- Ali Ghodsi, the CEO of artificial intelligence firm Databricks, was so paranoid about an economic downturn that the company raised two rounds of funding in under six months from the likes of Andreessen Horowitz.
- Now the company has "well over half a billion dollars on its balance sheet" left to ride out the recession to an IPO, likely next year.
- This week the company is unveiling a new concept called "the data lakehouse" that will allow companies to search data more quickly and efficiently than previously possible.
- The new approach is a combination of expensive, easy-to-analyze data warehouses and cheap, harder-to-manage data lakes.
- Visit Business Insider's homepage for more stories.
Ali Ghodsi, the CEO of Databricks, a startup that helps companies move massive data sets to the cloud and analyze them, spent the last several years conducting "the sky is falling" exercises where he ran through ways that the company could be "prepared for the worst."
The coronavirus crisis has justified that paranoia, and this week, in the middle of a downturn in which many companies are just trying to survive, the well-funded company is rolling out a new vision for Databricks' product that makes it easier for customers to analyze their data.
Because of its doomsday planning, the company raised a war-chest of $897 million in venture capital, most recently at a $6.2 billion valuation, including through two funding rounds led by noted tech investors Andreessen Horowitz less than six months apart. The company still has more than $500 million of that funding left, which it hopes to ride all the way to an initial public offering.
With 2019 annual revenue of $200 million that's "growing very fast," Databricks plans to be "IPO-ready next year," CEO Ghodsi told Business Insider. He says that investors accused him of being like "the boy who cried wolf, asking for money in case of downturn" last fall when he was raising a $400 million F round, but it turned out to be a prudent move. Similarly, Databrick's decision in mid-2019 to back out on plans to lease a $120 million office space in San Francisco because the team felt like it was "likely that the market's going to take a downturn," now seems like incredible foresight, too.
Ghodsi attributes part of his "worst-case-scenario" planning style to the childhood experience of having his family flee their home within 24 hours to escape the Iranian Revolution in the late 1970s.
"You kind of always know things can suddenly take a turn for the worst" after an experience like that, he says.
This week the seven-year-old startup, which has 1,300 employees and 6,000 enterprise customers, is riding high as it hosts an online conference where it will officially launch a product strategy around the idea of a "data lakehouse." Ghodsi describes the concept as "a new paradigm that could bring a seismic shift to the market."
For decades companies kept data in "warehouses," often costly on-premise servers, where it was periodically structured and filtered for specific business information purposes, like analyzing sales data. "But they were flying blind between quarters," Ghodsi says, because the data was highly structured around past events. More recently, "data lakes"— vast sets of raw, unstructured data — have become popular too, because "it's a really cheap way to store data," Ghodsi says. Data lakes might hold key insights, but they can be very hard to manage. Databricks is rolling out a combination of the warehouse and lake – the lakehouse — that combines the structure of the warehouse with the vast, up-to-date data of the lake. This will allow companies to quickly search vast data sets and put the information to use predicting future trends.
Ghodsi says that data lakehouses will allow companies to more easily analyze large data sets eight times faster than is now commonly possible.
While some customers have used the product in beta for a few months, it will officially roll out publicly this week during the Spark+AI Summit 2020. The free online conference includes participation from Microsoft, Starbucks, and data expert Nate Silver, as well as sessions on AI's role in social justice and policing.
Data lakehouses will allow "millions of companies on the planet — eventually, down the line — to be able to ask questions not just the past, but also the future," Ghodsi says.
Join the conversation about this story »
NOW WATCH: We tested a machine that brews beer at the push of a button