All right, great. Why don't we go ahead and get started? So we're just delighted to have Snowflake with us here at the Citizens JMP Technology Conference at the Ritz-Carlton Hotel in San Francisco. Thank you to Mike Scarpelli for coming and spending some time with us. How are you doing?
I'm good, other than that light is blinding.
I know it's blinding. I was.
Terrible.
I know. I know. I know. It's really.
If I'm not looking there, it's because I can't really see.
It is blinding.
Maybe this way.
Yeah, so much talk. We're just going to dive into it. All right, let's start at the top. How's business?
Good.
It is good, right? In fact, it was great, right? Was that overstating it? It seemed like Q4 was pretty great.
Q4 was good from a booking standpoint. A little bit lighter on the revenue than what internally I was forecasting. I think that was just because it seemed like there was more of a prolonged holiday impact. The last week or two of January is kind of where we were expected, but it's a little bit slower in the first part of January.
OK. Why was that?
You know it's the nature of a consumption business.
Yeah.
It really is.
This whole projecting these consumptions, it turns out they're really difficult to project.
It is definitely a lot more difficult to predict. You know when you're in a SaaS business, you know to the penny 97% of your revenue at the start of a quarter. That's not the case with us. It really is human interaction with our system. Yeah, about 70% of the work is scheduled work, but that human interaction has a big impact on consumption.
Ok. And so.
By the way, it was still good. It just was slightly less than what we would have expected internally. Bookings were phenomenal. It was a record bookings quarter. We signed some really big deals. A lot of people have asked, does that mean you've drained the pipeline of big? No, we have a lot of customers this quarter. I know three of my top 10 customers have to do something this quarter. But bookings doesn't mean revenue. Bookings is just showing you the customer commitment to Snowflake longer term. But revenue is where we expect it to be right now. I feel good about that. And we have a new CEO starting.
Oh, yeah. We're getting a lot of new products.
We have a lot of new products coming on board. I think the guidance is appropriate coming into a new year with all the change that we have. We'll see where we are at the end of the quarter.
Yeah. Yeah. Ok. So stocks off basically 20% from when you reported, primarily I think because of Frank leaving and the CEO change. When did you know that there was going to be a CEO change?
Well, you never know until he actually signs his employment agreement and Frank resigns, which happened on Tuesday. But you know I've been through a few CEO transitions. And Frank, all along when he joined the company, he said there is a time frame that I'm looking at here. It's not cast in stone, but we need to at the board, we need to really take succession serious because Frank believes that that is one of the primary roles of the board, is succession planning, hiring, firing the CEO. And we've been talking about it at the board for two years now, looking at both external and internal candidates. And when we acquired Neeva last May, I remember telling Frank, you've got to spend some time with Sridhar. And Frank quickly did. And we brought Sridhar onto our executive staff.
He quickly was running all of our AI and ML initiatives. The more time Frank and the board spent with him, that we realized that he could be a successor to Frank. Then talking about timing, the reality is good people don't sit around. They have other options. If Sridhar were not the CEO here, I would just tell you he'd be working at a very large company, maybe running all their AI initiatives.
Yeah. What's he like?
Sridhar is a driven individual. First of all, he's a super nice human being and humble. But don't mistake that with being soft. He is really driven. And he is very much about his big focus is we need to get product to market quicker. Why? Sales needs more product. And he's all about sales enablement, is making sure our sales team, our SEs, really are up to speed on our new products and know how to sell them. A lot of people mistake that he's a technologist and doesn't have operational chops. You know he ran a very, very, very large team at Google. They had like 12,000 people in it. And he scaled that business to $120 billion in revenue when he left. And he was very much involved in.
Say that part again. There were 12,000 employees.
$120 billion.
$120 billion in revenue. Google's what, roughly $300 billion, right?
The ad business. That's where they make all their money. Scaled that from $1.5 billion to $120 billion. So he's run big organizations before.
How many employees are there at Snowflake right now? Ballpark.
7,300, something like that.
Yeah. OK. How did your staying or leaving, committing for the three years fit into this conversation? Who discussed that with you? When did that happen?
I've had conversations with Sridhar and our board. Sridhar would like me to stay indefinitely. I'm like, Sridhar, we haven't even worked that closely together yet. I'll commit that I'll stay for three years at least. I'm also a big boy that if you think there's someone better, hire someone. I'm good to retire when you want, if you want me to retire sooner. I will tell you, I'm not going to go into another operational role, though.
You're not.
I'm too old.
What comes next?
You know I have enough to do on my own without managing all the people I manage.
Ok. So Sridhar's got his PhD from Brown, right? He spent an enormous amount of time at Google doing basically the most AI-intensive stuff.
His PhD is in databases.
Is it?
Yeah.
Did the board feel that the way the market is evolving, it's important to have someone with those kinds of technical chops at the helm?
We felt, with where the company at is and where it's going, you need to have someone with strong technical background to be able to run the company, to help be able to attract good people as well, too. He's very, very good at attracting the top technical talent. Without him, we wouldn't have got the five people out of Microsoft, the DeepSpeed team that he hired, to come and help with our AI initiatives.
When did that happen?
That happened five months ago, four or five months ago.
When he got them, and I think.
I think that was terrible. But I think last quarter, the last person just onboarded it.
Yeah. Who is this team out of Microsoft? What's their?
It's called their DeepSpeed team. Well, five of the core engineers there. They're very expensive people.
I shuddered to think. Ok. What does a sort of low-level, you just got your Ph.D. data scientist go for? And then what does an experienced DeepSpeed type one? What's the market for this kind of talent?
Some of these people you're talking when you're looking at the whole package coming together, it's $3 million, $4 million a year.
Yeah. Yeah.
A lot of that's equity, just so you know.
Yeah. And so Sridhar, he's really you said his big focus is we've got to get products to market faster so we can put them in the hands of salespeople so they can sell them. Demand's not really the problem. Yeah. So demand's not the problem. Were you guys not getting them to market fast enough before? And what were the roadblocks?
Listen, this is the year that we're delivering the most product.
Well, that's what I was thinking.
No, no, no.
OK.
You see, Unistore, we announced Unistore two years ago. We're just going to deliver that now. We need to shorten that time frame. Snowpark, there was a lot of stuff we've learned. Even though we went into GA, we've learned a lot of stuff with the migrations we were doing that we had to do more work on the product. So Cortex is one of the main things that Sridhar and his team really delivered. They delivered that into Private Preview in seven months from the start. It's going into Public Preview right now. And we'll be GA by Summit. That's a lot faster. That's what he's trying to do without sacrificing quality. Now, you can't do that with all things. Some things just take time. Like Unistore, there was a lot that we learned. That was a four-year process. But we now think we have that right.
Unistore in 2021 is four years. Cortex is basically 1 year.
Yeah. Not even a year.
Not even a year. So that's the order of magnitude we're talking about. What are some of the things you said we learned a lot on Unistore? We learned a lot on Snowpark. What are examples of some of the things you learned?
Well, I would say on Unistore, the performance that was required for certain things and having to take costs out because the costs were so high relative to to get to that performance, that we underestimated that. I would say one of the things that we Snowpark was really the product was missing certain capabilities. And I'm not an expert to go through those capabilities. That'd be a better thing to ask Christian or Sridhar. The other thing that we made a mistake in terms of the data scientist persona, they really want a notebook. We were leading towards partnering with people like Hex and others. And we really realized from customers, they want us to have a notebook. And they just want to have it to be able to use. They don't want to have to go procure it from someone else.
We have been working on a notebook for some time now that will be in Public preview very, very, very soon.
I can't remember who it was, but I was talking to one of your partners. He told me there's one of these notebook companies that works well within Snowflake that's gone from $3,000,000-$12,000,000 in like some incredibly short period of time.
Hex, probably.
Yeah. Yeah. But that's something you should own. That's pretty core.
Yeah. Yes. Data scientists want a notebook to play with.
Yeah. Ok. So how are you feeling about the business sort of new leadership, speed of getting products to market, losing Frank's fairly awesome set of go-to-market execution skills? And you balance it all out. How do you feel about it?
You know what I would say is everyone has this perception that Frank is driving all the sales and stuff.
It's true. We do have that perception. It's not the case.
So you know when Frank joined the company, there were maybe 1,000 people. And the whole go-to-market had to get it realigned. And he was instrumental in doing that. But we have a team of people that are intimately involved in the selling. And yeah, Frank talks to a lot of CEOs and stuff. But he's actually not selling. There are many, many, many people who are doing that. So I'm actually excited about Sridhar coming because he is so driven. He's actually in the office every day. He's interacting with people. And Frank wasn't in the office as much as what people think.
Which office are we talking about that Sridhar's in?
San Mateo office.
Right. Ok. Great. That's great. Where are you most of the time?
I split my time between the Dublin office, San Mateo. And the last two weeks, I was in our Bozeman office. We do our earnings and stuff there. And I was just working out of there.
Yeah. Where do you live, though?
It depends where I have to be.
OK. Well, all right. Let me ask differently. So over the next couple of years, where do you think you're going to be spending more of your time?
I'm definitely in California a lot if I'm not traveling. I'm heading out on Sunday for two or Saturday, two weeks for Asia in front of customers, doing a lot of stuff.
Yeah. So I was doing some interview. And it was the day that I was being interviewed about something else. But your stock was down the next day. And I got a question about it. And I said, you know, go back and look where ServiceNow's stock price was when Frank left. It was $80. And pull up where ServiceNow's stock price is today. And on the day of that interview, it was $780. Let me finish. So I think we're going to be good, right? Frank leaving is not a sign that what is a great company on a great trajectory suddenly is that trajectory is derailed. How good is my analogy?
They're very different businesses. But I think Frank would tell you he'd be upset if the stock traded up on his announcement of his departure. So you know I don't think anyone likes change. Investors really seem to like Frank. But just like me, we're just one person out of many on the team. It's a big organization. And I do think I personally think the selection of Sridhar as the CEO is the right choice for what the company really needs. And I think that you saw ServiceNow did really well when McDermott came on board. He's very much a sales-driven CEO. I think John Donahoe was a great. He's a really good human being. But he's not a technology guy. And Frank, when we started talking about succession, he was adamant it had to be someone that was deep in technical chops and with operational skills.
Sridhar, from everyone we saw, is the right person. We whiteboarded many of the top technology people in the world when we were kind of planning out who was the right person to go after.
If you just simplify it, right? So if you're coming from a position where you really own the cloud, the data cloud, and you want to move more into where Snowpark is, right? What's the fundamental sort of advantages that you have from having the data?
I would say Snowpark is fundamental to the cloud data platform that we want to be. I would say the biggest place we are moving is some of the things around generative AI that Sridhar is planning on doing. He envisions a world where a business analyst can just ask a question in simple English that can be translated into SQL code to get answers out of your data. That is what his vision is, using AI.
That's a good vision.
You know the other thing, too, what was really attractive about buying Neeva for us is customers don't understand all the data they have. Being able to find data, the enterprise search capabilities that Neeva built are core to what we are rolling out as well, too. That's another thing. Not just within your own data. We've heard this from customers for years. We have over, don't quote me on this, but I think we have over 1,800 different data set listings in our Marketplace. Well, customers don't know what is there. But also, through the use of AI, what data sets may be relevant to you, Mr. Customer, based upon what we're seeing in your data. That's what he's working on doing. That, I think, is going to be really meaningful. It's not going to happen overnight.
But that's what the team is working on.
Right. Make sure I understood it right. So number one was, instead of this idea of creating a place where all data science can come and do all their stuff, the vision is, no, just make it so that the business analyst can just talk to their data and get an answer that you would otherwise need a data scientist to get.
Correct.
Right. OK. Check. And then the second piece was around Neeva's, it was supposed to be basically a replacement for Google, right?
It was originally going after consumer search. Then they pivoted and said, you know, we're going to go after enterprise search.
OK. I didn't realize you were saying that.
And then taking that enterprise search capabilities into Snowflake for our customers, both within their own internal data but then also looking at the marketplace but being able to understand what data could be relevant to you based upon your types of data that you have.
How are you supposed to figure that out now, today?
Right now, it's the human that has to figure that out. That's the issue.
There's 1,800 data sets.
Yeah. So it's pretty overwhelming.
All right. Great. So the other question I've got—oh, you know what? Maybe I'm going to stop. Does anyone in our audience have questions for Mike?
I like your questions better. They're easier.
No. Probably. Yeah. The hard ones always come from the audience.
Silence.
Well, no. You know, usually, honestly, I tell my associates to count to 10. Then they have to ask a question. But I think, is it possible none of them are in here? Oh, there we go.
Out of some of your newer products that aren't including guidance, your Snowpark is 3% of product revenue. What other products that aren't included do you think have the highest potential to outperform Container Services? Is it something else?
For the current year or longer term?
Current year and longer term.
Current year, Cortex could be meaningful because that will drive a lot of compute and GPU compute if customers are bringing their large language model of choice into Snowflake to fine-tune that data. Longer term, Snowpark Container Services really enables native app developers on Snowflake to bring in their applications into Snowflake and monetize them through the marketplace. And that will have a very large impact. Unistore has the potential, not necessarily this year. We'll start to see consumption ramp up. But longer term, Unistore will be big. We are starting to see Dynamic Tables ramp up. That's not in GA yet. But just based upon the Public preview that we've seen, there's an uptake on that we're starting to see right now. Streamlit is another one, too, that is in GA this year. And we're starting to see more and more customers use that.
But definitely, Cortex would be, I think, the most meaningful driver this year of revenue.
When you did the guidance for this year, what assumptions did you make that could turn out to be conservative that maybe gave a little more headroom for your new CEO?
Assuming similar consumption patterns that we saw in 2024. I would say the low was probably Q2 of 2024, not factoring anything associated with the new products we have coming on. If our assumptions on the product headwinds associated with the timing of new performance improvements and migrations of hardware platforms where there's improvements, that gets delayed, that could cause conservatism.
All right. Number one was consumption was brutal last year. You're assuming that for this year, right?
Well, not the whole year. But we're still not assuming any type of rebound or anything this year, similar consumption to last year.
Yeah. And then nothing on the new products. And that includes Cortex, which you just told us could be meaningful this year.
Yes.
And then what was the third one?
The third one. Now I'm going blank on what I said the third one was.
I didn't get it down.
Well, you were too slow.
Oh, no.
Now I forget. I'm getting old. See, I'm getting old.
All right. Well, Mike, thanks so much for coming. We really appreciate having you here.
No problem. Thank you.
Yeah. It's good to see you.