The $38 billion business that became Databricks Inc. started with the swipe of a credit card.
The card belonged to Ali Ghodsi, Databricks’ chief executive officer, and he and his co-founders believed in 2013 that the quick and easy purchase of cloud storage on Amazon Web Services Inc. would pave the way toward faster customer adoption for its fledgling data lakehouse service.
“When I swiped my personal credit card on AWS, some of my co-founders did the same, and we started building,” Ghodsi recalled. “You can get end users much quicker access to the thing you are building. We just built this startup from nothing on AWS without even talking to anyone there.”
Ghodsi’s story is testament to the rise of “superclouds,” substantial cloud-based businesses, such as Databricks, Snowflake, VMware, and MongoDB, that are built on top of major cloud provider platforms. In advance of AWS re:Invent, which kicks off in Las Vegas this week, SiliconANGLE industry analyst and theCUBE host John Furrier sat down with Ghodsi for an exclusive interview that covered his firm’s evolving partnership with AWS, meeting customers’ multicloud needs, and why major cloud providers must innovate.
Robust platform on AWS
About four years after Databricks’ founding, AWS noticed that the startup was driving significant amounts of compute and data resources on the cloud provider’s platform. The company reached out to Ghodsi and a partnership was born.
“Around 2017, they suddenly saw this company with maybe $100 million in annual recurring revenue pop up on their radar and it’s driving massive amounts of compute, massive amounts of data,” Ghodsi said. “It took a little bit in the beginning for us to get to know each other, and over the years the partnership has deepened. We have multiple offerings on their Marketplace, we have a native offering on AWS.”
What Databricks successfully built on top of AWS was a robust platform. The company combines data warehouses and data lakes into its lakehouse architecture that unifies data, analytics and AI in one cloud-driven service.
“The strategy of Databricks is to take five or six services in the industry, but do them as one platform that is integrated,” Ghodsi said. “Our power is doing those integrated together in a way in which it’s really easy and simple to use. In one word, the strategy of Databricks is unification.”
This platform-based unification approach has created a difference of opinion over the evolving role of independent software vendors or ISVs. In Furrier’s exclusive interview in advance of re:Invent with Adam Selipsky, the AWS chief executive insisted that enterprise customers building on top of the cloud provider, such as Goldman Sachs & Co. with its Financial Services Data Cloud, were ISVs.
“Goldman on top of AWS is an ISV,” Selipsky said. “It’s a SaaS offering. We’re the cloud.”
Yet Ghodsi sees a subtly different dynamic emerging when it comes to the role of ISVs. His company is playing host to ISVs as they leverage Databricks services on its own platform.
“We have actually a whole slew of ISVs on top of Databricks, that integrate with our platform,” Ghodsi said. “We host those ISVs that then work on top of the data that we have in the Databricks lakehouse.”
Facing multicloud reality
Since the founding of Databricks in 2013, the cloud landscape has shifted. Microsoft Azure, Google Cloud Platform and Alibaba have grown substantially since then.
“Things have changed over the years, AWS is not the only cloud anymore,” Ghodsi noted. “Azure is real, GCP is real, there is also Alibaba. Now, over 70% of our customers are on more than one cloud.”
This multicloud reality has led Databricks to adjust for changing customer desires. The Databricks service now supports AWS, Microsoft Azure and GCP in over 50 regions around the world. Databricks can use a single code base to bridge compute and storage across public clouds for data federation and disaster recovery.
“This data stuff and analytics is not a walk in the park, it’s hard,” Ghodsi said. “If you have to do it again on a second cloud with a different set of services and then again on a third cloud with a different of services, that’s very costly. Not only do customers want a simplified, unified thing, but they also want it to work across the clouds.”
Why hyperscalers must adapt
Will this dynamic lead the major cloud providers to build multicloud bridges as well? Ghodsi believes that the hyperscalers will have to make services multicloud compatible, a position further outlined in a whitepaper — “The Sky Above The Clouds” — that he and other researchers published earlier this year.
“The writing is on the wall, it’s super-obvious what’s going to happen next,” Ghodsi said. “Customers will say: ‘For any service I am using, it better work exactly the same on all the clouds.’ You are going to see the cloud vendors changing the existing services they have to make them work on the other clouds.”
Hyperscale adoption of a multicloud model will likely result in more focus on the infrastructure layer, according to Ghodsi. The competitive differentiator among major cloud providers will be in compute performance rather than in services.
This focus on the infrastructure layer is becoming more apparent with new initiatives in processor development, such as AWS Graviton, designed to deliver optimal price/performance for cloud workloads on Amazon EC2.
“Probably 70% of cloud providers’ revenue is in the lower infrastructure layers – compute, storage, networking,” Ghodsi said. “You can see lots of innovation happening there. With Graviton, we see huge price/performance benefits in those chips. We see Azure doing the same thing now, they are also building their own chips, and we know that Google builds specialized machine learning chips, Tensor Processing Units.”
Ghodsi believes this evolving scenario among major cloud providers will place Databricks in an advantageous position. The current state of the economy is leading enterprises to reduce expenses and Databricks’ message of cost-cutting simplification could continue to find a receptive audience. The company reached $800 million in annual recurring revenue last year, an 80% increase over 2020.
“The ecosystem right now is full of lots of different software and people are looking at how to consolidate, simplify and cut costs,” Ghodsi said. “They are adopting the lakehouse because they are thinking: ‘Instead of using five vendors or three vendors I can simplify it down to one with you and I can cut my cost.’ If you want something simply integrated that also works across the clouds, then I think there is a special place for Databricks.”
Here is John Furrier’s complete interview with Ali Ghodsi: