00:00I do want to talk a little bit about kind of what you're seeing with regards to the demand
00:04for the consumption model. The Databricks is sort of, you know, I don't know if you made it famous,
00:10but certainly that you've managed to capitalize on that. And yeah, well, okay, I mean, you can
00:14take credit for it if you want. But I am curious as to just how that's holding up and whether
00:18you're seeing any shifts in consumer behavior. First off, thanks for having me today. And look,
00:25I think things have been going well on the consumption model, candidly. You know, we
00:29recently raised around, and one of the things we were noting about was that we've seen consumption
00:34accelerate across the board, right? I think we said something like 6.9 billion in ARR growing 80%,
00:41you know, year over year. And I think one of the big drivers is because we've seen it across all
00:46regions and all product lines growing. The way I like to think about it is people were using data
00:51and AI beforehand, but now effectively, everybody's kind of like doubled their headcount without knowing
00:56it, right? Because you have agents who are basically generating more data, driving more
01:00consumption, going after more use cases. And behind all of those, you need basically a database
01:06to power those. People want to talk to their data. And now you kind of referred to it, alluded to
01:11it
01:11earlier. Things like how do you control costs in that world and have governance are really important
01:17as well. Well, I am curious about kind of some of the chatter. And this is mostly anecdotal at this
01:21point, as I'm sure you know, this idea of people kind of going from the token maxing where, you know,
01:26every sort of corporate head was telling all their employees, you know, use this stuff as much as
01:30possible to now a little bit more of a value oriented approach to how they're using these large language
01:36models. How are you seeing that shift if that is indeed happening? And does that have an impact
01:42on your potential revenue generation? I think we're absolutely seeing that shift. Actually,
01:49you know, one of the big drivers is that we have a capability called Unity AI Gateway exactly for
01:55this. Because you saw people go from token maxing to what they're now referring to as value maxing,
02:01right? Which is what, you know, what is the value I'm going to get out of it? How do I
02:05make sure that
02:05there is ROI? And we're already seeing in our customer base, you know, 78% of them are using two
02:11models
02:11or more. And it's an and story, right? I think that the American frontier model providers are building
02:18incredible, basically, models. And there's some capabilities that they are really, really good
02:23fits for. You absolutely need those. At the same time, just to be candid, everybody does not need
02:29the best, best model in the world to generate an email draft, right? And so for that, they're going
02:34to want to leverage open source models that are, you know, far lower cost. And so they want to balance
02:40the right model for the right use case at any given time. And we've actually seen that be an
02:44accelerant because for us, we're big believers in kind of marrying both of those worlds. And the
02:49fact that we allow customers to choose and build their applications, leveraging the right model at
02:53the right time has been an accelerant for us. Do you think we're nearing the point, there's just been
02:58so much spending on AI infrastructure, these AI budgets at enterprises across the globe have, you
03:05know, increased their CapEx spending forecasts. Are we almost at the point now where CFOs want to see
03:13all of that spending actually turn into a real profit?
03:17I think we're absolutely at that point. But I think, obviously, demand is still very high. But
03:22in many of the conversations that we have with customers, it's why I mentioned, you know, the
03:27governance piece is so critical. CFOs are now asking, what is the ROI that I'm getting, right? And so,
03:34you know, capabilities that they've been asking for is, how do I give budgets to folks? How do I have
03:41guardrails on what they can do? And how do I intelligently route things to basically the right
03:46model that makes the most sense? I mean, that's why we've seen adoption of that unity AI gateway go
03:50so high. And so CFOs are then asking, how do I make sure what is that ROI? But that requires
03:56budgets, understanding of costs, and knowing that users are being routed to the right capability at the
04:01right time. So that is definitely a conversation I'm seeing across kind of all of the, you know,
04:05executive conversations with customers today. Hey, Arsalan, I got to ask you, obviously,
04:10the question everybody wants to know, are we going to see this Databricks public sometime soon?
04:16Look, I always give the same answer. It's a when, not an if. But to be honest, right now,
04:22our focus is, as I mentioned, you know, how do we basically get databases for these applications?
04:27How do we help everybody, you know, the Pradas, the 7-Elevens, the Pepsis of the world? How do we
04:32help
04:32them talk to their, you know, data? And how do we help them govern this AI world? That's our focus.
04:37And when the time is right, we'll be public.
Comments