SNOW: updated beliefs

Investment Analysis Clubs Saul’s Investing Discussions
1

Thought I’d rehash an old idea (https://discussion.fool.com/fsly-and-updating-beliefs-34643625.a…) as I digest these numbers from Snowflake.

Prior Belief: Snowflake would add fewer customers percentage-wise, but the decline would be slow.
Update: After adding almost 15% sequentially on average all 4 quarters of last year, they added 9.5% last quarter and 10.1% this quarter. That’s more decline than I’d like to see.

Prior Belief: NRR would remain incredibly strong at 160% (honestly 165%+)
Update: No update. This happened – 169%, in fact.

Prior Belief: Snowflake’s Revenue growth would be 19% or so sequentially.
Update: No update.

Prior Belief: Full Year guide would be at least 1.060b but not 1.080b+.
Update: No update. It was 1.070b.

Prior Belief: Snowflake’s RPO would grow not at 40%+ sequentially like it did in Q2, Q3, and Q4 last year, but certainly faster than the 7% sequentially it did in Q1 and maybe as fast as 20%+ sequentially.
Update: Wow, they only came in at 7% again. Not what I was expecting at all.

Prior Belief: Snowflake would show continued strong growth in $1m+ customers
Update: I’m surprised these only grew 11.5% sequentially after growing 16%, 18.5%, and 35% the last 3 quarters. Maybe I was expecting too much.

Bonus Prior Belief (not mine): WSM expected $275m+ revenue. (https://discussion.fool.com/bear-can-i-ask-specifically-what-num…)
Update: It was 272.2m.

Conclusion
Everything I believed before Snowflake reported was either unchanged or negatively updated. That means I have to lower my expectations…in other words, I don’t think we can consider this quarter as Snowflake “holding serve” (to borrow a term from my friend StockNovice). At least not based on what I was expecting. Therefore the options are hold, trim, or sell. I wouldn’t be adding, obviously.

This is an impressive company with impressive growth and impressive visibility into that growth. I think they’ll hit 100% next quarter and 95%+ in Q4. As the CFO says, their NRR will come down from ~170, but if data consumption trends continue it will stay at 150+ next year, so even if customer growth continues to underwhelm this year revenue will probably still grow around 80% next year which would mean around $2b+ revenue. Further, I believe 10 years they’ll be something like a $500b company. So I’m seeing a nice 5x as likely long term…but I feel I can grow my money faster elsewhere in the near term. Maybe at some point the valuation will entice me more, but I’m out for now.

I reiterate that anyone who wants an easy long-term win will likely do well to keep their money in Snowflake. That can also be said (with more certainty…but less upside) of owning an S&P index fund.

Bear

179 Likes
2

First of all, Bear, I love the format of this post with your prior belief and updated belief clearly stated.

Prior Belief: Snowflake would add fewer customers percentage-wise, but the decline would be slow.
Update: After adding almost 15% sequentially on average all 4 quarters of last year, they added 9.5% last quarter and 10.1% this quarter. That’s more decline than I’d like to see.

I’m sure you listed to the conference call, and you would know that Snowflake is mainly targeting larger customers, with their focus being on the Global 2000.

Adding larger customers takes more time. So I’m not that concerned (yet) about the slowing customer adds. Even with slow customer growth, revenue growth should remain strong. If I remember correctly, CFO said less than 1% of in quarter revenue came from customers added in the quarter. And less than 10% for full year revenue will come from customers added this year.

My personal opinion is that Snowflake could target mid-market customers and grow the customer base faster. But they are choosing to go after the larger customers, which they see as a big market opportunity.

14 Likes
3

My personal opinion is that Snowflake could target mid-market customers and grow the customer base faster. But they are choosing to go after the larger customers, which they see as a big market opportunity.

Isn’t that what Fastly did? Net ran the other road and see how much better they did. So my Question is who is Snow’s Net?

Andy

12 Likes
4

The problem is, if customer growth is going to slow due to a focus on larger customers, then we should see corresponding bumps in RPO to go along with these new, bigger customers, but we’re not seeing that. We can assume that it’s offset in part by some RPO of other customers moving to current quarter revenue, which is probably what is happening to an extent and explains the slower deceleration in revenue, but it obviously is not enough to translate to accelerating revenue growth, at least on a % basis. I don’t have the raw $$ figures in front of me at the moment, but irregardless, I’d ideally like to see much greater QoQ growth in RPO, which would give me confidence that they are signing these newest larger customers to big, multi year deals. The lack of greater QoQ growth to correspond with the 41 or so new F500 clients gives me pause.

Maybe it’s just the land, and the expand, what with 167% NRR will cause RPO to jump in later quarters and this will be just a temporary blip. Maybe too this is amplified by the change in incentives last year that caused those spikes in RPO growth they’re lapping now, since there aren’t as many renewals this year as in the past, but still, I wouldn’t be surprised if the next quarter or two (or three) are similar before things accelerate again. I’ve trimmed my position a bit today but still have a significant position (~7%), for now.

7 Likes
5

This has already been posted but it is very important when looking at the RPO numbers. Please read the thread below.

https://twitter.com/jaminball/status/1430706374237777921

Basically there were incentives initiated last year to sign multiyear deals and this is the first quarter where Snowflake lapped the first quarter with the incentives. Bottom line: RPO YoY is currently deceiving.

I agree there are other metrics that could be viewed as concerning, but the RPO number needs this clarification.

19 Likes
6

Also, from the conf call:

“As a reminder, in Q2 last year, we sold our largest multi-year contract ever, a three-year $100 million deal. While the multi-year component of new booking sets up a difficult comparison, we saw a net - we saw new annualized contract value accelerate compared to the year ago period. This is why RPO and revenue must be evaluated together in a consumption-based business model.”

7 Likes
7

Hi Junomean,

I did see that earlier and read the Twitter thread, but it doesn’t fully add up in my mind. Yes, we’ve lapped the incentives, meaning there are fewer renewals this quarter because they won’t renew for another year or two, and these are already reflected in the existing RPO. I get that. But they consistently add 400-500 customers every quarter including 13-19 F500 (excepting Q1) since Q2 ‘20. Those first 3 quarters of multi year, larger client focus resulted in quarterly RPO increases between 220-405m. I would expect the %s to decline from 47/35/44%, but the net increases in dollar terms have dropped to 99 and now 97m. As a said in a post last night, I expect this may be lumpy, but if RPO doesn’t jump again next quarter in sheer dollar terms, we’ll be looking at RPO growth rate declining from 206% to 122% to ~75% next quarter. That’s not what I want to see, and with a usage based model instead of recurring SaaS revenue, we can assume wilder swings in market sentiment if they don’t beat earnings handily. Please, if I’m way off base with the above, tell me why. I’m trying to learn and surely am not accounting for some variables of import.

2 Likes
8

Further, I believe 10 years they’ll be something like a $500b company. So I’m seeing a nice 5x as likely long term…but I feel I can grow my money faster elsewhere in the near term.

SNOW mgmt said in June that it expected $10b in revenues in 7 years, or 2028.
Let’s say in 10 years (or 3 years after that) they actually wind up at $20b in revenues.

Salesforce is a titan and at $20b now. Their valuation has shot up quite a bit since the 2020 boom for growth software/cloud stocks, and after a good ER they are about $265b.

A decade is forever in tech, and I am just not sure SNOW has such a moat that would keep customers from eventually migrating to newer/better cloud-based data storage solutions that comes out in 4-6-8 years from now.

But even if they are still dominant, they would be valued at double what CRM is today for the same revenue, and CRM is actually SaaS.

If I was going to own SNOW, it would probably be over the next 12 months, when stock most likely to irrationally keep going higher due to crazy growth. Once their growth starts to naturally decline, the stock upside may be much more limited. So you could own it for 10 years, but if it does grow from $90b to $500b over 10 years, that is about an 18% CAGR.

Dreamer

16 Likes
9

I can’t seem to find the posts talking about this, but I recall there was a discussion here regarding how Snowflake gets or reports revenue. From what I remember, they are a quarter or so behind. This means that even if they landed those large clients and contracts, we will not see it in their reporting for at least a quarter.

I could be very wrong here, as I’m going off of recollection. So please correct me if I’m wrong.

-Griz

10

The thread and explanation from Jamin Ball is helpful. Thanks for posting it.

At the same time, it’s worth noting the analyst comment he attached to the beginning of the thread modeled for $1.8B in RPO this quarter. So, while figuring out the revenue/customer/RPO relationship for one of our companies is always worthwhile, we still have to acknowledge these mental gymnastics weren’t needed for SNOW in prior quarters.

Take that as you will. I’m still mulling it over myself.

16 Likes
11

So, while figuring out the revenue/customer/RPO relationship for one of our companies is always worthwhile, we still have to acknowledge these mental gymnastics weren’t needed for SNOW in prior quarters.

When the RPO in Q1 2021 was tepid as well, we explained it away due to seasonality and expected it to rise this quarter. However, this quarter the explanation has to do with the way management incentivized multi-year contracts.

Now RPO’s or Remaining Performance Obligations, represents the total future “performance obligations” that come from signing new contracts. So it’s just the sum of the invoiced amount and the future amounts not yet invoiced for a contract with a customer. This works best if you have signed up a customer with a fixed subscription, such as Cloudfare tends to do.

But Snowflake pricing is based on consumption, not as much on subscriptions. Is RPO a good measure of Snowflake’s forward revenue, or was it always flawed to use it for $SNOW to begin with? I’m not an expert on this, so please let me know if I am wrong in my understanding.

Here’s an explanation of their pricing structure: https://www.in516ht.com/snowflake-pricing-example-size-matte…

Key Takeaways

  • No fixed charge per user or per CPU core. At Snowflake, there are no fix charges – all charges are usage based.
  • No hidden quotas or price bumps. When you run queries in Snowflake, there are no added usage quotas or hidden price premiums. You pay only for what you use.
    - If no queries are actively running, you can automate your warehouse to shift into suspend mode. Once suspended, charges are also suspended for idle compute time.

Additionally, Snowflake as also indicated that revenues don’t meaningfully show up until 6-9 months after onboarding a new customer and that the longer they stay, the more they end up trusting the platform and spending more. Their CFO indicated that this will cause lumpiness in their revenue numbers. Since they onboarded a large number (19) of Fortune 500 in Q4 2020, and another 18 this past quarter. Neither of these quarters has shown up in revenue yet.

What we do know that they have stickiness, once a company is onboarded, they tend to keep their platform. This is supported by their NRR of 169%. This should bode well for a nice lump increase in revenues in Q3 and Q4.

So where does that leave us? Snowflake is an incredibly fast growing company with a easy-to-use, scalable data cloud platform that their clients appreciate. After their latest earnings call, analysts have raised their price target to, on the low end of $295 to $375 on the high end, with most around $330. However, if you are unsure, waiting a quarter or two for the “lump increase” in revenue will ease your anxiety. This latest earnings report didn’t shed as much light as we would have wanted, which is not necessarily a negative. We are simply in chapter 4 of book 1 of a large trilogy, something like The Baroque Cycle by Neal Stephenson (which I heavily recommend).

Quite a bit of data to mull over here and I am looking forward to more responses or constructive criticism of my understanding.

33 Likes
12

Hi gibbersome,

I could not agree more with your post. It does take a bit of mental gymnastics to follow Snowflake’s business model. I think that’s how those on this board, utilizing crowd-sourcing of sorts, have an advantage when evaluating the choice to invest or not in Snowflake. I had not put my notes together into a cohesive post before you wrote yours. It’s difficult to keep in mind what Snowflake management has been trying to be clear about, including what you wrote here:
Their CFO indicated that this will cause lumpiness in their revenue numbers. Since they onboarded a large number (19) of Fortune 500 in Q4 2020, and another 18 this past quarter. Neither of these quarters has shown up in revenue yet.

I’d only add that Snowflake management has also said on each of the Conference Calls since IPO, ‘it’s standard for enterprises to sign up for one year the first year and then renew with multi-year contracts.’ I think this only adds to the ‘lumpiness’ pointed out repeatedly throughout this Q CC.

I don’t believe,we need to try and somehow group together Rev growth, Customer Growth and RPO together; but, I do believe we need to keep in mind how easy it is for Snowflake customers to use up their pre-paid account balance ahead of schedule and how many Fortune 500 companies are going to be ramping up their utilization this December quarter.

Thanks again for your post,

Jason

23 Likes
13

Thank you to both gibbersome and Jason, very helpful stuff that addresses a lot of what I was concerned about.

‘it’s standard for enterprises to sign up for one year the first year and then renew with multi-year contracts.’

  • this would explain why new customers are not making RPO ‘pop’ as much these past two quarters, and should bode well for renewals next year (possibly compounded with renewals of 2-year contracts from last year for an extra large pop?).

‘What we do know that they have stickiness, once a company is onboarded, they tend to keep their platform. This is supported by their NRR of 169%. This should bode well for a nice lump increase in revenues in Q3 and Q4.’

  • I think this is the million dollar question. I was hoping for more this quarter and it did not materialize. If RPO or revenue does not have a nice lump increase next quarter, it’s possible investors may be in for some short term volatility.

  • Frank

4 Likes
14

It may be helpful if I borrow from a couple reliable posters here to explain what I meant when I wrote, in my prior post, I do believe we need to keep in mind how easy it is for Snowflake customers to use up their pre-paid account balance ahead of schedule and how many Fortune 500 companies are going to be ramping up their utilization…

Revenue recognition, Diablito
Snowflake is Alteryx (AYX) in reverse in terms of its revenue recognition method. As Alteryx-shareholders painfully learned, recognizing a huge portion of revenue upfront accelerates reported revenue growth in good times but also creates amplified problems when new deals and renewals get scarce. Snowflake investors do not have to worry about that. In fact, because consumption = revenue, investors have perfect visibility into how Snowflake’s business is actually doing each quarter.

From Snowflakes Q1 Slide deck:

“Primarily billed annually in advance with some on-demand in arrears.”

Pricing, Coldmountain

Snowflake user here – I’d like to highlight a couple of things noted by either other users of snowflake, or in the earnings q&a.

Snowflake charges on use, not on storage (or at least the primary part of their revenue is use). It will take time for new and large customers to figure out just how easy it is to use snowflake. Once these large users catch on, snowflake revenues will REALLY ramp up.

At a certain point our data we hosted on snowflake started to get more popular among users, so more people started to notice it was a bit ‘slow’ (very relative term…). Snowflake makes it very very easy to increase the horsepower on your applications, and it is in terms of 2x, 4x, 8x and maybe now 16x or more. Each of those costs 2x the tier below, so moving from the slowest rate doubled our cost per mile (so to speak). It is still usage based, but the rate is now double. You can turn it up or down any time, for any reason, and billing is immediately reflected. It was therefore painless to double our payments to Snowflake! I didn’t give it a second thought when I upped our horsepower until the IT supervisor asked me if anything had changed, because we were now paying snowflake double what we paid a month or two ago! Very clever of snowflake.

Second, the new compression technology that snow rolled out will make it cheaper for them to hold data, and easier for users to use it (again, snowflake revenues ramp up).

Long snow

Me here:
I think what we saw this quarter was: 103% increase in YoY revenue, spend as it was utilized by customers. RPO +107% YoY, a drop compared to last q could be explained, justifiably, in many ways given how Snowflake on-boards customers and allows for utilization of contracted spend.

I don’t hear anyone saying this isn’t a world changing company that’s not going to be amazingly profitable. I added before earnings, making it a 14% position and I’m holding for now.

Best,

Jason

22 Likes
15

My personal opinion is that Snowflake could target mid-market customers and grow the customer base faster. But they are choosing to go after the larger customers, which they see as a big market opportunity.

Isn’t that what Fastly did?

Fastly and Snow do have some similarities:
• Both have innovative platforms and are commonly regarded, technically, as best-of-breed
• Both have a consumption revenue model (not contractual recurring revenue)
• Both have stronger appeal from large customers/users

There are also some important differences:
• Snow has a superior, experienced, management team; Fastly has an inferior management team
• Snow has contractual spend agreements with its customers; Fastly was pay as you go (or don’t pay if you don’t go).
• Switching costs for data storage are high. One has to not only migrate the data itself, but also re-build the applications that use that data for the new platform. OTOH, switching costs for CDNs are low. Despite differences in their underlying technology, CDNs have been around so long that despite under the cover differences, it is easy to spin up a new CDN on your existing web pages, which don’t need to be modified. This has also lead to many customers using multiple CDNs simultaneously for their pages - that’s how easy it is to adopt CDNs.

Now, Fastly is trying to build up an Edge Computing business, which would be stickier than CDNs, but Fastly’s product is way behind Cloudflare’s in terms of customer availability. The longer it takes Fastly to spin that up, the more customers will in the meantime adopt other Edge Computing solutions, which will make it harder for Fastly to “steal” them away to its platform, even if it’s better. It would have to be a whole lot better, which doesn’t seem likely for some/many edge compute applications.

And, probably the biggest difference is that Snowflake customers of all kinds love the product and regularly come back for more than they originally contracted. Fastly appeals to a more narrow set of potential customers (ones with rapidly updating content, like NYTimes or TikTok).

That said, I do think one has to understand that Snowflake’s targeting large customers results in not just a longer deal-signing time, but also that migrating existing workflows to Snowflake takes months. This is actually great (it’s not years), but it’s also not immediate. I haven’t yet listened to the recent conference call, but back in March, CFO Mike Scarpelli said:

“I want to stress, it takes customers, especially if you’re doing a legacy migration, it can take customers six months+ before we start to recognize any consumption revenue from those customers because they’re doing the data migration. And what we find is they consume very little in the first six months, and then in the remaining six months they’ve consumed their entire contract they have.”

Snowflake is not the kind of business where you ramp up marketing or sales efforts and you see results within the quarter or even next quarter. That said, one would have expected that RPO would have increased more than 7% or so they reported last quarter. At this point, however, my inclination is to give SNOW management the benefit of the doubt and take my time to analyze what was said, and how it was said, in the recent call. This is not a Fastly stumbling around trying to figure out how to run a business around a superior technical solution, this is an experienced team planning and executing for the long haul.

60 Likes
16

But Snowflake pricing is based on consumption, not as much on subscriptions. Is RPO a good measure of Snowflake’s forward revenue, or was it always flawed to use it for $SNOW to begin with? I’m not an expert on this, so please let me know if I am wrong in my understanding.

I’m surprised many of you think that RPO is not important or not a good measure for Snowflake. I suggest you guys read the 10th page of Snowflake’s earning slide - https://s26.q4cdn.com/463892824/files/doc_financials/2022/q2….

RPO and Product Revenue are the two most important metrics they focus on, said in Snowflake’s own words! We would not know more about Snowflake than the management in the company. Snowflake is a company with such an exciting story that people don’t believe they could really slow down and try to find excuse for Snowflake to hide its slowness. IMHO, this could be dangerous for investors. The numbers would almost always tell the truth and what Snowflake’s numbers tell us is that the customer growth has slowed down, so does RPO.

Someone brought up that enterprise tends to subscribe just one year and then potentially renew with a multi-year contract. I don’t think this is a good thing for Snowflake. Don’t that mean the company can also choose to NOT renew in one year? Why would a customer try Snowflake for one year if they’re convinced by Snowflake’s product? Couldn’t it be that the customer is not fully convinced by Snowflake actually and they may potentially have other options to replace Snowflake in one year? As comparison, ZSclar in its last earning call highlighted that they signed several multi-year contracts.

Some other comments in this thread reminded us of that Snowflake takes almost a year to see the revenue generated from new customers actually ramping up, so the revenue from the new 18 Fortune 500 customers will pay off. Well, isn’t the delay of reflection in revenue exactly why they saw slow down in both RPO and customer growth but not in revenue growth? Won’t the slowdown of customer growth cause a slowdown on revenue growth in a year? Think about that! In other words, Snowflake’s hyper growth is mostly from D-NNR, while there’s less and less from new customers. Investors will eventually realize that from Snowflake’s future earning calls, when it could be too late.

Good luck,
Luffy

14 Likes
17

Having read all threads and posts on the subject, there are two themes in play here: performance and potential. There is a third one that was brought late last week but ignored after the ER: competitive landscape and use cases. Different questions, different answers.

PERFORMANCE:
I fail to see any solid argument for an issue with SNOW. Assuming J. Ball and the market are right on how the RPO numbers are to be interpreted, then everything else are just guesses. Since the market spoke loud and clear today, the folks still going for the RPO numbers need to make a stronger case. Why is Ball (and Altimeter, one of the larger shareholders) wrong?

The bigger, strategic issue that came out here before the ER was the usability of SNOW for machine learning. If Databricks plus your choice of storage is better for ML, as some have argued here and elsewhere, then that would presumably exclude SNOW from the use cases requiring the most usage. Is that so?

It was interesting to hear management address ML right away in the prepared remarks. There is of course a good chance that what they said was just “CEO speak.” But at least it was said convincingly: short, clear, firm but no over the top statements. Clear case was made for future revenue: big customers are spending more and more. It takes a while to see that usage materialize, ITL is growing great, you can use Snowpark and do all the ML work you want, etc.

Then no analyst really drilled into that–or really into anything else. I wish they did. As is, all SNOW had to reiterate in response to a mild question about Snowpark was developer feedback and partner integration plus the feasibility of what they offer.

As the results, so the call lacked excitement on both the management and the analyst side. But both had a machine-like precision to them: non-Gaap operating expenses continuing on a textbook trajectory and management was as professional as any. If the goal was to hide issues, mission was accomplished–but thinking that way would block any decision making on the investor side.

Compare that to the ESTC report that I listened to right afterwards. ESTC were kind to offer illustrative examples of adoption (something that would have made the SNOW presentation more relatable for non-tech folks), but for the rest it was essentially “we are a security company, too” PR pitch (plus guidance for lumpy revenue growth in Q3 and Q4).

Those who say that SNOW is facing trouble (and even is the new FSLY!) may turn out to be right but if so that will be either because they know something they are not sharing or because they got lucky. No solid bearish argument has been made so far here. Gotta start by explaining why Ball and the market misunderstood the RPO numbers.

FUTURE:

On the other hand, the question whether SNOW has much of a reasonable upside from here is practically unanswerable. SNOW came out of the gates at valuation+market cap combo that challenged everyone, including Saul’s model and Saul himself. To assume that SNOW can greatly outperform and that Bear is wrong is to assume that SNOW will do something nobody else has done–and that CRM (or ADBE) is not a good comparison (because if it is, then Bear is too optimistic, as others have pointed out already).

So if someone is sure they have better ideas as of today, that’s hard to argue against.

OPEN QUESTIONS:

I would love to hear from muji or someone else with deep understanding of the product and the use cases and the competitive landscape whether:

–SNOW is right and Snowpark is all you need for complex ML uses or critics are right and Databricks plus your choice of storage is much better for the latter. I guess if there are only two serious players, there is nothing to worry about. But is there more to the story? There usually is :slight_smile:

–If SNOW’s competitive advantage is chiefly about allowing data sharing and offering a data marketplace, then how big of a guesstimated TAM is that–especially if it is technically true that Databricks has the advantage in ML? How close to realization are those uses considering that enterprises are just getting started with the “base case” uses?

41 Likes
18

I’m surprised many of you think that RPO is not important or not a good measure for Snowflake. I suggest you guys read the 10th page of Snowflake’s earning slide.

Thank you monkeydluffy, I overlooked this.

Concerning the IPO issue, provided we agree with he CFO about the lumpiness and about Jamin Ball about management incentivizing sales last quarter, we do no have to make excuses for the poor RPO growth.

monkeydluffy correctly pointed out that RPO and product revenues have been identified by Snowflake as the two growth metrics they are focused on (Page 10 of their earnings presentation). If they cannot execute there and combined with the slowing customer growth, can the market justify their high valuation?

I don’t want to wait to find out when I am more confident in other companies. In the meantime, with RPO slowing and customer growth decelerating, I see a ceiling on their stock price in the near future as the company grows into its lofty valuation. If I am wrong and the NRR(168%), Revenue growth (103% YoY), Operating Leverage (9%) should have been trusted more, I’m willing to re-enter the stock after the next earnings call.

I also exited my position in $SNOW.

4 Likes