Trends

Snowflake Inc. (SNOW) The JMP Securities Technology Conference 2023 Transcript

Trending 1 year ago
beritaja.com

Snowflake Inc. (NYSE:SNOW) The JMP Securities Technology Conference 2023 March 6, 2023 5:00 PM ET

Company Participants

Michael Scarpelli - Chief Financial Officer

Conference Call Participants

Patrick Walravens - JMP Securities

Patrick Walravens

...finance and to my right, of course, is Mr. Scarpelli, nan CFO. And we'll commencement astatine nan top, which is -- well, first of all, really are you?

Michael Scarpelli

I'm good. Can I opportunity thing earlier you do this? I person to show you.

Patrick Walravens

Yes. I can't wait.

Michael Scarpelli

I've been dealing pinch -- this is really nan 5th nationalist institution I've been dealing with, analysts pinch callbacks for many, galore years. And I person to show you, Pat did thing nan different time that was nan first...

Patrick Walravens

I can't wait, I can't wait. What was it?

Michael Scarpelli

It was unbelievable. He really had his son, who is simply a information intelligence who useful astatine really a customer of ours, travel connected nan telephone and asked each nan questions of Christian astatine a method level, which was I've ne'er ever seen an expert person their boy do that. That tells maine 2 things: A, you're old.

Patrick Walravens

Really old.

Michael Scarpelli

I've known Pat for galore years; and B, nan worth you get retired of really, if you're a information intelligence and you admit that, and hopefully, you sewage thing retired of that, that your boy study and you could admit nan quality betwixt Snowflake and what others are doing retired there. So I really did admit you doing that. I wasn't expecting that.

Patrick Walravens

What he...

Michael Scarpelli

And I besides recognize Pat's inexpensive because his location was truthful acold that his boy has a overgarment connected and he has a overgarment on, too. So…

Patrick Walravens

I wrote it -- I wrote it down. I'm not going to find it successful time. I wrote down Zachary's reaction. I put it arsenic a affirmative information point. So yes, truthful I person a -- we person 4 children, but 3 of them are -- hopefully, nary 1 is in. They're astir present somewhere. But 3 of them are normal, but 1 of them is just really, really disconnected nan charts, right? And he is simply a information intelligence and he talks slow. Did you announcement that?

Michael Scarpelli

Yes, he's very methodical.

Patrick Walravens

Very slow, very slow. And truthful he would beryllium pausing earlier asking nan adjacent question. I was like, "Oh, my God. Is location different mobility coming?" And past it was conscionable for illustration great.

Michael Scarpelli

He asked bully questions.

Question-and-Answer Session

Q - Patrick Walravens

It was great. It was great. And he -- truthful erstwhile I e-mailed you guys, I said you get my information subject master on, I didn't stock who it was. Yes. So Snowpark is simply a large deal. That was what I took distant from nan speech pinch Zach, and we'll get to that. So let's commencement pinch wellness business and past possibly we'll spell consecutive to Snowpark aft that.

Michael Scarpelli

Well, we conscionable gave an update connected nan business past week, and nothing's changed since then.

Patrick Walravens

Yes. Give maine your characterization.

Michael Scarpelli

Q4, from a gross standpoint, we beautiful overmuch deed our soul guidance from nan beginning. And erstwhile I opportunity internal, what we coming to our Board. And truthful it's wherever we expected it to be. What we did announcement was different was nan -- our older customers look to person contributed to much of nan maturation and nan newer cohort are increasing overmuch slower. And we deliberation that is simply a usability of newer customers coming is ramping connected Snowflake.

There's a batch much knowledge retired location astir group who person been moving connected Snowflake for a play of time, person learned nan pitfalls of not deploying Snowflake properly, nan pitfalls of not putting successful spot nan due governance. We person a batch much partners who are trained connected really to instrumentality Snowflake properly. We are doing a batch much training pinch customers connected really to usage Snowflake retired of nan gate.

We person built a batch of capabilities wrong Snowflake to guarantee that group are utilizing it decently for illustration car suspend. It was ever easy for group to you rotation up a warehouse. There's 2 things you want to do. You want to beryllium capable to prime nan correct size warehouse. And it was easy for group to prime a really ample warehouse. It was easy for group to disable nan car suspend function. So a, you could beryllium moving a storage bigger than what you want.

Now it's overmuch harder to do that, and we do that for you. We person each kinds of alerting erstwhile personification disables an car suspend function. So that group cognize that -- and there's a question, do you really want to do that? And b, group are conscionable utilizing Snowflake much efficiently.

But I think, too, is nan earlier customers, were those digitally autochthonal customers that deliberation of nan Instacarts and others of nan world that were conscionable calved successful nan unreality that really it was maturation astatine each costs. The cohort of customers we person now are nan much established, mature companies. I don't want to opportunity they're nan laggards aliases nan precocious adopters, but they're not those -- needfully nan fastest-moving companies. They've ever had costs controls successful place. And truthful we're seeing nan newer cohorts conscionable ramping much slowly. They still each person nan aforesaid extremity authorities they want to get to. They're conscionable going to spell astatine their gait and successful a very controlled manner connected controlling costs.

Patrick Walravens

Okay. So you foretell nan future. You were predicting nan early based connected cohorts. And nan caller -- nan cohort -- nan caller customers are really not matching nan aforesaid behaviour arsenic nan older cohorts. Is that a adjacent assessment?

Michael Scarpelli

Correct, correct. And erstwhile we were going done and spending a batch of clip successful nan 2nd half of January and February arsenic we're rolling quota -- retrieve astatine nan opening of nan twelvemonth is erstwhile we person to rotation quotas retired by rep for nan afloat year, and we rotation retired depletion quota. There's a batch of chat that happens betwixt my finance squad and nan salespeople connected accounts, and that's wherever it became evident a batch of those newer cohort of customers are conscionable not increasing astatine nan aforesaid pace.

And I don't mean customers we landed 3 quarters ago. I'm talking to customers we've landed complete nan past 1 to 3 years are conscionable increasing slower.

Patrick Walravens

Over nan past 1 to 3 years. How overmuch slower are we talking about?

Michael Scarpelli

Well, capable that I lowered nan guidance to 40% maturation adjacent year, which, by nan way, is still bully maturation astatine nan standard we're at. I'm not going to apologize for that type of growth.

Patrick Walravens

What was it before?

Michael Scarpelli

In nan extremity of November, I had -- erstwhile we were doing our preliminary planning, we were reasoning 47% growth.

Patrick Walravens

Yes. Okay. So arsenic an bid of magnitude of what nan cohorts are seeing, is that a bully indicator? Or is location thing other successful there, a furniture of other level of conservativeness aliases thing connected apical of it?

Michael Scarpelli

That's nan biggest point there.

Patrick Walravens

Okay.

Michael Scarpelli

And by nan way, we're mostly beautiful bully astatine forecasting depletion connected an yearly basis. And I'll show you, we had group our 2022 plan. In February of 2022, we went to nan Board. We did miss that scheme by 2%, meaning afloat transparency, astir $42 cardinal aliases immoderate it was. And that was earlier you had Ukraine, that was earlier you had liking rates, that was earlier you had nan crypto implosion, that was before...

Patrick Walravens

You're talking astir 4%?

Michael Scarpelli

2%.

Patrick Walravens

2%, yes.

Michael Scarpelli

So I deliberation we do beautiful -- my constituent is I deliberation we do a beautiful bully occupation astatine forecasting.

Patrick Walravens

Yes. Okay. So you piqued my liking erstwhile you started talking astir nan quotas. So yes, really do you -- really do nan quotas activity for your salespeople?

Michael Scarpelli

Salespeople get 2 quotas. They get a maturation quota, and they get a depletion quota. And it could beryllium anyplace from 50-50 to 50% of their pay. Their adaptable salary is based connected growth, 50% of revenue. It could beryllium arsenic precocious arsenic 90% connected growth, 10% connected gross if you are genuinely a huntsman going aft caller logos, aliases it could beryllium 10-90, 10% connected growth, 90% connected consumption, if you're conscionable managing 1 of our apical 10 accounts.

Patrick Walravens

Do they kick that it's not adjacent to expect them as, hey, conscionable Mr. Salesperson to beryllium capable to power nan depletion astatine immoderate awesome bank, for example?

Michael Scarpelli

They -- immoderate complain, immoderate don't.

Patrick Walravens

Some different really...

Michael Scarpelli

They don't kick straight to maine because they cognize what happens erstwhile they kick to me. But astatine nan extremity of nan day, though, they tin power it.

Patrick Walravens

Can they? How do nan bully income guys do that?

Michael Scarpelli

Here's a premier example. What they -- if they're retired location and they're educating customers connected really really to usage nan product, really to usage Snowpark, we really spent our past income kickoff connected educating salespeople and really to spell successful and inquire nan correct questions to customers to place whether there's a Snowpark opportunity wrong that account. A Spark replacement successful AMR, Cloudera, whatever.

And they don't really person to waste anything. They conscionable request to amended nan customer and get a customer using, and that will lead to a customer having to bargain much capacity sooner. That's what we request our income reps to do. And by nan way, AWS, Azure and GooglePay each nan reps connected revenue, not connected bookings.

Patrick Walravens

I missed nan value of that.

Michael Scarpelli

What we're doing is what is successful nan industry, you're saying reps can't power a customer...

Patrick Walravens

Oh, you're saying they each do that?

Michael Scarpelli

That's really they're paying each nan reps.

Patrick Walravens

Yes, yes, yes. That's really they each did it. Okay. So let's displacement to Snowpark, which honestly, I don't get astir arsenic good arsenic I should. So this will beryllium nan opportunity.

Michael Scarpelli

Well, if you don't get it, I'm astir apt not -- I'm not arsenic method arsenic your son, that's for sure.

Patrick Walravens

Well, yes, if only he were here. But truthful what was nan opportunity pinch Snowpark that you felt for illustration you were not capturing before?

Michael Scarpelli

Spark workloads connected -- wrong information engineering. And what was happening is earlier we had Snowpark, customers would return their information retired of Snowflake, move it into different system, tally nan Spark Workloads and past move that information backmost into Snowflake. Well, that's very costly because you're paying to move that data.

By moving it correct wrong Snowflake, you cognize exactly, a, you're not having to salary for moving nan data; but b, you person this information and governance of Snowflake. Once you move your information extracurricular of Snowflake, unless it's successful a strategy that's highly secured and governed, you don't cognize what whitethorn hap to that information and wherever it could go. That's nan large use to customers.

Patrick Walravens

Yes. And what are nan -- astatine nan very precocious level, what are nan benignant of workloads that you would return retired of? What's nan benignant of study that you would do successful Spark that you wouldn't conscionable beryllium capable to do successful Snowflake?

Michael Scarpelli

As I said, I'm not a method person. I'm nan accountant.

Patrick Walravens

I don't know. It’s okay, it's okay. What do you do successful Spark that you wouldn't do astatine Snowflake?

Michael Scarpelli

No, you do analytics successful Snowflake each nan time. That's what we do.

Patrick Walravens

It's instrumentality learning stuff. Exactly.

Michael Scarpelli

Okay. And is that -- were they often doing it successful Databricks?

Michael Scarpelli

It besides enables you to constitute applications straight successful Snowflake by having that, and they thin to beryllium heavy analytics applications.

Patrick Walravens

Right. So -- but support it really simple. So fundamentally we're talking astir there's this instrumentality learning opportunity, right, that you guys were not capturing arsenic overmuch of arsenic you could. And successful fact, it was worse, right, because your customers would return nan information retired of Snowflake, put it location other and past there'd beryllium each these inefficiencies successful doing that. When did Snowpark travel out?

Michael Scarpelli

Our GA for Python conscionable came retired past quarter, and really conscionable pushed pinch nan income unit astatine our income kickoff successful February.

Patrick Walravens

Yes. I mean, truthful we're correct there, I guess.

Michael Scarpelli

We've had it for Java and Scala earlier, but Python is nan astir communal programming connection that nan information engineers want to usage today.

Patrick Walravens

Yes. By nan way, a small speech to that kid that you're talking about. I went, Zach, what's everyone utilizing ChatGPT for? He goes, "Well, Dad, I wrote a programme successful SCALA, and I needed it to be, I think, astatine Python. And truthful I asked ChatGPT to rewrite it for maine successful Python." And I'm like,"Did that work?" He goes, "Oh, yes." He submitted nan codification yesterday and sewage accepted. Crazy.

Michael Scarpelli

It is . That's really what we spot arsenic 1 of nan biggest usage cases for AI is really helping create code. I'm still trying to fig retired really ChatGPT itself is going to make money.

Patrick Walravens

Yes. Do you perceive to this week is all-in podcast?

Michael Scarpelli

No.

Patrick Walravens

No? I can't judge that I -- honestly, it's truthful freaking good, right? And it's truthful bully because that 1 feline is -- Dave's entree nan caput of task firms, truthful what are 40, 50 portfolio companies, and he conscionable benignant of tells you what he's seeing, right, which is very uncommon to get a VC to do that, right?

They're clasp things person to nan vest. But anyway, this week convention is each astir they're not judge either is nan bottommost line. They spot wherever there's a batch of benefits coming to consumers, right? They're not truthful judge that, that's not going to get absorbed by nan large tech players. They're not truthful judge either.

Michael Scarpelli

It was interesting. There was a -- it was good publicized. There was a Princeton grad student complete a weekend. He wrote an exertion successful Stream fto utilizing ChatGTP to show whether a insubstantial was machine-generated aliases written by humans.

Patrick Walravens

Yes. We...

Michael Scarpelli

That had complete 8 cardinal views.

Patrick Walravens

Yes. So we had this conversation, and our 14-year aged is like, "Yes, why are each nan teachers truthful freaked retired astatine Redwood High School astir this ChatGPT thing? What is it anyways, right?" And truthful we person nan information scientists and nan 14-year aged astatine nan table, right? And truthful Zachary goes [ph] inquire ChatGPT -- truthful she's successful ninth grades, he says, he goes, "What are you studying successful schoolhouse correct now?" and she goes, "Global warming." He goes, "Ask ChatGPT to constitute a 5 paragraph." She goes, "ChatGPT, constitute a 5 paragraphs astir world warming that's due for a 10th grader."

Michael Scarpelli

Yes.

Patrick Walravens

Right. That's nan beauty, right? That's due for a tenth grader. And past nan adjacent illustration is moreover crazier. So my woman is an writer and she's sewage a bunch of articles connected nan web, and she was astir to question and reply personification for a people that she's doing, who is nan CEO and female laminitis of a video crippled company.

And it was constitute nan question and reply questions that she should inquire nan CEO, laminitis of nan video crippled institution successful nan sound of Samantha Parent [ph], my wife. And it goes -- and she gets a small bossy erstwhile she's connected this topic, there's a batch of musts. And truthful literally, nan questions travel retired successful her voice. But really is that each going to use Snowflake? I don't know.

Michael Scarpelli

So what we spot happening, and I'm conscionable going to -- I'll telephone them ample connection models. We're ne'er going to beryllium nan 1 processing these ample connection models. Why? To create thing for illustration that is like, it's -- could beryllium $100 cardinal successful compute to create these models, but these models request to beryllium fine-tuned connected existent data.

We person nan data. So nan large point that we deliberation is enabling those models to tally straight connected nan information successful Snowflake is really we deliberation we will benefit. So companies will return those ample connection models. They licence and tally it against their Snowflake information for their business to fine-tune nan models for them. That's really we believe.

What's absorbing to us, though, is group to beryllium capable to constitute their ain queries and Snowflake utilizing ChatGPT to create nan queries. That's -- we're already seeing group do that. There's each kinds of YouTube videos astir that retired there.

Patrick Walravens

Give america an example.

Michael Scarpelli

You want to query a definite information and you want to constitute nan codification successful Python? Similar to what you saying, but not person it, really constitute nan queries themselves, you tin do.

Patrick Walravens

Wow, wow.

Michael Scarpelli

By nan way, they're not ever 100% meticulous [indiscernible] too.

Patrick Walravens

No, no. Okay. Cool. So let's spell backmost to what you're seeing successful nan macro. So lengthening income cycles, much approvals, each that worldly that everyone other is talking about.

Michael Scarpelli

Sales cycles themselves, I've been saying since time 1 of these ample Global 2000 accounts are 1- to 2- to 3-year income cycles, it really depends, but they each commencement small. We mostly -- erstwhile we onshore a Global 2000, nan mean size is for illustration 100,000. It's not for illustration they're big. It's their follow-on deals typically are bigger, don't spot thing changing there. What I do spot is customers wanting, and it's much pinch our existing customers wherever they've consumed faster than their statement rate.

They want further discounts. They want an economical use to do a caller deal. If not, they're conscionable going to proceed to bargain nether their existing contract, and we're conscionable saying, "Fine, bargain nether your existing contract." It's conscionable buying capacity, and that's what we saw. Some of our largest customers conscionable bought capable capacity to span them done to March alternatively than do an yearly contract. Why? Because they tin nether their contract, and they tin proceed to do that until July.

But successful July, they person to do thing of adjacent to aliases greater than their aged woody aliases they suffer nan discount they have. And truthful I'm proceeding much and much customers wanting different costs terms. If you went backmost complete a twelvemonth ago, liking rates were virtually 0 and holding rate wasn't a large deal. Today, you tin gain 4.9% overnight money.

Patrick Walravens

Okay. Good. We person 5 minutes. So let's unfastened it. I person a mates much I want to inquire you astatine nan end, but let's unfastened it up to immoderate questions from our audience.

UnidentifiedAnalyst

Who's your most, I guess, aggressively increasing competitor? And really is your strategy addressing it?

Michael Scarpelli

There is nary alteration successful nan competitory landscape. It continues to beryllium Google, Bitquery, number one. It has been since nan clip we went public. Databricks is astir apt -- I don't moreover want to opportunity number 2 because Microsoft, I would say, would beryllium number two. Databricks is nan different 1 and our full Snowpark really goes straight astatine them. But reality is they coexist successful galore of our accounts. Why we brought them into galore of those accounts early on, and we collaborated pinch them.

But it's decidedly Google, I would say, is nan astir competitive. And Google and Microsoft person nan champion -- they tin do a batch of bundling, which they do. Databricks doesn't person that pricing unit that a Google aliases a Microsoft tin put connected us. And by nan way, we've been dealing pinch that for years pinch those guys offering worldly for free, and free isn't free. You sewage to look astatine nan full costs of ownership.

UnidentifiedAnalyst

I deliberation astir nan nationalist opportunity and did you expect to…

Michael Scarpelli

So...

Patrick Walravens

You person to repetition nan question.

Michael Scarpelli

Yes, I'll do. The mobility was is, what do I deliberation astir nan nationalist assemblage opportunity and erstwhile do we expect to person FedRAMP precocious certification? I expect to person FedRAMP precocious certification very soon. We've submitted everything, and it's conscionable waiting connected nan -- getting awarded that.

It could beryllium -- arsenic acold arsenic I know, it could beryllium a weaker, it could beryllium 2 aliases 3 months waiting connected them. The nationalist assemblage opportunity is only upside because they're beautiful small. It's little than 1% of our business today. I do deliberation nationalist sector, successful general, and I'm not conscionable talking nan U.S., is simply a large opportunity for us. I conscionable came backmost from Korea, and I was really gathering pinch a consortium of companies that are advising nan Korean authorities connected their integer strategy and what they want to do. And this is going connected astir nan world.

There are information sovereignty issues that we're moving through. But UK is different bully 1 there. But overall, nationalist assemblage could beryllium 10% of our business, but it's going to return immoderate clip to get there. We're doing very good wrong authorities and section conception of authorities that doesn't require that FedRAMP high. But it almost seems for illustration nan goalpost's person moved wherever we thought FedRAMP Moderate was going to beryllium enough, and now they're saying, no, they request FedRAMP High. So hopefully, it's very soon. And past IL-5 will beryllium next.

UnidentifiedAnalyst

What's that five?

Michael Scarpelli

It's different higher certification.

Patrick Walravens

Is it different level? Is it really?

Michael Scarpelli

Yes. IL-6 [ph] is -- I don't deliberation we'll ever get there. It's astir apt nan astir rigorous.

Patrick Walravens

Yes, I sewage to study more. Okay. I americium funny though. So erstwhile you opportunity we're waiting backmost to perceive from them, who's them? Is it for illustration a peculiar department? Or is it -- really does that go?

Michael Scarpelli

So what happens is you person to -- you person to do each this worldly and show that you person each nan controls and everything and procedures successful spot to run successful nan FedRAMP High. And past you virtually manus each your files complete to a 3rd statement to do an audit. That 3rd statement -- and you request a sponsor wrong nan government. And we person a sponsor wrong nan authorities that has to salary for that 3rd statement to do nan audit of everything for them to motion disconnected on...

Patrick Walravens

Oh, is that really it works?

Michael Scarpelli

Yes.

Patrick Walravens

They person to salary for it.

Michael Scarpelli

It's a batch much than that, but nan sponsor has to salary for it.

Patrick Walravens

But they person to -- they person to salary for it. Yes, yes. Interesting. Okay. 1.5 minutes. What do you deliberation investors don't get arsenic good arsenic they should astir this story?

Michael Scarpelli

I get asked this mobility each nan time.

Patrick Walravens

Well, I thought it was -- I thought I had nan astir original question. Is it not that original?

Michael Scarpelli

I deliberation successful general, investors get it. I'm not going to opportunity they don't get anything. I deliberation a batch of group don't admit that really agelong immoderate of these migrations are going to return pinch customers and that we person immoderate customers that is going to return them 10 years earlier they person each their information move to Snowflake because their on-prem information authorities is truthful large and they move truthful slow.

I deliberation if you look astatine -- and a batch of group deliberation that we've been conscionable an on-prem information storage migration. Less than 20% of our business has really been these large on-prem information warehouses. We person a batch of caller digitally autochthonal companies that were ne'er an on-prem migration. And if you look at, we've signed up complete -- good complete -- I think, it's complete 1,500 on-prem information storage migrations of principally Teradata, and little than 100 of those person customers person really unopen Teradata down. It's a agelong tail.

Patrick Walravens

Yes, yes. That's a durable opportunity. All right, Mike. It's ever awesome to person you here. We admit it.

Michael Scarpelli

Thank you.

Patrick Walravens

Thanks for coming.

Editor: Naga



Read other contents from Beritaja.com at
More Source
close