The fastest revenue engine in SaaS history: $5.4B run rate in 10 Years - Ron Gabrisko [Databricks]

0:00Ron Gabrisko:It's all about trust and adding value, right? I'm there to teach customers how to get value out of our product. What you want to do is tie pricing to your most basic unit of value. And for us, that was consumption. Don't give your day to us. Don't give it to anyone because that's going to be, like you said, your goal. That's your secret sauce. I mean, that's what I'm here to help build. Today on Billions, I'm sitting down with Ron Gabrisko, Chief Revenue Officer at Databricks. In 2016, Ron joined the startup with less than$1 million in annual recurring revenue, a product billowed by engineers who had never heard from a sales rep before.

0:3910 years later, Databricks runs at$5.4 billion revenue run rate, growing 65 % year-on-year at a$134 billion valuation. No CRO in SaaS history has built a revenue engine this fast from this early at a starting point. Ron, thanks a lot for being here.

0:59Ron Gabrisko:Yeah, thanks for having me. Great to be here, Guillaume. Yeah, I'm super excited to get into, you know, like the dirty details of what it takes to grow a company to that stage. So when you joined, you know, it was below one million in ARR. How was it? And what is the first thing that you decided to do? Yeah, I mean, it's funny. you know uh i was looking at kind of later stage companies at the time and i met ben horowitz from a16z and he's like hey i have this little startup called databricks it's way earlier stage they invented spark it's like the greatest piece of software on the planet and gave it away for free and i want to i want someone to help him build a business from it so uh he's like it has more upside than any company in my whole portfolio.

1:48Ron Gabrisko:So, uh, you know, I, I jumped in the boat, we have seven PhD founders and, uh, yeah, it was, it was, uh, basically, you know, how do you create a product? How do you create the monetization? How do you create the initial pitch deck? I mean, uh, we had to build everything from scratch. So, uh, it was pretty fun. It was pretty exciting. So, so lesson one is, uh, when Ben says something, you listen and you, you go Yeah, exactly. He's done pretty well, you know. Indeed. Yeah, exactly. So it was good advice for sure. And how did it look like at first, like when you joined the company and how exactly do you structure the sales motion and the whole like revenue team from the start?

2:35Ron Gabrisko:Yeah. I mean, when I first joined, like, I mean, the company was mostly engineers. So I think we had probably 40 or 50 engineers, a couple salespeople. But the initial thesis was like, go to all the Silicon Valley startups, because they're going to be the earliest adopters of new technology. And we were actually, the company was actually trying to do what they called zero touch. So it was like product led growth. But, you know, the product was pretty complex. and so like you know if you have a product that's like intuitive it's easy to try to do product-led growth in early days it wasn't that intuitive right it was fairly complex so you know step one was like let's learn you know how everybody's using spark uh how you know uh open sources is being used what people will pay for so you know i hired the initial sales team just to go out and literally talk to all the open source users and say, how are you using it?

3:35Ron Gabrisko:What would you pay for? What are your challenges? And that's how we built our initial pricing models and initial pitch deck and all that kind of stuff. So not rocket science, but trying to find product market fit for sure. And I'm quite curious because you mentioned like, okay, you have your sales team and you reach out to developers who are using an open source project. But for me, there is something to unwrap here because how do you reach out to developers and get them to answer? And especially like talk to sales rep. So, yeah. So what were like the method you used like to get them to reply?

4:13Ron Gabrisko:I mean, listen, like I think Databricks, and I've talked about this a little bit, but, you know, we're very technical sales team, right? Like I think, you know, kind of old school, just relationship sales, you know, in this world, in a very technical world with technical products, it doesn't work. Right. So you have to have pretty technical salespeople. And the biggest thing we're doing is trying to offer to help. Right. If you have a challenge, you know, we'll bring a technical resource or we'll try to help with it ourselves. And so that's the biggest thing I think when you sell to engineers, of course, you know, they don't love salespeople.

4:49Ron Gabrisko:They don't necessarily trust salespeople. And, uh, you know, for me, I've done sales for 25 plus years. It's all about trust and adding value, right? I'm there to teach customers how to get value out of our product. And, uh, and that's the biggest thing, you know, it's, uh, how do you break that myth of, you know, trying to sell used cars or something like that, right? It's, uh, it's a very professional discipline, uh, in sales. And so, you know, we took the approach, let's hire really technical sales folks to be able to add value to these engineers. So, and in your like outreach method, like, do you see something that helps you build trust a little bit faster?

5:30Is it like using people's jargon? Is it like, I don't know, being super specific about the problem that you can help them solve? Like, what are the things you've tried that allowed you to get like more replies and more meetings?

5:43Ron Gabrisko:Yeah. I mean, originally we had, you know, community edition, which is kind of like our free edition, right. And people would sign up for that. We mostly would just reach out and say, Hey, just, you know, saw you signed up. Like, you know, is there anything we can help with or any technical challenges? And so if they ran into, Hey, can you help me configure this or set up this? Or, you know, I have a memory issue with this. I mean, you just add a little bit of value. you you gain trust you know what i mean versus like hey can i come pitch you on why you should buy databricks and so very different approach is you know you got to give to get back um and so that's what we started with just you know very small like reach out to try to help folks and uh you know that worked really really well for us so and to build up on what you're saying i think like what you did very differently it's like in 2017 i think like obviously every sas company in the world were selling per seat and selling seats you built basically a consumption model instead can you walk us through like how exactly did you build like that's like a consumption model and how exactly did you choose where to price, what to keep free?

7:00Ron Gabrisko:Yeah, no, that's a great question. Because I think one of the early, I mean, Databricks is an innovation machine, you know, in terms of product, but we innovated quite a bit on GTM too, right? Like I think we're the first pioneer to build a cloud service around open source. And that became like how you monetize open source. And so we kind of created that playbook for the market. I think, you know, a lot of startups and early stage companies, what you want to do is tie pricing to your most basic unit of value. And for us, that was consumption. You know, seats has a limitation, especially, you know, there's so many people in an organization.

7:43Ron Gabrisko:but you know tying ourselves to consumption when data is growing queries are growing you know ai is going and agents are going to do even more queries like like that was kind of a genius move that we did as a team and the market especially around cloud technologies was kind of going that way and so that's where we started you know our first kind of proprietary product was our data science notebooks. So we kind of created this entire category of data science. So, you know, we would enter in through kind of the machine learning and data science teams that were building these models to do predictions.

8:24Ron Gabrisko:And, you know, so we grew a lot from kind of bottoms up with that specific product. And so we would call it open source core, right, where we continue to put more and more in open source, but maybe keep some of the proprietary features so that we can continue to build a business and then we keep putting those in open source and so you know people have you know uh listen the great part about open source is you're not locked in you can move away if you need to and so we kind of embrace that community and that interest but also built a business around it with other things we thought had value so at the same time you know like uh what's interesting in what you're saying you know about open source is the fact that as you said like people can move if they want to move but at the same time like if you look at your retention number i don't have like the exact one but i'm assuming it's like a north of 130 40 something like this yeah we're north of that okay so that's like super impressive which means that revenue keeps expanding your turn is extremely low so how how do you think the the two are are connected or maybe more specifically, like how do you make sure that you give enough value to people so they actually like stay longer?

9:43Yeah.

9:44Ron Gabrisko:I mean, part of it's just, you know, customer obsession, making sure customers are successful and they love the company. And, um, you know, I think, uh, the entire company, including engineering and sales are trying to make sure that, you know, they love Databricks, but the bigger part is we didn't just do one open source project. There's a lot of open source companies, they build one product and then they stop. And it's hard to be sticky because then they can probably use that elsewhere. What we did was we built Spark and then we built Delta and then we built MLflow. And then we combined Delta with Iceberg when we bought Tabular, which was also open source.

10:27Ron Gabrisko:We bought Mosaic AI, which was open source. Like we continue to do these open source projects. And so now it's a fully integrated data and AI platform that's all open source. Right. And so I think that's the key to retention is you can't just have one trick. You got to, you know, keep layering on these new innovation engines. And that's what we've done really well at Databricks. Right. Is, you know, created this innovation factory where we're constantly, you know, reinventing ourselves and releasing new technology. So, and as a CRO, like how much involved are you in the decisions of launching like a new open source project or even acquiring an open source project?

11:14Ron Gabrisko:I mean, a lot of that's driven by Ali and the founders, right? Like they have specific cultural things they're looking for in acquisitions as well as technology fit. And, you know, I would say our seven founders are visionaries for this space. I mean, they teach at Berkeley and MIT and Stanford and most of the, you know, other innovations come out of those schools and they've probably taught most of those students. But, you know, so they have their own vision on where they want to go with the data and AI platform. My role in that, I mean, I do get involved in the acquisitions to understand the sales teams, customers, you know, even try to sell the, you know, the companies on Y Databricks versus going alone.

12:00Ron Gabrisko:But they're really making the decisions on the technology and the acquisitions. Um, the part GTM solves is like, you know, recently we released Lake Base and Genie, uh, and, uh, Lake Watch. And so how do I get, you know, all the enablement done? How do I make sure we know how those products add value, how they differentiate with other products in the market? those are massive innovations for us in the market and so you know we want to try to grab market share and grab customers and get customers using those as fast as possible and so a lot of the new product and introduction materials and enablement and all those things we go pretty fast on that stuff so and what what would you say like from your side like what is the focus on new businesses versus existing customers and expansion with other products?

12:57Like how do you split kind of the effort?

12:59Ron Gabrisko:Yeah, I mean, I spend probably, you know, 30 % plus on new business acquisition, meaning in terms of resources, people, because I know if we want to continue, you know, we've publicly, you know, 65 % plus year over year growth, you got to have new customers because those new customers turn into million, 10 million,$100 million customers. You can't just milk your existing base. I think that's another, you know, mistake that some more mature companies make. I mean, obviously our existing customers, because our net retention's, you know, crazy high, like we're getting lots of revenue and in a lot of our new products, we're selling into those existing customers, but you got to do both.

13:40Ron Gabrisko:So, you know, for example, like Genie, like unlocks such amazing use cases, allows you to talk to your data using AI, does calculations, predictions, it'll do graphs, you know, so CEOs and CFOs are now using Databricks to, you know, be able to get all those answers that they need for, you know, doing predictions and running their business. And so each one of the new products we release, we're telling that story inside that customer, but we're also using those to land new customers. So, you know, you definitely have to do both. So. Okay. Interesting. And when it comes to, uh, you, you were, you were mentioning the fact that you're growing like 65 % year on year or like North of that, um, every year it gets harder to reach like, uh, that, uh, that revenue, uh, milestone when it's the beginning of the year or like before the beginning of the year, like when you're setting up what's going to happen for the next year, how exactly can you walk us through maybe like, how do you define where I'm going to spend money, how I'm going to do it, who I'm going to hire, like what's your reasoning about all of it and how do you plan for it?

14:53Ron Gabrisko:I mean, the good part is like, we're, you know, a data and AI company. We do data science on every single customer. So I can, you know, I can kind of predict plus or minus a certain percentage, how fast I think each customer is going to grow. And so I know just based on our current products, like how much each customer is going to grow just organically through more data, more queries, all that kind of stuff. then we'll layer on okay these are the new products we're going to release okay let's try to predict okay how much you know you know should we be able to sell 100 million plus a lake watch or you know what are the goals for each one of those other products um and then based on that you know we have a model of like okay how many new customers do we want to add we know the productivity of the new the hunter reps versus the core reps like how many core reps do we need and obviously we're putting some assumptions in there in productivity improvements.

15:51Ron Gabrisko:And so, you know, at this point it's, it's a science, you know, we're doing, you know, like I said, you know, 5.4 billion plus publicly is the number, but you know, it's kind of a science how we do that planning. And again, there's a little bit of predictions on what are the new products going to drive that year. But once we get kind of a model for that, we'll know how much that's going to go the next year in each customer. So it's pretty cool. And yeah, it's definitely like really cool. But I'm just wondering, do you do like outbound on brand new customers that you're going to try to like basically move from a competitor or else to your platform?

16:33Because I'm assuming like this predictive model works well for existing paying customers because you can tie obviously all the data to the usage and you can look at how fast the usage is growing. And then eventually you can, I guess, draw a line if I simplify it to make the prediction. But do you also think of, okay, in terms of new customers, how many new customers do I need? What's kind of my target addressable market? How am I going to split it? Like, how do you think about all that?

17:03Ron Gabrisko:Yeah, exactly. So we'll be like, you know, hey, we've segmented the market. we know 7 ,000 companies in the world make up about 90 % of the TAM. And we've done that using data sets, you know, that we bought, you know, looking at cloud spend, at different product spend, things of that nature, right? And so we're focused on how do we get all of those companies as customers? So assuming we have, you know, a certain percentage of those, we want to go get the rest. We know each Hunter AE can land, you know, this many new customers, you know, say like, you know, depending on the segment, you know, if it's like a strategic or named segment, they might be able to land one or two in a year.

17:48Ron Gabrisko:If it's like a, you know, kind of startup or D &B, they might be able to land 10 or 20 in a year. And so we're measuring those. Now, the actual dollar impact of those inside that year is not huge because they're just starting to consume. but we also know by segment right we know for like hey a retail customer of this size is going to grow approximately this much the first year this much the second year so again we're analyzing we know the lifetime value of each one of those new customers based on the segment and so we can kind of use i mean that's what i'm saying like i'm investing ahead of the curve on new business acquisition we're investing you know approximately 30 percent to make sure we get all those new customers because we know lifetime value of those is in the tens of millions of dollars for us so wow yeah so you can spend you can spend a bit of dollars for acquisition exactly exactly and and how exactly is uh the the like the the revenue team how is it split exactly between uh Do you have like marketing, go-to-market sales?

18:59Like how exactly is the whole structured?

19:03Ron Gabrisko:Yeah, we have sales. We have field engineering, which is pre-sales. We have marketing. We have SDRs. We have sales ops. Those are kind of the key components, I would say, in partnerships. We also have vertical specialists and product specialists. And so as you get larger, I mean, we're probably, you know, 5 ,000 plus in the field. You know, you need specialists, right, for certain products and certain verticals. So we're verticalized. You know, it's awesome because data and AI like literally, you know, will transform these companies. and we can tell a story on how you can use Databricks to, you know, increase revenue, reduce costs, transform these companies, right?

19:52Ron Gabrisko:And so, you know, for example, like we work in, you know, pharma and in the genomes, like now they're doing new R &D for drugs virtually. They're, you know, doing data and data and AI and data science on different compounds, matching it against medical records and, uh, you know, Regeneron found a new liver cancer cure, uh, that would have taken them, you know, years and years. It accelerated R and D by 10 X. Right. And so we can tell that story and you want to be able to tell that story for every pharma company. And if you're in retail, you want to be able to tell how they're managing, you know, increasing revenue for, you know, how they maximize revenue by where they put products on the shelf, what promotions they're running.

20:41Ron Gabrisko:Um, you know, I mean, every industry has different use cases, right? So we've mapped those use cases to know, Hey, this is how this vertical uses data and AI to make the biggest impact on their company. And so we want to go sell those outcomes to the CEO, to the CIO, uh, to the CDO of those companies on how we can help accelerate those. And so we're organized vertically. I mean, obviously we have North America teams, AMIA teams, APJ teams, but it's all organized vertically. So we can tell that outcome-based story for each vertical. And when you say you're organized vertically, does that mean also that, for example, in the marketing team, you're going to have like specialists on, let's say, this pharma vertical, or do they work across like several industries, but the verticality is more about the product specialist and the pre-sales or sales.

21:38Ron Gabrisko:No, we have both. So like for retail, we'll have a GM and a tech GM that will understand the entire market, what are all the, call it different use cases that say a large retailer would want to use data and AI. And we can even calculate what we think for each specific company, what it's going to be worth to them to implement data and AI. Um, and then on the same, in the same way, we'll have like a genie or lake based specialist that can go really deep on how does genie help with their specific datasets? You know, can we configure it to make sure that, you know, our biggest retail customers can take advantage of it.

22:20Ron Gabrisko:And so we kind of do both, but the vertical specialists sit inside of sales. Uh, you know, I compensate them basically based on the success of that particular vertical. And so we've aligned everyone on how do you make sure the company continues to grow super, super fast. Very interesting. And I think recently you've done a partnership with Anthropic to make sure that all the Databricks users were able to leverage it directly. like how exactly do you position yourself when it comes to this kind of partnership because Anthropic is obviously right now like they have surpassed you know like open AI in terms of revenue like are you whenever you do a partnership do you only work with one specific company or do you also consider others like how exactly do you do you create this kind of uh yeah good good question.

23:20Ron Gabrisko:So, I mean, we partner with Anthropic, but we also partner with OpenAI and Gemini and GCP. And so like a lot of our customers want kind of model optionality or coding optionality, depending on, you know, what they're using. It's similar to how we partner with the clouds, right? So we're a first party on Azure, which is unique. But, you know, we obviously run on AWS, Azure, and GCP. They might have their own native technologies too, but a lot of customers, you know, one, they think Databricks is the best data and AI platform and similar on AI. Like, you know, again, I think, you know, the value of AI is how do you apply it to your proprietary data to make decisions and recommendations and predictions for your company.

24:11Ron Gabrisko:And, you know, Databricks is the data platform in most of those companies. And so for a lot of enterprises to get value out of AI, they need to connect some kind of AI system to Databricks. And so that's why we partner with all of them. And yeah, I mean, they're growing super, super fast, obviously, but it's exciting times, for sure. And do you feel like the AI wave we're seeing right now, do you think that it's helping accelerate Databricks business also? Yeah, for sure. I mean, we've already, I think publicly said, we've done 1.4 billion just in AI already, right? So that's probably one of the fastest growing parts of our business.

25:01Ron Gabrisko:And I think it's just the beginning, right? Like companies are just starting to adopt this technology. It's absolutely game-changing. So we're excited about it. I mean, Agent Bricks and Genie and a lot of the products that we offer leverage AI. Again, we're focused on enterprise AI, not necessarily consumer-based AI, but how does that AI with your data, how does that help you make decisions and predictions and really at a CEO, CIO level? That's the whole focus of Genie. So, and can you walk us through maybe like a few use cases that you found like a great and you were like, ah, okay, like this, this is next level of what we're going to be able to do.

25:46Ron Gabrisko:Yeah, it's so, it's so cool. And so, and we build out these like custom demos and then proof of concepts for our customers as well. But like, you know, looking at like a big retailer, you know, they were looking at, okay, how do they, you know, try to maximize revenue from their, you know, short stock, right? So if you're a grocer and you have certain things that are going to expire, what kind of promotions? And literally, you can see which stores in the network have potential short-life products. And then there's agents that will go look at that data in that store and recommend what promotions they should run for that particular store based on the demographics and all the buying.

26:35Ron Gabrisko:I mean, that's crazy, right? Like you're basically, it's a, you know, a retail knock, right? Network operating center, which is pretty cool. I mean, I gave some of the pharma examples, you know, in banking, they're doing all kinds of fraud detections. so obviously you know when you see uh you know uh uh an email on your phone or a text on your phone that hey this might potentially be fraudulent that's you know usually databricks in the background running machine learning models to determine this is way outside your your normal purchasing uh patterns but again like if you think about it like a ceo if they're like hey what's happening with my business?

27:21Ron Gabrisko:Uh, you know, again, these retailer, what's happening with, with my business. Why is, you know, the Southeast slowing down on sales? Like normally they'd go get an analyst. The analyst would run a bunch of, you know, reports, send them back a report. They would look at it and say, okay, that makes sense. Like change this, add this. And they go back and they come back and like the data now is stale. So update the data. Now they literally can just start typing English and say, Hey, what's going on in the Southeast? You know, it might be like, Hey, these are the beverages aren't selling as well. You should try doing, you know, different shelf placement.

27:59Ron Gabrisko:I saw that increase in the Northwest region, right? Like they're literally just interacting with the data and asking questions. It's, it's absolutely game changing for CEOs, CFOs, CIOs. So it's, uh, it's incredible. So it's exciting, uh, exciting times ahead. and for sure um i'm just wondering like um do you sometimes have like uh because you're focusing on enterprise and very large enterprise deal so the data for some companies is equivalent to gold i would say and they don't want to share it with uh with everyone around them so is you know like do you sometimes struggle to close deals because of how the data is handled and this kind of thing because we've seen it a lot some companies typically don't want to use like Anthropic don't want to use like OpenAI etc because they don't want to share kind of like their data with some as some like AI companies so how exactly do you navigate this from your perspective and the enterprise clients you work with?

29:11Ron Gabrisko:Yeah, I mean, we never, you never give your data to us. Like that's the whole, you know, great part about Databricks is you stored in open storage, you own your data. Like there's a lot of companies out there that are like, you know, you give your data to them and then, you know, you have to make sure it's safe, all that kind of stuff. You never give your data to Databricks. It stays in open storage, in open data formats, Delta or Iceberg. Again, that's the key, right? And, you know, that provides an even playing field too for the entire industry. Like let the best engines and the best products win that work on top of that data.