage comparison is the thief of joy certainly holds true when Cursor is hitting $300M in less than 2 years.