Snowflake hit $600M ARR at IPO growing 115% annually, with 162% net revenue retention and 93% consumption-based pricing. Gross margins sat around 60% with 3,500 smaller customers plus major enterprises, though 85-90% of revenue came from the US.
Questions this video answers
Approximately $600 million ARR.
~162% NRR (noting a 189% figure in 2020 described as an anomaly).
~93% consumption-based revenue (about 7% under fixed contracts).
Gross margins approached ~60% with a projection of almost 70% for the following year.
About 90% in North America in 2020, declining to ~85% later.
Chapters
A Snowflake. What must have been the hottest IPO of the last 12 months. What everyone was talking about, even if folks were always quite sure what a data lake was or what Snowflake did. But it went public and quickly rocketed to an $85 billion market cap valued even higher than Zoom. It settled a bit lower since then, but it deserved it. It deserved it. Snowflake IPO'd at 600 million in ARR, incredible as it is, but growing 115%. At 600 million, 115%. With 162% NRR. This and UiPath have NRS and growth rates that we've really never seen before in business software. More than doubling at 600 million in ARR and existing customers growing, adding 62% more revenue each year. It's crazy. It's crazy. And we can't all be Snowflake. Most of us will not grow 150% at most of the cloud leaders today didn't and they're doing just fine. But there's a lot actually we can learn from Snowflake, even in the enterprise. Number one is that 93% of the revenue is consumption based, yet it still has a SaaS multiple. And let's slow that down for a minute because I think one thing a lot of us have learned, especially over the last 24 months as more and more cloud companies have gone public, is revenue doesn't all have to be under this traditional per seat license and it doesn't all have to be truly recurring as long as it recurs. So what do I mean is, yes, Snowflake is 93% of its revenue is consumption based. Only 7% is under fixed contracts. And in fact they can only recognize the revenue as this consumption occurs. They don't smooth it, they have rollovers. Yes, they bill in advance, but they have to true up the numbers. And so it's not what we used to love, what we were trained by Salesforce to do, which is, which is charge folks for seats even before they even use them. That was the classic Salesforce model to have this hyper locked in, latched in per seat model. Datadog's the opposite. They're like, yes, as you scale, you're going to prepay a lot of it annually sometimes, but it's a consumption base. That's how you're going to pay. But Twilio showed us it works and Datadog shows it works. And what it really shows is do pricing the way your customers want to pay, make them comfortable. And with Twilio, what We saw when we talked with Jeff Lawson at the last big Saster event is he's like most Twilio customers prefer to pay consumption because they're part of the cost of revenue that they forward bill. They forward bill to their customers. So it's so if they're going to end up billing a client a dollar a minute, paying Twilio 2 cents a minute or whatever it is, it's very natural, it's very aligned. And Snowflake's not the same but the customers want to pay on consumption. We've seen the same thing with Datadog. Most of Datadog's customers want to pay consumption. That's how their developers want to pay it with fastly. It's been mixed. With Shopify it's been mixed. But if your customers want to buy into consumpt model, let them. Let them. And the real key from that Snowflake shows us is if the AR is as high as description model Wall street doesn't care 162% NRR. I don't think Wall street cares if it's under a contract, if it's written on paper, if it's an email, if it's a. If it's a pinky promise. 162% NRR is what creates these crazy multiples in SaaS. Okay, point two a snowflake gross margins. We just chatted about this a bit before with Palantir, but Snowflake has a lot of costs. It is a data lake and that's significant compute and storage costs. And so its gross margins were not high historically, but again going into IPO, going into hundreds of millions of AR, the gross margin approach 60% and they're projecting almost 70% for next year. So it's a reminder that again get your house in order the early days. If you want to do give away professional services, if you want to run lower margins in the early days, just get it done. So long as as you get bigger, especially as you approach ipo you have the same sort of margin other structures as basically everyone else. The third interesting learning from Snowflake. It's just a reminder that world class NRR does not have to come down. Not really. So Snowflakes NRR is at a stunning 162% each. For each hundred dollars again of existing customers they pay $162 next year. Now let's come down a bit. It was a jaw dropping 189% in 2020 and it but that may be looking at the data that may be more an anomaly of how folks upgraded because it's been pretty consistent in that 160 range since then. No matter what, it is a reminder that as you approach Almost any amount, 10 million, 20 million, 100 million, 1 billion ARR, like Snowflake, your NRR does not have to come down. And it's a reminder that if it does, maybe it's a flag. Maybe you're falling behind competitively, maybe you're not staffing up as much in customer success as you thought. It's a reminder that you probably that these accounts can can grow forever. Now, because of the nature of products like Snowflake, you may not see 162% NRR, maybe you see 120, 110, 130, 140. But aim for it to stay that high or increase for essentially forever.4. Just a small one from Snowflake about going global. But interestingly, again, in part because of the nature of the product, they've been slow to go global. And even as late as 2020, 90% of the revenue was in North America. Today, it's still 85%. So we've seen this all over the place. There are folks that have started outside of the US like Monday and Wix that often have the majority of their revenue outside the US there are folks like RingCentral for telephony reasons that were slower to go outside of the US and there's folks like Snowflake that have a reason. But wherever it is, chase it, grow it, lean into wherever your customers are. And the last interesting learning from Snowflake is just around the long tail. Snowflake on its surface is very enterprise. They have 65, $1 million a year customers and they're the fastest growing of all. They're growing 110% year over year, so very enterprise. But they're fueled by 3,500 smaller customers in this long tail. The overall customer base is still growing at a stunning 84% at almost a billion in AR, at over 600 million. So slow that down for a minute. A lot. What we see with a lot of folks is that sometimes customer count slows down, slows down as you approach 30, 40, 50, 100 million in ARR. But it doesn't need to. Snowflake's customer count is still growing 84% at 600 million in revenue. And to make that work, you want a long tail. You want thousands of customers loving you, trying you and buying you, even if, even if the end goal is to get as many of them as possible up to a million. So pretty cool. With Snowflake, we're not all Snowflake. It is an outlier. Right? And we are not outliers. But we can learn from outliers in terms of terms of KPIs and goals that we can aspire to emulate. So keep 3,500 customers happy. It can happen. Those small ones can grow into large ones in the enterprise. Even if they start small, they can grow with 160% NRR. It's crazy to see that it's Snowflake and UiPath and you can have you grow 100% at 600 million AR. Well, maybe we all can't do that like Snowflake. But what it means is there may be no natural ceiling. For most of us, there may. We can, we can get better, we can grow our tams, we can increase. And Snowflake should inspire us to really never slow down.
