Why cloud darling Snowflake took Salesforce’s money
With almost $1 billion in cash in the bank and usage increasing, Snowflake CEO Frank Slootman believes his cloud data-warehouse company is well positioned to navigate uncertain times ahead.
When Snowflake raised $479 million in funding just two months ago, CEO Frank Slootman thought the earliest the company might file to go public would be sometime this summer. Then the world changed.
"Where things are right now, the IPO market is effectively closed," Slootman said in a recent conversation with Protocol. Snowflake is an enterprise software rocket ship, building and honing a cloud-based data warehouse service that has seen a lot of traction as companies look to move huge repositories of data to the cloud.
That last funding round, led by Dragoneer Investment Group and Salesforce Ventures, valued the company at $12.4 billion. It brought the total amount of money raised by the company to $1.4 billion and welcomed a very interesting strategic investor in Salesforce, whose customers generate a lot of data and need a vehicle for the analytical queries that can unlock the potential of that data.
And it's fair to say that Snowflake has gotten the attention of all three major cloud players: Snowflake's service runs on all of them, which means they have some insight into exactly how many people are using the service. No one knows for sure what will happen over the next six months as the world grapples with the fallout of COVID-19, but crucial data will be generated that needs analysis, and Snowflake offers one of the more modern ways to do that.
Slootman, who once wrote a book called "Tape Sucks," led ServiceNow through its 2012 IPO and has now been at Snowflake for a little under a year. Via (what else) Zoom, we talked about life after COVID-19, database competition, and when the IPO window might once again crack open.
This interview has been condensed and edited for clarity.
I've been asking this question in every conversation I've had for a month: Have you seen any business impact from the pandemic or changes in the way people use Snowflake?
We're a consumption-based company, not a SaaS company. It's a utilization model, so we can literally see from one day to the next how we're doing, and in that regard, our consumption has actually paced ahead of our forecast, or our pre-COVID-19 forecast. Not crazy, but substantial.
Our company has exhibited very high net revenue-retention rates, and what that means is that our average customer close to doubles in consumption on an annual basis. So we're used to high consumption rates. And we're just pleased to see that the crisis has not affected that.
The place where we are seeing some drop-off is just a knock-on effect from industries that are on life support right now, there's travel and hospitality and all that stuff. So we're feeling that, but at the same time, we're also seeing the interest in what we do sort of catalyzing because of the crisis.
When things get dislocated like this, the need for data and analytics becomes much more pronounced. They're just trying to understand what the hell's happening to them, and also figuring out what are the inflections in the data, and trying to do very detailed regional county by county analysis on the data, and join that data with supply chain and economic data.
So with the consumption model, do people pay as they go with that? Do they lock in contracts for three months, a year or a couple years, with pre-agreed upon rates of consumption?
We have an on-demand business whenever people want to pay as they go, but most of our larger enterprise customers have contracts where they buy in volume, in exchange for discounts on that basis. It used to be they were one-year contracts, and now you see customers go into multiyear contracts to try to lock in better rates and terms over time. We're happy to try and accommodate that if we can.
Do most of your customers run on just one cloud?
Yes, I would say so. But I would also say that almost all of our customers have a multicloud posture, meaning that they are committed to a multicloud strategy. Now to what degree they execute on that, that's another question. But there's not one who will say "oh, we're 100% just on one platform." Literally nobody says that, everybody wants to sort of keep their options open.
With respect to the fundraising over the last couple of years, it has been pretty aggressive for the sector. Why has Snowflake chosen that strategy, and what is the plan for all of that capital?
I've been here almost exactly a year, so I can't be fully blamed for what happened before that. It doesn't really matter what I would or would not have done in that period of time, it is what it is. We have not accessed the prior round [October 2018] of funding yet, in terms of cash consumption.
The only reason that we raised more money is not because we needed the money, it's because we were pursuing a strategic relationship with Salesforce, and it was just part of the relationship coming together that they wanted to be part-owner in the company to potentially benefit from that relationship, which is a fairly common thing to do to.
So that fundraise wasn't really a fundraise; it was our strategic relationship, and it was the cost of doing business for us. And now we're sitting on close to $1 billion in cash, which, considering the crisis that's going on, gives a lot of people a higher level of confidence — especially inside the company — that we can weather a storm here over a prolonged period of time.
How did the Salesforce relationship come about?
I knew Salesforce from my prior life as CEO of ServiceNow and have had interactions over the years with the leadership there, and ended up having conversations with them fairly shortly after I joined the company. I actually teed up the idea that we make the integration between Salesforce and Snowflake a lot more seamless and frictionless.
Salesforce is a really big data set that gets used on Snowflake, and it is fairly difficult and painful to get that data onto Snowflake. It uses a lot of third-party tools, it has latency issues and causes frustrations with customers. So I said, "Hey, we can work together." We can make that a full thing where you go into Salesforce, you surrender your credentials, you select the clouds you want to have on Snowflake, and it just shows up there.
That would really un-silo the data, if you will. It's a term we use a lot: un-siloing data. Then Salesforce data can be combined with other data — marketing data, COVID-19 data, supply chain data, original data — and it will just create this effect, where your data doesn't sit in containers and can only be marginally be used. Salesforce agreed with that perspective.
And I have always commended them for having that kind of an enlightened perspective, because a lot of the big SaaS companies, they have a reflexive posture where it's like, "no, nothing ever leaves my platform." That's sort of how most SaaS companies react.
But Salesforce is much more confident about their role in the world and their data, so we were thrilled to be able to do this with them. We're in the middle of this project, we've done the relationship, obviously though we haven't delivered on the technology part of it; that's still ongoing.
There was a cost to that in terms of the equity piece of it. If they really wanted to integrate with Snowflake so badly, why wouldn't they just integrate? I mean, there's enough of a case for that as it is, I would think.
Well, it's just not abnormal for companies to want to participate in the value creation that happens in companies like ours, in part as a function of the value that they bring to it. We don't mind that, either. And I'll tell you why: I mean, the more Salesforce becomes bought into our outcome, you know, the better off we are, right? We'd like to have alignment with them.
We have enough enemies out there as it is. So it's nice to make some friends.
Who are your enemies?
Well, anybody that wants to own the data layer in the cloud, right? So that's quite a few people.
Do you find yourself competing more against traditional database players? Or, for lack of a better word, the cloud-native database market?
Most of the competition is with what we call the first-party product from the public cloud companies. So it's not the traditional database companies.
Are you finding that most people who pursue Snowflake already had their data in the cloud? As opposed to the transformational companies who are moving from on-premises to the cloud.
It's very common for us to engage with customers who have legacy data warehouses, the Teradatas and [IBM] Netezzas but also Oracle, SQL Server … in other words, they are analytical workloads, but they live on regular database management platforms.
But typically, these companies already have cloud strategies in place. So there's trying to move workloads to the cloud, but from the on-premises platform, so we see both. Some of them start in the cloud, some of them start on premises, but they already have cloud strategies in place. So it's not a stretch for them to make that move. And most people are not all cloud or on-prem, they're typically in transition and in a mix of both modes, that's the most common scenario out there.
Do you expect that to continue, like hybrid cloud being the natural endpoint for some of these companies that are a little bit later in their transition?
Well, it's not an endpoint. I think it's just in transition. We have like 30 years of database that is gradually going to move toward the cloud, it's going to take a long time. It's not just a quick switch that you throw; some of these transitions take years. And it's not that easy, so it costs money, it takes time and it takes attention.
A lot of these systems are heavily grafted into business processes. It's not easy to decouple these things and manage through that. So it's not a low-friction world, in terms of just saying, "oh, yeah, we're going to be on the cloud by October." It's a serious undertaking.
When we set up this interview, the world was a slightly different place. And obviously none of us can really understand what's going to happen over the next six months, but where are you thinking right now in terms of Snowflake going public?
Well, the strategy is the same as it was before the crisis. And the strategy was: let's get the company prepared to go public in terms of risk control, compliance and all that stuff. Which we weren't, so we will achieve that milestone in the next several months.
Where things are right now, the IPO market is effectively closed. Henry Kissinger once said, nothing clears the mind like a lack of options, you know? [Editor's note: Close enough.]
We're well capitalized, as you pointed out, so there's really no pressure whatsoever. In that sense, we'll wait for things to take a turn.
I lived through the year 2000 meltdown. And the IPO market was closed for five years, if you recall, and traffic was great in Silicon Valley for years. Obviously, the sentiment in the market has to be to our liking, that the multiples are in a place where we think it's compelling to do.
I don't spend a lot of time thinking about this, I only think about building the company one step at a time. And an IPO is just a liquidity event that comes and goes very quickly. It's not that big a deal.
Get in touch with us: Share information securely with Protocol via encrypted Signal or WhatsApp message, at 415-214-4715 or through our anonymous SecureDrop.
It's a big deal for employees.
Sure, but one thing we've done for employees, we executed, over the years, a couple of tenders where employees can sell some of their shares to take some of the pressure off the liquidity that they feel, especially people that have been there a lot of years.
I am not going to hold this IPO up if I can help it, but at the same time, we're not going to sell our equity at a ridiculous discount, right? This is a valuable company, and we feel it's important for everybody that gets recognized in that process.
- How to IPO in the middle of a pandemic - Protocol ›
- HashiCorp's big step toward a central role in the cloud - Protocol ›
- Big data has eaten the world. Now what? - Protocol ›
- Snowflake’s historic IPO signal of surging demand for cloud-based enterprise software - Protocol ›
- Move fast, break nothing: Why investors like JFrog’s approach to modern software development - Protocol ›