Going Deep on Cloudflare and Datadog with Mostly Borrowed Ideas (Transcript)

Transcript of Podcast #17

Dec 13, 2022

You can listen to the audio here:

🎧 Going Deep on Cloudflare and Datadog with Mostly Borrowed Ideas (MBI)

Liberty:

Hello my friend. Thank you for joining me today for... We were talking about before this recording about how we're going to try to recreate back when you lived in Ottawa, we used to take walks and just talk about everything and about business and companies and all that kind of stuff. So today we're going to try to recreate some of that. It's not an interview, it's a conversation about a couple of companies that you've covered in detail. Small disclaimer. First, this is not investment advice. This is a general thing about everything we say, but also it's not because we talk about something that we're saying, "You should invest in this now." I think we're both the kind of investors that like to learn about interesting companies and then eventually they may become good investment opportunities, but it's better to learn about them in advance. But anyway, I just want to welcome you. Thank you for joining me today.

MBI:

No, thank you so much. I think I was the first guest to appear twice on your podcast. And then David Centerra kind of came out of nowhere and married to [inaudible 00:00:57]. So I had to be here to make sure I'm at least at [inaudible 00:01:01], but given how relevant David Centerra is, I don't think I'll be able to keep up with him.

Liberty:

It's a real horse race now. There's competition. That's good. So the businesses, I guess everybody has seen the title of the podcast by now, but we're going to talk about CloudFlare and DataDog and there's not going to be exhaustive if people want all the detail, all the history, all the financials and all that stuff in the show notes, I'm going to link to your two excellent deep dives and maybe to some other resources. But your deep dives are full of resources and links, so that's good hubs too. It's so funny how things can change quickly. Because I think when we first started talking about these businesses and when you put them on your own docket for deep dives, they were trading at incredibly high multiples, but everybody loved them and the multiples kept going up, which means that people kept recalibrating their opinions of the companies in more positive lights. This thing is worth 25.

No, it's worth 30 times, worth 35. People kept loving them more basically. And now just a year later, all these stocks are down like 50, 60, 70, 80% in some cases, and the multiples have compressed tremendously. That's not the performance of the revenue growth and all that stuff. The gross margins haven't fallen off a cliff. A lot of stuff has still kept going, but the stocks are trading kind of terribly. So funny how those things change. But anyway, I guess that's part of the discussion today. But we want to mostly talk about the businesses, not the stocks.

MBI:

No, you're absolutely right. It's incredible how things can change within just 12 months and you have to live through it to be able to appreciate how much things can change within 12 months. There are probably a couple of things I want to say about this rapid change. I remember having conversation about some of these companies, even before I studied these companies, I was having some conversation with some of my friends who also invest, and I remember one of my friends told me that both these companies are probably trading at, I don't know, maybe 30, 40, 50 times revenue a year ago. And if you think about it, when something trades at 40, 50, 60, 70, whatever revenue multiple that these companies are trading at, the kind of shareholder base that you have is just, it's very different. Your shareholder base mostly consists of very valuation insensitive investors. So one is basically, let's say obviously they are passive investors who are legally required to be valuation insensitive.

That is what they're trying to do. And similarly, investors who care about valuation, they probably get nervous even when something starts trading at 20 times, 25 times. So by the time it crossed that 20, 25 times from revenue multiple, anyone who cares about valuation are probably already priced out. They are not there. So anyone who has left as shareholders, either they're very low cost basis, so they don't want to pay taxes or passive investors or just people who are momentum traders or people who are just think about this business can be so much bigger in 10 years, 15 years. So if you think about, and if you notice the kind of characteristics of all these types of shareholders that you may have when you are trading at 40, 50 times revenue multiple, it does show that they are perhaps valuation insensitive. So it probably at some point it stopped mattering whether these companies are trading 40 times or 140 times.

It's the same thing, you couldn't probably make the math work at 50 times revenue. We also couldn't make the math work at 70 times more revenue multiple. So there obviously there's a huge difference between 50 times and 70 times, but the kind of investors who are on that level, they couldn't care less, whether it's 50, 70 or 90. So that's one. And I think now that the stocks are kind of down at 70, 80%, now it has definitely attracted tension from those kind of investors who care about valuation and now they're kind of prodding along and they are kind of arguing about term economies, what it can happen, what may be eventually the margin structure look like.

So all those things that I think many of the investors currently are talking about, the reason they are not talking about that is probably they just weren't there at all in these companies. Obviously, we are all kind of drawing calculators of the market, who really knows who those investors are, what the current base of investors are. A lot of this is just driven by sentiments. When 10 year yields goes from close to 0% to 4% within 12 to or 24 months, that's a massive change. And a lot can change within that timeframe. When rates go from such low level to such... 4% itself is not a high number, but to go from 0.5% to 4% in 12 to 18 months, that's a massive change.

Liberty:

It's not always the absolute number, it's the rate of change. And the big fall that we saw in these stocks was kind of set up by what you said about if most of your shareholders are super insensitive to this stuff, what else do these shareholders own? And if all that stuff is going down at the same time elsewhere, they're probably going to be forced to sell there anyway. If people were full on crypto and AMC and all these types of things, well, it's not like losses on the SaaS side was segregated and the rest of the portfolio was doing great at the time. And if people are so aggressive in that manner, are they also aggressive in other aspects of their investing? Are they on margin? You hear about people that were buying crypto on credit cards. That's not diamond hands that can withstand big volatility.

MBI:

You can have that diamond hands, but the world will not let you to have diamond hands if you are buying crypto or anything that's very volatile with your credit card, the world will not let you have diamond hands. It's not possible.

Liberty:

The bank is going to come and repossess your diamond hands.

MBI:

The one thing I... Sorry, I would like to mention here, I only applaud the fact that even we are having this podcast, even having this episode, I remember 12 months ago I saw a lot of people who claimed to have deep interest in internet infrastructure and what's going on in the cloud infrastructure and 12 months in later, I feel like a lot of that was just basically the stock price was going up and people are not as interested in this kind of geeky, nerdy topics. Like what's going on our internet infrastructure, our cloud infrastructure level at security level. And the fact that we are still having this conversation you are interested in having this conversation kind of shows me that you were always deeply interested in this kind of nerdy, geeky stuff. These are one of my big takeaways from studying these companies. And I would also include CrowdStrike, which I covered in I think mid 2021.

These are important companies, these are companies worth studying and following. These are also interesting companies, regardless of what happens to their stock price, they may not be good businesses or they may be terrific businesses. That's kind of TBD. But I thought that was interesting. I thought all these businesses are very interesting. I didn't know a lot about these companies or a lot of how internet infrastructure or cloud infrastructure is evolving or has evolved over the course of the last five, 10 years before starting this company. So I really enjoyed studying these companies and hey, like I said, I applaud your continued interest, continued eagerness to learn about this. So I think you were one of the people who persuaded me, not directly, but indirectly. As I was reading your newsletter, I was following some of these companies like that you wrote about, especially CloudFlare. And I was like, "I probably want to understand what this business is."

I see you writing about it. I don't quite understand all the stuff that you write about on CloudFlare before starting this. And maybe even still, I don't quite understand everything that goes on in CloudFlare. But yeah, after starting my conclusion was, is they're very interesting companies and worth spending time on. Even if I don't buy the stock or whatever, is this stuff worth learning? And it's much more enjoyable, just focus on learning. I'm pretty sure making money is also super important for anyone including me. I'm not avoiding that responsibility and just trying to use learning as an excuse to lose money. But over the course of in a long period, I feel like both are important. You have to enjoy learning and you also have to make sure you are making money if you want to survive in this world of [inaudible 00:09:43].

Liberty:

Yeah, no, you make a great point. It's like in crypto when the prices were going up, everybody was so interested in cryptography and there's private key here and the public key there, and then when it starts going down suddenly people don't care about that. So same for internet infrastructure, you find out who the real nerds were. And in my case, I got interested in these companies because of the tech angle and only looked at them as stocks afterwards. So it's when I write about stuff, many of my biggest investments are boring. I never write about them even though they may be my biggest positions because nothing is going on. And these types of companies I write about all the time, not because they're investments or they're big investment or whatever, but because they're so interesting, there's always stuff going on. CloudFlare has development velocity that is hard to match by anyone, including much bigger companies with multiples of the number of employees.

So it's ridiculous how much stuff they come up with in a year. It's almost impossible to keep up with. We're not even going to try to cover everything today. So to me there's a big difference between what's interesting and then what you invest in. But I find that from my wiring, from my personality, I much prefer to start with what's interesting first, because that gives me the fuel to learn about it deeply enough that if it ever becomes a good opportunity, I'm going to be able to recognize it and then have the ability to hold on for the long term, even if it's volatile. Because if you buy something quickly just because the price was going up, you have no confidence in it and you have no understanding. So it may be a great five year investment or 10 year investment, but if every time there's a bump in the road you sell it, you're never going to get to that five or 10 year.

Not saying it's any of these companies are going to be great five or 10 year investments, I don't know that. But regardless there's still a word talking about. So that's what we're going to try to do today. I want to start with a quick overview of CloudFlare because I think they may sound like something, oh everybody knows them by now, but I think most people have kind of no idea what they're doing. Their reverse proxy and their infrastructure is touching something like at this point, I haven't seen the latest number, but probably like 25% of internet websites. That's huge. The internet is really big, but most people have kind of no idea how that stuff works. So the kind of short overview I would give is that CloudFlare started as a content delivery network. That's a CDN. That stuff is like if you have a server with a bunch of files on it and all of a sudden lots of users are trying to access the same files at the same time, it could overwhelm the server.

You could have terrible performance or even not be reachable by your users. So what you do is you put someone in front of your server, someone like CloudFlare or Akamai or some of these other companies that do the same kind of thing, the big hyperscalers also do it like Amazon and you put someone in front of your server and they get the file from your server and they cage them into their own infrastructure, which is much bigger than your server. So when there's a flood of users, or something goes viral or whatever, there's a huge spike on Christmas day or whatever it is, your server survives. So that's what they were doing for a long time and as they grew they had enough infrastructure to end those very massive traffic. So they went into DDoS protection. DDoS is distributed denial of service attack. So that's a way that nefarious actors can try to take your site down.

So they take control over thousands or millions of computers via viruses or hacking or whatever, and then they flood your server with traffic coming from all over the world and you can't block it all. You can't find the source. It's not one source that you can block, it's distributed. So they used to take down website, it could be a bunch of script kitties doing that just for fun. It could be blackmail, some Russian mafia type of black hackers would tell you, "You pay us this much or we're going to take your server down on the Valentine's Day when you make half of your sales." Or something like that. So CloudFlare, what they would do is similar to CDN, they would put their servers in front of you and they would kind of act like the big bodyguard with a Kevlar vest and they would jump in front of the bullet and catch the bullet for you. So they would catch the attack for you and you'd survive. And because they had this huge distribution network, they could better find the sources of traffic and block it more easily.

MBI:

Yeah, I think one particular thing that kind of stood out to me when I was studying CloudFlare, how they kind of approached this problem, how come this small startup out of nowhere were able to build such a distributed large network all over the world? So I think that I found that unique. So they partnered with all the ISPs all over the world, more than a hundred... In a hundred cities around the world to have co-location facilities on this ISP. And it was kind of a weird relationship for both the ISPs and CloudFlare. So ISPs, by allowing CloudFlare to have their servers on their facilities, it could lower their bandwidth costs and CloudFlare could also help them from DDoS attacks and all that. So once CloudFlare had their servers on those ISPs and facilities, they kind of have their CDN business at the core product.

And on top of that, because these servers are kind of programmable, they could build a lot of the stuff on top of it. So it's just in the same base layer of infrastructure and they're just building on top of it at a rapid pace as they were saying that it is so hard to keep up what exactly they are building on their platform. So they had their DDoS attacks, security products and a lot of other applications. And now they're also talking about wireless platform, which I'm pretty sure you'll touch on what people look at CloudFlare today. I think very few people are very super excited about the CDN business.

They're much more excited about all the stuff that CloudFlare can build and probably will build and has been building on top of that base layer of platform that they have on this different iSpace around the world on a distributed basis. And I thought that was very unique. I mentioned this kind of anecdote on my deep dive like Matthew Prince and Michelle Zatlyn, the two of the three co-founders other being Lee Holloway. So I know Michelle and Matthew Prince were meeting one of the advisors, one of their kind of advisors.

And when they kind of explained what they're trying to do to solve the problems in the infrastructure level, the advisor was like, "You're trying to build a CDN." And they didn't know what CDN is, so they kind of were nervous about it and they kind of nodded yes, very nervously said yes. And then after the meeting was over Michelle and Matthew kind of looked at each other, "Hey, what is CDN?" And they had no idea. And Michelle is from Canada and she was like, "Maybe it has something to do with Canada." So the fact that these people didn't even know what CDN was, I think in retrospect it was probably helpful to approach this problem from a very unique vantage point.

Liberty:

First principle.

MBI:

First principle's perspective rather than, oh we already have this, let's just kind of make it incrementally better. No, because they didn't know any of that stuff in the initial years. They kind of approach it in a very, like you said, first principle basis and approached problem in a very different way. In a way only probably in know startups could come up and disrupt the incumbents. If, they were trying to do something what Akamai was already doing I don't think we would be talking about CloudFlare today.

Liberty:

Yeah, for sure. If you just copy what someone else is doing in CloudFlare's case would've a very different infrastructure and it wouldn't be flexible in the way that allowed them to level up with all the other stuff. I think you said it very well, the CDN stuff was the Trojan horse that got them in the door because from an ISPs perspective, if every commercial customer, every just private citizen connected to you is going all along and surfing the web and every time they're going somewhere that you have to connect to the outside internet and go somewhere and go back, that's all bandwidth cost. Connection costs. If you have CloudFlare that is caching like thousands and thousands and thousands of website inside of your co-location, the traffic just doesn't even go out. It's free basically. That's what Netflix and Google, they all put a bunch of racks inside of ISPs and when you're watching Netflix, you're not going out on the open internet on the server to try to watch your movie.

It's all inside of your ISP. And CloudFlare kind of did that, but instead of just for one company's stuff for Netflix movies, well they're doing it for everybody that they're doing caching for. But once you're inside, you have very low latency. You're inside literally of the ISP, you have very low bandwidth costs. And then the way CloudFlare designed its network is they use commodity hardware everywhere. They didn't build custom special server for CDNs that were super specialized and could only do this kind of storage bandwidth. They built servers that had a bunch of other resources that then over time they use for other things. On calls, they often talk about how when they decide to build a new product, they will often look at what's our idle capacity right now? What do we have an extra, or even if it's not literal capacity because you don't overbuy that much, but it's kind of latent capacity?

Well we could easily add that part without rebuilding the whole thing. And so for example, first the CDN stuff was using a lot of bandwidth but outflows and storage. But then the security and zero trust stuff that they built later was more compute intensive. With zero trust every connection has to be encrypted separately and you have to authenticate people all the time. The zero trust basically is a security model that assumes that everybody is denied everything all the time until they can prove that they have access to something. The old model was kind of like the castle and moat thing where you have a firewall and a VPN and people outside can't access stuff and then you pass the gate with your credentials. But once you're inside of the castle you can do basically anything. And that model doesn't work anymore. People are distributed, people bring laptops home and hackers are much better at getting inside.

And so they can get inside, they move laterally and they can screw up your whole business. So now the new model is basically, okay, everything's going to be encrypted, everybody's going to need authentication for every single thing they do, every server, every service, every file, every app. CloudFlare got into that and that's kind of like the way they describe it is that they're stacking S-curves. So the CDN stuff in the US was the first S-curve, but that S-curve is kind of plateauing out. It's pretty mature. On top of that the second thing is the security stuff and that one is still kind of early and that one is bigger than the CDN stuff. It's a bigger market, it's a bigger TAM, if we can still use the word TAM in 2022, maybe TAM has been banned or canceled, I don't know. So that's the second one, but they are already working on the next one.

The third one is the worker's stuff. So they were looking at their... Now they have 275 locations around the world full of servers. And those servers, they still have a bunch of resources and so they figure out that rather than be kind of dumb servers just serving HTTP requests as a CDN, well what if you could do a bunch of compute under server, basically an edge platform instead of having to go back to the central server to do the logic of an application or a service. If you could do that logic on the edge, it will be faster for some stuff that's very latency dependent. But also, and that's kind of the big use case that they discovered in this world of regulations. And every country in the EU has different data laws and privacy laws and that stuff keeps increasing. Well if you can process the data locally and follow the local laws and regulations and all that, that's a huge value add.

But the big hyperscalers, their whole thing is scale. They try to be centralized, they try to have a big DC for the whole region and they're not kind of fully set up. They can do some of it, but they're not fully set up to have 35 different jurisdiction. And now maybe is every Canadian province going to do something different? Every US state. And so CloudFlare with the third S-curve that they're trying to do is the workers platform. And in theory that's even bigger than the security stuff because they keep adding stuff to it. They have a database with D1, they have object storage and they have all kinds of stuff where over time it's kind of becoming not fully equivalent but reproducing a lot of the big central clouds. But at the edge. They even have a partnership with NVDIA. They put a bunch of GPUs in there, that's the third one.

And the fourth one, which is on the horizon, they've only talked about it a bit and they started doing it. It's kind of becoming more of a global telco. If you look at what the IT span of companies like 50%, is telco stuff, just connectivity and all kind of products there. So they have CloudFlare for offices where they basically bring a connection directly to your office and instead of connecting to an ISP, you connect directly into CloudFlare. So they keep adding stuff on top of each other to build that meta S-curve or something like that.

MBI:

Yeah, no, I think you sum it up so well. No, I think they're definitely trying to stack one S-curve on another. That's what literally they're aiming for. I think two things that I want to highlight is region specific regulation or the different laws in different jurisdictions that you kind of hinted at. So I remember reading interviews of AWS executive, Google Cloud CEO, Thomas Kurian's interview and Microsoft President Brad Smith and Matthew Prince's interview. All of these people are interviewed by Ben Thompson over the course of last 12 or 18 months. And Ben Thompson later highlighted this point. He kind of asked a similar type of question on, "Hey, can you follow different regulations in different jurisdictions if they make you do?" And everyone except CloudFlare was like, " Yeah, I know that's hard. It's going to be complicated. It's super hard. I don't know whether we'll be able to match these different regulations for different jurisdictions. It's going to be super complicated if they make us do that."

And CloudFlare's like, "No, we can do that. And I think that's where the internet is heading." And it does seem there's definitely a grain of truth to that. China is already a separate sort of internet infrastructure that they have. And more and more it's possible that we may not have a global source of value culture. It may have very different values in many different countries and regions. And if that happens then yes, CloudFlare seems to be in a better spot to go along with that sort of changed reality in the internet infrastructure level. So obviously that's something that won't show up in the next quarter's revenue or next year's revenue. It's definitely something that caught my attention, it's definitely possible that CloudFlare may be potentially an unintended beneficiary of such regulations and laws by different jurisdictions.

And another thing is you mentioned the workers platform. The companies who are already in cloud are like [inaudible 00:24:18], bigger enterprises who are thinking about shifting to cloud. It's unlikely that they will start with workers' platform because it's obviously not fully fledged and now they don't have everything that AWS or Azure has. Those are much more developed platforms. Lot more developers are working on it. Lot of innovation is happening in those public cloud level. But if more of new and up and coming startups, they are the ones probably who may think about building on top of workers' platform. And I think Matthew Prince mentioned this, they studied a lot of different developer platforms and it usually takes eight to 10 years to catalyze innovation to kind of spin the wheel faster.

Liberty:

And get the tools mature too.

MBI:

Yes. And I think it's been what? Five years since they launched. So yeah, we probably have much better clarity probably three to five years down the line. What exactly is the potential for this platform? I think, if I remember correctly, 15% of their customers are building on top of workers' platform. It's probably a bit dated because I looked at CloudFlare a few months ago, so probably, I don't know what the latest number is by latest quarter, maybe it's a bit higher. So, it's a 15, 20% of their customers are building on top of workers' platform. And if they can get this number to 50, 60, 70% and as they kind of platform... The number of customers gets bigger and bigger, that can be something quite, I feel like those are the kind of stuff, what is the potential for workers platform, how big it can be. Those seem to be harder to model.

Liberty:

They kind of don't even try, they often talk about their time and they're like, "We don't even try to give you a number for workers." And then they say something, "Because the numbers get so ridiculous that you wouldn't believe us anyway." Or something like that. But I kind of like the long-term thinking on workers when they talk about it's always like, "We don't try to squeeze the lemon right now. We don't try to get money from it right now because all that matters is adoption from developers."

And so they announce a fund with VCs. They were trying to raise a one billion fund from VCs to finance startups building on workers, which is kind of great. Because they're leveraging other people's money to help build up their own platform. So if you can do it, that's a good position to be in. But then the fund was oversubscribed. It was like one point 25 billion. Okay, great. And then a month later was like, oh yeah, the fund is now two billion with 40 VCs. It's like, well if CloudFlare can have access to two billion of VC money to spend on the workers' platform to help train engineers to know all of the tools. Because that's the thing with engineers, well, if they have a choice, sometimes they don't. But if they have a choice and you give them a tool and they learn it and they like it...

Liberty:

Sometimes they don't but if they have a choice and you give them a tool and they learn it and they like it, they could stick to it for a long time. They move company and they try to use their favorite tools at the new company. And that's kind of like how AWS started at first. It was more of a startup thing and over time a bunch of these startups became the huge companies of today. And I think that's kind of like Cloudflare trying to see this platform with that kind of energy. Right? Over time we're trying to become respectable and track the big companies, but also we want to plant a bunch of seeds, let a thousand flowers bloom in there and see what happens and maybe the next Uber or Airbnb or something is going to build on workers.

MBI:

What do you think is the likelihood that some of these more let's say successful startups who end up becoming a lot bigger over time may graduate to let's say more established public cloud? If it's easier to move from one of the things that Matthew Prince talks about, like AWS has this egress fees so hard to take the data out and if Cloudflare makes it easy and keeps it as easy as it's to move in and out, can't some of these bigger customers just graduate to let's say AWS or Azure? Because these public cloud companies, they're obviously investing hand over fist on this platform. It's hard for me to believe that companies like Cloudflare will be able to match the level of innovation that we will see on the public cloud level, like the top three that we have, AWS, Azure, and GCP. Do you think that that's a potential road block for Cloudflare workers platform to be successful?

And before you answer, I want to make sure that anyone who's listening to this podcast knows that before this recording, I actually explicitly mentioned that I want to have this conversation, so it's not going to be like I'll do most of the talking because it's a conversation. We don't know who is going to do more of the talking. It could very well be Liberty doing 60% of the talking. I'm here to talk about this companies because I'm actually interested in hearing also Liberty's view, so I'm not just expressing my opinions and views, which I can do from time to time, but I'm also very much interested in listening to what Liberty thinks.

Liberty:

Thank you for mentioning that. I'll mention it in the show notes too, just so people know going in, it's a conversation about these companies. It's not an interview of NBI about these companies. It's a really good question. I think that's a huge challenge for them. I think if all else is equal, it's harder for Cloudflare because it's not as mature. So if all else is equal, I think a lot of companies building on workers will do it because not all else is equal. They're going to take advantage of some of workers' specific advantages. So for example, the way it's much more decentralized and distributed at the edge, maybe some applications need that to perform best. So moving back to a more centralized cloud, maybe kind of a step back, maybe it's going to be the regulation stuff we were talking about where you need kind of the logic that's present in its jurisdiction to do the app or the service properly.

Cloudflare also seems to have a very, very efficient network, partly probably because it's collocated in tight ISPs and all that. So that's why they have free egress on R-2, their storage that's S3 compatible. They're trying to compete with Amazon. The dynamic they're going for is like, okay, we're offering you something similar to S3 but without egress costs, right? Because a bunch of people have huge egress costs on AWS, and the way Matthew Print said, he's like either we win some customers because we are offering something cheaper or we push Amazon to lower its egress costs, which will make our network more valuable.

If the egress costs are lower, all less equal, you're probably going to have more egress, right? You're going to distributor your apps more, maybe part is inside AWS, part is inside Cloudflare. Maybe a part is on-premise, but you use the Cloudflare network to connect everything because one thing they often say at Cloudflare is that the network is the product basically. So they're not trying to be so much a competition to the centralized DCs that the hyperscalers have. They're trying to be a glue that ties everything together, like the open internet and the hyperscalers and every office with its own secured private network. And also the glue becomes more valuable if you push the centralized player to not hang onto their data quite as much because that's what egress fees are. Like every time, you send something out of the AWS via S3, you pay something. So it's encouraging you to keep everything inside, right?

MBI:

Yeah. But what is the public cloud's incentives to play along in Cloudflare's terms? Switching costs or lock in I think is one of the key reasons why investors love this in public cloud companies. So it is very convenient and easy to move from one cloud to another. It will definitely diminish the switching cost and it can become a pricing war down to the bottom. So I'm pretty sure this is not a very unique take, public clouds probably know this and that's why they probably focused on building their platforms in a way they have built. So what do you think the incentives are for AWS to play along with let's say Cloudflare's terms?

Liberty:

So far they seem to have taken the middle road where they reduce their egress fees but not fully because AWS is so big that they're looking at Cloudflare. It's like, oh, it's a billion dollar total revenue company. But if you desegregate the parts, the part that they're fighting for is much smaller. AWS is not going to lose 2 billion to win a hundred million there. So in a way, that's kind of what Cloudflare is probably hoping for a long time. They're going to be able to build stuff from the bottom, trying to disrupt things by building cheaper products, by using all the success capacity that they have on their network that was built for something else. And by the time they become big enough to be noticeable by the huge players, they're already big enough. They already have the scale to be a real competitor.

I think that's what they're hoping to do, like classic low end disruption. That's also why Cloudflare has a huge free tier. They have probably around 5 million free users. That's insane. No, nobody else is like, we're giving away really good versions of our services. The free tier is not just a trial. It's not like, oh, try it for 10 minutes and then you have to pay us. I'm on the free tier of Cloudflare for some stuff. I'm on War Plus and a bunch of stuff. And the thing is, it's very synergistic. Can I even say that word? Is it TAM as a bit banned? But anyway, I feel like the free tier of Cloudflare is brilliant because the way the R&D works at Cloudflare is they create a bunch of products that are the minimum viable product and then they put it all to the free tier.

And free people are just glad to have something, right? If you were putting the minimum viable product to a paid enterprise customer that's paying you a hundred thousand dollars a month or something, things may not be great. But if you use your free tier as a bunch of better testers, you're going to get much, much more feedback that you could get with a QA team internally. You're going to be able to iterate on the product much faster, make it better much faster. And then once it's great after a shorter period than otherwise, then you can put it out to the paid tiers and the paid customers are just happier with more solid product. I think that's how Cloudflare could do what it did so quickly. Two years ago, they basically had almost no security products. All the zero trust of didn't exist. And now they have built a Z-scaler inside of Cloudflare within a couple years they're already selling a bunch of stuff to the US federal government and they're not even FedRAMP approved yet.

They're really, really fast at that stuff. And another reason why they're so fast, I think, is that they're dogfooding their stuff. So that's an expression that not everybody may be familiar with, but eating your own dog food is the expression you often use in tech where it's like if you use your own products, you're going to be much better at figuring out what's wrong with them or how they could be improved, right? Once in a while, I'm sure we've all bought something and it really sucks and you're like, did the person that made it even try it? You try for five seconds and you figure out all the problems. Well, Cloudflare almost all of their products now, the Zero Trust stuff is built on the workers platform. So they're building their own products on their own platform that helps them build the products faster, but it also helps build the platform faster. Workers are improving faster because it's used in the house.

The engineer that's working on the Zero Trust thinking can turn around and tell the workers, engineers, hey, this thing you could make it better this way or that way. And so this kind of tight feedback loop like John Boyd would say, the OODA loop of Cloudflare is very, very tight and very quick. And part of that is building on your own platform. Part of that is the free tier. Part of that is that the CDN stuff, the network is also kind of a sensor network. It's not only sending traffics and bits and pipes, they can also monitor the whole internet in a way that few companies can route around damage and traffics and clog parts and they can see from where the attacks are coming so that the security staff can be trained right there. Machine learning models and everything a bit CrowdStrike can be trained with a bunch of the data that they get from having millions and millions of users on their network connected at all time.

So I feel like that's part of what makes Cloudflare so interesting. It's like a super organism. As a company, if you look at the whole network, it's a very, very complex things, but lots of interlocking parts that are very well designed to go together. Is flywheel another banned word? I don't know. But everything improves everything else. And this elegance that I like is probably why I write so much about the company. The more I learn about it, the more I'm like, that's clever. It's like for another quick example, the commodity hardware that they use everywhere, every server that Cloudflare has can run every Cloudflare product. So they can dynamically adjust things and they're like, okay, we need more worker staff right now. So we can take a bunch of servers that used to do CDN, but now CDN has less demand in this area.

They cross compile their code. Maybe not all of it, but a lot of it. So it can run on arm and on X86, so they can buy the cheapest servers from either ISA. They're very, very flexible and the free tier for most companies that didn't design stuff from the ground up to be this flexible, if you were giving away your products to millions of people, it would be hugely expensive. But Cloudflare's gross margins are great. Their cap expending is not crazy. Partly because the free tier is, they give it to the idle capacity of the rest of the network.

So the paid user get priority. So if you're paid user, you're never going to be slowed down by the free users and they send the free users to South Korea's asleep right now, let's send free reserves to the DC somewhere in South Korea or yeah, let's use idle capacity in this location that, I don't know, for some reason right now, some idle capacity so they can move stuff around the network. So imagine the code that it's the old telephone system switches, you have to route all of the things all around. They're routing stuff between data centers and between products all the time in real time while taking account of traffic on the internet and congestion and this fiber optic cable was cut. So that's really a big brain overlaid over the internet.

MBI:

Absolutely. And I think one thing that probably should be highlighted even more, the fact that they chose this off the shelf, commoditized hardware is definitely a key factor for their subsequent success or subsequent products that they had built over on top of their products. Like you said, they can just move to stuff from one geographic to another, one region to another based on where the ideal capacity is. It'll be a lot harder if it were more custom made and differentiated servers and different platforms and it would probably be not as efficient and not as convenient to move traffic from one region to another or one server to another. But the fact that they have chosen to build based on an off the shelf commoditized hardware, made it possible of all the stuff they're doing right now.

Liberty:

And it does show long-term thinking, right? Because at the time they made that decision,

MBI:

They didn't know all these details, right?

Liberty:

Yeah. But when they were just a CDN for example, or just doing EDOS protection, maybe they could have tweaked their DCs to be even more efficient at doing CDN stuff, but they would've precluded some other paths that they took later for, like an hour or two in workers. And so this long-term thinking that we're hearing now for workers, I'm sure it was present five years ago, it was present 10 years ago. It seems to be part of the DNA of the company, which is a good sign for the future. It's not just something that they stumbled into and were like oh, what an accident, we have this stuff that we can use for something else. It was, as you said, they probably didn't know what exactly they would do with it, but they knew that they had the capacity to invent something. They often talk about how they do these innovation weeks and these release weeks eight times a year or something like that, where in a week they release 36 features and they write long blog posts about everything, all it does and tons of people are like, that's a bit overkill.

What kind of marketing is that? But when you listen to the founders talk about it, they're like the target audience is engineers, who want to work for Cloudflare and for the best engineers in the world, they can go work almost anywhere. So we want to show them that we're working on interesting problems that affect millions and millions of people. It's the foundation on the internet, where everything else is built on top of us basically. And they also want to show them that we can ship stuff. We're not married in bureaucracy and you're going to be in committees and meetings forever before you ship a tiny feature. It's like, no, if you come work for Cloudflare, look at us, we're shipping stuff every other day. So I think as marketing, that works pretty well. I remember almost two years ago in call Matthew Prince said that I think in 2020 they had 200,000 applications for engineers and salespeople to work at Cloudflare.

MBI:

200,000?

Liberty:

They probably can have a good pick of some of the top engineers in the world in those fields, right? I'm sure AWS and Azure and Google Cloud, all those are doing great too. But because Cloudflare is much smaller, but they have to compete with the big guys, they need something differentiated. They need to be able to attract people and they're not all going to be able to pay them necessarily quite as much. You wrote about this recently about all these big tech monopolies are making it more expensive for everybody else to compete with them by giving so much comp and all that. So Cloudflare has to figure out a way out of this problem.

MBI:

Yeah, I personally have owned some of the big tech companies for almost last four years, and I feel like I still haven't appreciated how big of a competitive advantage they have in recruiting until recently. When I sat down and looked at their operating expenses per headcount and gross profit per headcount, it's crazy. You know basically that shows you, it kind of cost 270,000 to $300,000 operating expenses per headcount for almost all tech companies in the valley because good talent is expensive. Talent in general in the valley is expensive, but not all companies can have 500,000 a million, or even for Meta's case it was 1.4 million gross profit per headcount. So to me it become obvious that big tech and the rest of the tech companies are operating in a very uneven field. Big tech companies basically set the price for the talent or the raw materials to build these great companies, but only a handful of them have those incredible gross profit per headcount.

So I have noticed a lot of investors, analysts are kind of angry like these managements in different tech companies are stealing money from us. But my question to them, how exactly are you planning to hire engineers and salespeople when a big tech company is basically offering them 400,000, $500,000? That's total compensation. How are you going to convince anyone to work for Okta, Cloudflare, DataDog, all these up and coming tech companies for $150,000? Because even if you recruit them today, if they know their friends and their peers are getting paid like 300,000, $400,000 from Meta, from Google, from Apple, Amazon, AWS, right? What they will do, they will probably work for these companies for a couple of years or so, and then they'll just apply to these big tech companies and then move. I'm not saying everyone will do that, but enough of them will do that to make sure that the management of these companies like DataDog and Cloudflare of the world will start thinking, oh, you know what? We have to basically pay them to keep this talent pool.

And I think there are a couple of things that happened. Let's say last five years or so. If I were, let's say Mathew Prince or Olivier Pommel trying to hire someone, my pitch would be, Hey look, we are up and coming. We can't pay you as much as Google or Meta. We can't do that. What you can do is I can promise you a better work environment, less bureaucratic environment. You can build things, you can see things, you don't have to go through jumps and hoops of five different committees to make sure that this fits our chips. And actually many of the big tech companies don't have probably alternative because they are scaled platform, they have billions of people on their platform. So any even minor mistake is amplified and gets a lot of attention. If you are making a minor mistake in let's say in some of the smaller tech companies, you may get some hit, but it's not going to be a Wall Street Journal front page news.

So yeah, that's number one. Hey, you get a better work environment, less bureaucratic work environment and we can pay you our stock market. Google's market cap is probably now going to from 2 trillion to 4 trillion or 1 trillion to 3 trillion. You can't really compound our capital 3, 4, 5, 10 times. But we are just five, 10 billion, 20 billion market cap company and if we really build a great company, we can make it a hundred billion dollar company, $200 billion company in five years, 10 years. And you can grow your wealth along with the companies that we are building. And I think for a long time that seemed quite a credible pitch. I was just imagining if I were in this kind of employees shoes, I would think that's credible. Because if you think about it, there are a lot of companies, Shopify, Twilio, Cloudflare, DataDog, I think Shopify and Twilio specifically. I actually remember the number on Twilio because I kind of mentioned this in one of my deep dives.

If you joined in sometime in 2017 at Twilio, the start basically became 20x from sometimes in 2017 to top of 2021. So I'm not saying everyone got 20x, but I'm pretty sure most of the people who joined Twilio got 5X, 10X, 15X over the last 3, 4, 5 years if we are taking 2022 out. So it was a very credible pitch. People and employees could actually believe that pitch because it's credible, it's not something made up. They could see that their peers are actually making more money by working at this company because the stock is going up a lot faster than these big tech companies. But big companies did fine, but it's not nothing like 5X, 10X, 20X over the course of three, four years. And it is also true, these companies are less bureaucratic.

So this was a credible pitch. Now I guess my concern is after looking at these numbers in more details, I think at one hand these companies are getting bigger. So these are not a hundred million, 200 million revenue companies. These are a billion dollar, 2 billion revenue companies. So over time they may become slightly more bureaucratic, but definitely not as bureaucratic as let's say big tech companies. But they'll be slightly more bureaucratic. And there's always this new and up and coming other startups growing at hundred million, two hundred million. So if you are playing with that, you know can probably join a startup, a private company. Although they're lots of other problems, let's not get into that. But then the other thing about the stock price, that is also a big question mark now whether these companies can be, theoretically yes, they are probably in this, many of them are probably around 2018 even some of them are probably 2017, 2019 around this level in terms of market cap.

But there is a lot of question mark because of the rates, and a lot of investors are now asking about margin structure and all that. So, I think one of the things that I have been thinking probably a lot more deeply than I used to is how exactly a company can be able to recruit talent in a more efficient way. I'm know Aian has an advantage because most of their employees are in Netherlands. Even Shopify has a lot of their workforce in Canada, which is clearly not as expensive talent market as the Silicon Valley. But the thing that really concerns me that many of these up and coming Silicon Valley tech companies will have a really hard time to lower their compensation because if I were Zuckerberg or Sundar Pichai, I would actually maintain that. I would actually maintain the high cost for this.

What I would do is basically if I were Zuckerberg, I would basically try to make sure how to improve the productivity of my workforce. And if I were Zuckerberg or Sundar Pichai, I would think this is an excellent opportunity to hire the top talents from anywhere from other parts of Silicon Valley. I don't have to necessarily increase my workforce. I can just show the door for let's say below performers or under performers and create some space. I think many of these companies are probably doing that meta just laid off 10, 11,000. They may do that gradually over time more, but it also creates space to hire back a lot of top talents that are out there all across many Silicon Valley companies. That's something I'll be paying more attention to than I did. But I think I should mention because these things can be influenced by what you own.

I do own Meta, Google and Amazon shares, so it's possible I'm just over extrapolating some of my existing biases. But I do think it's definitely a concern if I were in the talent market. There is probably a bit of a lol moment here in the sense I think these companies, obviously big tech is not hiring right now. They're kind of in a retreat mode, but I doubt they'll be in the retreat mode for too long. So they'll probably come back to the talent market maybe after a year or so. So now this up and coming silicon tech companies, like smaller companies, they would probably still be able to hire, well let's say next six to 12 months. But beyond that, I think the big table will probably come back at some point when the economy sort of stabilizes. So the question about total calm, a lot of the time I myself used to think, these companies are investing aggressively and over time as they scale, they'll be much more profitable.

I'm questioning that assumption deeply. And it's not the stock price, I don't think it's the stock price going downwards that's influencing me. I have every incentive to buy these companies if they are lucrative and compelling. But these are the kind of questions that really give me pause. If you want to believe the profitability of many of these tech companies, you have to believe two things. That the total compensation for the engineers will be more or less similar or somewhat downwards. It can't go up like 10, 20% from here, let's say from now to next five to seven years. That's one. And the number two is you have to believe this new up-and-coming companies will be more efficient than their predecessor software companies. See, if you look at most of the software companies and see how many employees they had when they reached this a 10 billion revenue or 10 billion gross profit.

And if you put that number and multiply that with the average compensation that the engineers or salespeople get today, and then you see that there's not much profitably left even after five, seven years. So you have to assume that these companies will be much more efficient in terms of utilizing their workforce. Maybe they will be, but that's what something basically, we'll have to assume that so far I personally haven't seen much of an evidence. So for example, one thing that I did, let's go back when they were as small as these companies are now. So let's say a billion dollar of gross profit. And if you go back in those times it kind of looks similar. It looks they are scaling in the same way when they are smaller. But we don't know the trajectory from 1 billion to 10 billion can be vastly different for some of these companies.

The companies that we are discussing, maybe I'm a bit biased in the sense that because I have a slightly positive perception about both of these companies because I studied them and I'm not making broad stroke judgment on these three companies. For example, Cloudflare, if Cloudflare can pull off workers, then yes, their journey from a billion dollar to 10 billion would be probably vastly different from let's say Salesforce or ServiceNow. ServiceNow is probably four or 5 billion gross profit. So yeah, so Cloudflare journey can be different. That's what basically you have to assume or believe, and I'm not sure how comfortable with this assumption, but then those are definitely in the realm of the possibility that both those assumptions can be true. It's possible that compensation for engineers and sales people is just too high and they'll go through some sort of rationalization, which like I said, I'm not sure because big tech probably should maintain those level of salaries. It is a competitive advantage for them. And the second is their trajectory from here to visit 5 billion, 10 billion gross profit. That is possible I guess, but we'll see how that evolves.

Liberty:

Yeah, you bring up so many good points. One of them is even if big tech is retreating somewhat in hiring, it's a relative thing. So if they're retreating by 1X but the smaller players are hurting more and they're retreating by 2X or 3X or going out of business, big tech is still gaining in that. And if they can come back quicker, well, I'm going to make an analogy in the same way that Google and Apple taking a big cut on their app stores is preventing a bunch of businesses from ever existing. If you have a business where, giving away 30% or 15%, you just can't make it, right?

It's not a high profit business, you just won't exist at all. So you end up with a bunch of stuff where selling coins on Candy Crush or whatever, like pure profit, but there's a whole cohort of businesses that may have existed without that cup. I feel like big tech keeping salaries very high for engineers and designers and salespeople, by keeping those salaries very high, it's kind of doing the same thing, right? It's like a tax on the industry and anything below that that may exist, if you could pay engineers a 100K but can't exist if you have to pay them 300K. Those companies won't even get started.

So the hope with this kind of slowdown and recession and the kind of tech bubble popping and all that stuff, the hope for me was well, maybe we're going to see tons of great new companies because startups will be able to hire now and it's going to free up a bunch of talented engineers that could have been doing more useful things for the world than trying to optimize by 0.00001%, some color on some Google ad page somewhere. Or I hope that's what's going to happen. But maybe the alternate scenario is just that big tech is so strong and profitable that they're going to keep the floor high and it's actually the smaller players that are going to hurt even more. And I don't know, I still believe that the very best startups and the very best ideas get stronger in these types of downturns.

But maybe big tech is the difference with 2002, right, or the other types of tech winters. And another very interesting aspect of this, how human psychology works, is that when it felt best to go work for these tech companies was when it was the worst time, right? In 2001, you were looking back and seeing the stock charts going vertical and you're like, "Oh wow, my stock comp is going to be worth a fortune," right? And that was the most dangerous time. And right now everybody's like, "Stuck SBC and I want to be paid in cash." And now maybe the best time to take some of these jobs if things turn around and in, I don't know, five years the RSUs do well, right? I don't know, maybe not. Maybe we're in a longer difficult period for tech. If you forget about valuations, if you only look at the businesses and the cash coming in, these are still very, very, very good businesses, right?

Super high recurring revenue, super high gross margins, potentially very good unit economics. They're reinvesting a ton in growth. But a lot of these businesses, if they took their foot off the gas with hiring, we would hope that the margins would scale well. We don't see it. And that's the thing, right? At some point, it's all like, "Well, is the theory correct, that they could do it if they wanted to, but they just don't because it creates more value to just keep growing faster?" I guess that's the question that someday we're going to see the answer to.

MBI:

Yeah, so I think obviously, some people mentioned this, if it's so hard to be profitable, how come this PE, private equity guys, just buy these companies and then make it profitable, right? I think yeah, there's definitely something to that. First of all, I personally don't think these are structurally unprofitable businesses, right? And I do think we have to level-set that to make sure that we are not arguing against a straw-man case. These companies are still worth 10, 20, 30 billion dollar, right? So obviously, market is not saying these are structurally unprofitable businesses. These are, the market still believes, and I do think these can be profitable business. Now the question, the crux of the debate I think, is how profitable.

So yes, PE companies can buy these companies, PE funds can buy these companies and turn them into profitable business, right, because if I give you an assignment, if I give Matthew Prince an assignment, "Hey Matthew, you have to go out there and make this a 10% gap EBIT margin business by the next three years," I guarantee you he'll be able to do it.

Liberty:

He could probably do it within six months, right? Just fire a few people and.

MBI:

Because that's the assignment I am giving you, in three years, in five years you have to do this. But when you are building a company with no time horizon in mind, you are not trying to get to a particular margin to flip it to public market again or to flip it to another PE buyer, right? You are with this asset, whether you like it or not, you have to keep building it. So if you are a public company, if you are especially a founder, I think they have a very different approach to in a situation like this. Yes, they can be profitable, but that is not that, right? I think a lot of the times people think they can be profitable. Of course they can be profitable. These are two-billion-dollar businesses. Do you think that a two-billion revenue business, one-billion revenue business can't be profitable if they want to?

Liberty:

Yeah, 80% gross margins, right?

MBI:

Yes, they can. Obviously, they can. But at what cost? If they become profitable, if Cloudflare becomes 20%, let's say, margin profitable in next two years, three years, obviously, they're not probably hiring as much, obviously, they're probably not doing some of the other stuff they would. So definitely there isn't trade-off here between top line and margin, right? It's a balance game. Like most things in life, you have to balance these things and last three to five years was just probably not so balanced. People were literally rewarded to accelerate their top line regardless of profitability, right? So it would probably be a different sort of balancing environment in the next, less than three years or so. So we'll have probably some clarity in terms of what is the extent of the trade off. But yeah, that is a key debate. Here's something that I want to specifically, and it's not about Cloudflare or DataDog, but I think somewhat related.

I was looking at CrowdStrike's, yesterday's earnings, yesterday being 29th of November, right? So they mentioned they have this slide of their long-term operating margin, and this is very common, most software companies have this, and they have long-term operating margin as 20 to 22%, but there's a non-gap number, right? What's their stock-based compensation as percent of revenue today? 23% in nine months so far. So even if you assume this 22, 23% of SBCs stock-based compensation as percent of revenue will go from 22% to, let's say, 12% or 10%, which by the way, never happened. If you look at last 10, 12 years of other companies who scaled from, let's say, a billion-dollar revenue or 2 billion-dollar revenue, 10 billion revenue, what did their stock-based composition did in this period? It didn't go down. It kind of stayed flat or actually went up, right? And I'm not saying it'll go from 22% to 32%, I think that it just doesn't work that way, but to assume that from 22% to 12%, 10%, even if you assume that, that's a 10% margin business.

And the thing is, management is not even attempting to make it. They're under the impression it's a 20% margin business. So if you don't measure it, you can't really aim it, right? So they may feel great, oh, we actually reached our long-term margin target 32%. What's the [inaudible 01:00:25] revenue? 20%. So basically, have a break-even business, right? If that is a long-term target, that is structurally destructive.

Liberty:

I feel like all these long-term target things are kind of meaningless, right? They're just telling a bunch of analysts kind of what they want to hear, but over time, they're all going to change them and it's going to fall somewhere very different, right? If you told Matthew Prince or George Kurtz, like okay, you can't grow faster than 15%, do what you want, but cap it at 15%, right, what would the margins be? They would have great, great margins, right, but the value of growth for these companies where revenue is extremely sticky and high margin, high gross margin, the growth is worth so much that they're not even thinking about trying to optimize for margin. So they give kind of all the same numbers, right? All of these companies have kind of the same target operating margin numbers. You look at Snowflake, they're just telling analysts kind of what they want to hear, but I don't put any stock in those numbers really.

I think they could tweak them, if they pulled a lever one direction or the other, they could get almost any number they want in between those things. The question is how much value is created by each approach, by each kind of set of levers.

MBI:

No, I think that's an interesting thought experiment. If you made Matthew Prince to grow 15%, what would the margin look like? Yeah, like I said, I don't think these are structurally unprofitable, this can be profitable business. My suspicion is obviously, that is subject to change based on how things evolve. These are actually 10% margin business, even terminally speaking. And why is that? Because the cost there, the raw material cost, is just so high, right? There are elephants out there who make sure the cost of these raw materials don't really go down rapidly or precipitously. And so one way, kind of if everyone majored in computer science, probably yes, then supply, you just go high. These things take time. I looked at the numbers, the numbers are not crazy. People are still not graduating with CS measure.

Liberty:

Yeah, maybe AI is going to make engineers much more productive, right? CoPilot. CoPilot is getting better all the time.

MBI:

So that's another thing. I feel like that's one way I can be wrong in my assumption. If AI becomes such a huge force, that makes a lot of the density of employees irrelevant, not irrelevant, less relevant. That's something I think I feel like yes, when there is a necessity, you tend to come up with things that kind of solves your problem. There's so many companies out there, so many companies in Silicon Valley, if they're all 10%, 5% gap, [inaudible 01:03:01] margin business, obviously, it's not super lucrative for the next set of founders to come in and build businesses, right, if the cost for the engineers remains so exceedingly high, right? Yeah, so it is very much possible that AI may be a lot bigger force than many of us seem to think, especially in the context of how many engineers you really need [inaudible 01:03:26] these products.

Liberty:

I think that's the other thing. I think as you are growing super fast, right, if you're growing 50% a year or something, that's real big, right? Organic growth of that level, I can understand why you need a lot more engineers and everything, but as these businesses mature, they don't all have to take the big tech road of we're going to hire everybody we can find and just figure out something for them to do. I think a lot of these people are not doing that much, right? They're defensive hires. But if you run the business in a different way, if you don't overhire, if you don't try to suck up all of the engineers from the pool so others don't get them, in theory, right, we're not seeing it right now, but in theory, all this tech should scale very well, right, to run the same service for 5 million customers shouldn't take 10 times more engineers than for 500,000.

It shouldn't. It just shouldn't, right? The same software can be copied to a bunch of new servers, you can auto-scale servers in the cloud. There's a bunch of leverage from the tech that should be a fixed cost, but right now that's not what we are seeing. But I feel like if they mature these businesses, and they're well run for the shareholders, they shouldn't need a sales force that keeps going up, right, all the time. At some point, you don't need half of the planet to be your sales force. I don't know, maybe I'm idealistic, maybe there's something that's going to keep everybody just hiring and hiring all the time, and when they have 20 billion in revenue, they're all going to have 200,000 employees, but at some point, it's like, what are all these people doing? I think what we're going to see with all the layoffs that are going on right now is maybe a lot of these people were not doing that much that created value for the companies.

MBI:

That's always true, isn't it? I feel like that's always true. I think that if you go to any company, it doesn't have to be a tech company. Any company.

Liberty:

Yeah, government and yeah.

MBI:

Right. That's how it is. It's not like everyone equally contributes to the bottom line. It's very lopsided in any company. And I think the harder thing is to figure out which people are basically making that happen. I think it's a very fair statement to say any company, probably 20% of employees are basically contributing 80% of the value or 90% of the value. And I think I guess the way kind of Elon Musk did, I don't disagree with him that Twitter was probably overstaffed, and I think even if Musk was not there, that even that previous management would fire people, a lot of the people that they had. The difficult part I think as kind of news flow came along afterwards it becomes apparent. He ended up laying off a lot of people that he actually needs. He was kind of hiring them back and giving new offers to come back. So that's the hard part. You can probably go into any company and just fire 20% of the people, but which ones is the hard part.

Liberty:

If in the week you fire half the people by looking at their printouts of codes, you're going to make a bunch of mistakes, right? But I feel like tech is probably a bit different if you're a, I don't know, a farming company or a contractor building houses, well, maybe 30% of your people are slow and they're dragging their feet and they're not doing their work or they're doing bad work that needs to be undone by other, and okay, you could remove a bunch of people, but it scales more linearly. If you are in tech, if you have a few really, really good engineers, WhatsApp was like, I don't know, 50 employees running a billion people service. You get tons of leverage from the actual tech. The computers and the software are basically virtual employees doing work for you. That's where you can get some of that leverage that is harder to get in other industries.

If you're Boeing and building planes and you fire half of the technicians, well the planes are not going to get built. Maybe Twitter went too far, but a bunch of other companies, the software and the servers and everything can probably be maintained by a lot fewer people. And the question is, how productive are the others, right? Cloudflare seems to hire a lot, but be very productive, right? They have a bunch of graphs during the pace of new release of products and you're like, okay, these people are building products and they're shipping them, but a bunch of other companies, you look at the number of employees and the curve looks exponential, right? And then you look at the output and it's like, well most of the service are kind of the same, nothing's improving that fast, things are getting worse, right? I don't know, I feel some companies are much more ripe for a kind of reset of expectations for how many people are needed to do the actual thing.

MBI:

I do want to highlight one particular point, and that is it is hard to paint a broad brush and say all this [inaudible 01:07:48] certain percent gap with margin. My guess is some of them, couple of them will probably prove to be lot higher margin business and a lot more profitable business. But I think I would be personally surprised if the average, let's say gap EBIT margin for most cloud or software companies in 2030 is like 20%.

Liberty:

Oh yeah. For most of them, if I was thinking just Cloudflare, those are some of the best ones, right? If we're looking across the board, a bunch of them have basically new mode, right? Monday and Asana are almost selling the same product, right, and they're growing super fast, but if one of them raises price, a bunch of those have a lot of competition and few barriers, some others are more differentiated and they have more scale advantages, right, or, say, CrowdStrike or it's machine learning stuff has more data points than a new competitor, a new startup, or you can try to differentiate between some of them that have no barriers to entry and some that do. Cloudflare's infrastructure and its talent pool, that's hard to replicate. So I totally agree with you that the average of a lot of these companies, maybe the average is zero.

I don't know because there's a bunch of crappy ones that probably couldn't exist without a lot of free money, a lot of high price stocks. A bunch of their customers are people getting VC money on one side and there's a bunch of recycling of money in the ecosystem in the tech ecosystem. If that goes away, a bunch of these companies probably don't have very real businesses. But if we're only thinking about some of the very best ones, I feel like, at maturity, some of those may have 30, 40% margins. I don't know, it depends what they want to do, what the rest of the opportunity is, right? If they start to saturate their markets and they're like, okay, now's the time to go profitable, well, I think we saw an example of this yesterday. Cloudflare raised some prices on some of their city and stuff for the first time in 12 years.

And in that 12 years the plan has the same name, but what's inside of the plan is 50 times better and bigger and more, right? Their free plans today have more stuff than their paid plans had 10 years ago, right? But if Cloudflare was like, okay, we want to step on the profitability pedal, they could raise prices here and there and a bunch of their stuff is pretty mission-critical, right? People are not going to be like, okay, we don't need DDOS protection anymore, we're just going to cut that, right? So I feel like keeping low prices is indirectly a reinvestment in growth, right, because they're probably growing more because the prices are lower so it's easier to acquire new customers. And then once people are on board, people stick around for 5, 10 years, the gross churn is very low, right? So I don't know some of the very best companies can do that.

Some of the others, if they raise their prices, people will turn around and go to a competitor doing the exact same thing. So it's really a case-by-case-basis thing. I think maybe we should talk about DataDog a little bit. We're forgetting them a bit. It's probably fresh in your memory 'cause you just need a deep dive. I don't know them as well as Cloudflare, but if we wanted to do a quick overview of them, I can give us a start.

MBI:

Sure.

Liberty:

The general idea of DataDog, the way I understand it, is that their bread and butter is observability. Today in the cloud you have all kinds of services and servers and apps and you have to keep track of a lot of stuff, right, make sure it runs well, and uptime is money, right? If your website goes down, if your website is slow and hard to reach, every second, every minute that's going on, you're probably losing customer money.

Your people may just quit your service and go somewhere else. So it's very, very important to keep all that stuff running smoothly. So what DataDog offers is basically kind of a dashboard and what you do is you connect their agents and their APIs and their frameworks and all that stuff, SDKs, into your infrastructure, right, so into your services and your software. And these kind of software sensors are sending the logs and the data back to DataDog and they do a bunch of compute and give you a nice dashboard so you can monitor, okay, there's a problem with this server over there. Oh, the latency on this app is spiking right now, what's going on, right? Do we need to fire up more servers? You can autoscale with that kind of stuff. So that's kind of where they started. But then there's another big trend that they're writing where it used to be that developers and the people in operations were kind of separate.

So you had the people on one side, they were building the software, and then they threw it on the other side and said, okay, you over there, figure out how to run it, right, to put it on servers and put it on something that customers can interface with. And that's getting more squished together, they call it DevOps, it's kind of one word, right? So the people making the stuff and running the stuff are often the same people or working closely together 'cause now that's all in the cloud, right? It's not a bunch of people racking servers on-prem, it's kind of a different job. And that's further evolving into DevSecOps where you bring in security too, right, because DataDog has gone into security recently.

And so they're kind of writing that secular wave of all the stuff going on the cloud, all the stuff becoming serverless and desegregated and instead of having one big monolithic app running on one server over there, well, part of the app is running in that cluster over there and part of the app is over there and part of the app is an API at a totally different company, all that stuff is kind of glued together. And so to monitor all that stuff that's kind of spread around the world and around data centers, you need that kind of central dashboard, a central view into all of it, and that's what DataDog is trying to do. And the security stuff that they're doing, it's interesting because all of the data that they're gathering for the observability part and the performance part, all that stuff is useful for security, right? It's a lot of the same logs. They like to say that they have beachfront access to your infrastructure, right? Once you put them in, that beachfront can be used to build all kinds of other products.

So like many others like CrossTrack, right, they are building these new models all the time. I think they have 14 now or something. Their customers are increasing the number of modules they're using over time. So a lot of their sales are to the same base of customers. And one thing that I learned from your deep dive that's kind of cool is that apparently just before IPO in 2019, they got an offer from Cisco, I think it's rumored, nothing is official, an offer from Cisco for $8 billion to be acquired. So I think Olivier, the founder, can use my French for once, Olivier, the founder, said to grow to a large company, one of the hardest things is not to sell along the way, right, 'cause the better you are as a company, the more people that are going to try to acquire you.

MBI:

I didn't mention this on my deep dive, but actually, after publishing my deep dive, I came to this other news set about potential accusation offer for DataDog. Apparently, Salesforce also tried to buy DataDog after DataDog went public, I think. If I remember correctly, for like $20 billion. So I think it's probably 20, 22 billion today. So Salesforce actually tried to buy them. Again, [inaudible 01:14:16]. I'm not sure. I just saw some press reports, media reports that Salesforce might have been interested in buying DataDog. So yeah, Olivier was absolutely right. It remains a very hard thing to not sell your company along the way as you kind of build your business.

Liberty:

I think the classic there is Yahoo's offer to Zuckerberg to buy Facebook for a billion dollar. And I'm sure that happens all the time with every company getting tons of traction, you're the belle of the ball, everybody wants to dance with you. The other thing, one of the reasons why we're talking about both Cloudflare and DataDog is they have some similarities. I think they're both very, very high velocity in development. They come out with new products all the time. I love their website. They have great graphic designers. DataDog has some of the best blog posts. They have all these cartoons. And anyway, that's a small thing, but got to have a company with a nice brand image, right? Just the dog is kind of cool. But yeah, they come out with new products all the time, but what's interesting is that the products that they come out with seem to have a pretty high hit rate.

And that seems to be because they're not pushing them, they're getting pulled, right? So customers are like, hey, we wish we had this new feature, or we wish some of the same dashboard would cover CICD stuff or even earlier in development, or customers come up with future requests and DataDog builds that stuff for willing customers, right? The products are getting pulled out of them. So that's always a good sign and may explain why their NRS is so high, right? They're selling a lot more to the same customers. On top of being a consumption model, they're not pure SaaS, which has pros and cons, but they're selling by hosts, not by customer.

MBI:

Yeah. No, I think one of the unique things about DataDog that definitely stood out to me at least, when I was looking at companies or any tech companies that are growing more than 50% over the last 12 months, DataDog is the only company who basically spends more on R&D than they spend on sales and marketing. So for example, most of the tech companies, a typical structure is basically they spend 20, 25% of their revenue in R&D and 40 to 50% of their revenue in sales and marketing. DataDog is the opposite. They spent like 25 to 30% on sales and marketing and they spend usually 40% on R&D. So they're the only company, and I think as you were mentioning before, the reason their hit rate is so high that these are mostly pull-through rather than push-to their respective customers. They just build their product and they don't have to necessarily sell them.

So one of the things that Olivier mentioned was even when they kind of scaled from SMB to enterprise level, and I think this is a very common story, and this is actually a common theme for both Cloudflare and DataDog, even for some of the other tech companies as well, they all started focusing on SMBs. And then it's funny, even in 2017, I think DataDog used to have like 40% of their revenue from SMB. Now it's like 17%. So it's not a long time, right, four years, five years. Similar story with Cloudflare as well, the smaller customers used to have significant contribution to their overall top line, now it's not so much. So the way kind of Olivier explained is there is no difference regardless of the size of the customers that you are selling to, it's very bottom-up.

So even if they start with a CIO at a large enterprise company, the CIOs basically go to market engineers, talk to them, you have to convince them. And basically, after they kind of sell to those developers, we're actually in a more of a bottom-up approach, DataDog ends up later with the CIO with the bills and forms and everything, right? So DataDog doesn't have to necessarily convince the CIO, it's not a very top-of, it's more of a bottom-up approach. That makes a lot of sense why they don't have to spend as much in sales and marketing and deserve very valuable dollars. If you can sell your products without sales and marketing, that dollar can be spent on R&D. That's a very valuable utilization of your dollar. The only concern I have is because it's unique, it's something like this is perhaps their mode, right?

For many of these small tech companies, it's so hard to figure out what exactly is their mode because they're still in the early stage, and it takes time to basically the modes to be formed, modes to be fully formed. [inaudible 01:18:52] to appreciate and understand, oh, that's why you can't really disrupt that business. That's why you can't. Now we have the signs of mode, right? We are not really sure whether this is going to be a much bigger mode in like 5, 10 years time. So for example, for DataDog's case, if they can maintain this operating cost structure, if they can continue to spend less in sales and marketing and more in R&D, as they scale from 2 billion to 5 billion, 10 billion, that will end up creating some sort of innovation model product mode over time ecosystem mode, platform mode, right?

But the only concern I have, that's the bull case, the bear case is basically, this is what Jeff Lawson used to say, actually. He even wrote a book, right?

Liberty:

Yeah, I read it.

MBI:

Right. So he used to be a big [inaudible 01:19:45] on product-led growth, and that worked, that worked for a long time. It did seem like Twilio was doing what I'm saying about DataDog, they're just building great products, they're investing more R&D. But the challenge with usage-based model is your customer will always end up having a very power-law distribution. The revenue distribution will have power law and attendency. So the top 10% DataDog's customers may drive like 80, 90% of their overall revenue. That's [inaudible 01:20:16]. I don't know, we didn't really disclose it, but my suspicion is that's how it may end up, like the top 10, 20%. And if you are top 10, 20% customer, you probably know that you are big customer for DataDog. Now the trick is...

Liberty:

You have leverage.

MBI:

Right. The challenge I think that DataDog may have, Olivier is explaining how they price their product, right? So they say, hey, before launching a product, we were thinking about launching this product at like $12 per instance, right, and the night before we decided, you know what, let's go with 15. And they started with 15 and probably a few months, years later they increase it to 18. And he said there's no difference. So one interpretation is yes, the customers don't really care about price, they think it's a good enough product.

MBI:

Customers don't really care about price, they think it's a good enough price, it's a reasonable price. The other thing is, when I realized that I am top 10 customers of DataDog, I want that to be $8 per instance. I will have that power. If I use 14 different products, 20 different products of DataDog, I want everything to be cheap. So my concern is, for them to generate outsized profit margins, they have to be able to withstand that pressure. And that is possible, if they think this customer is bluffing, I can price this customer at $18 and this customer can't do anything because this customer can't actually switch.

Liberty:

Yeah. You have to be differentiated in some way. You can't just turn around and go to Splunk or someone else and get the exact same thing for a cheaper price.

MBI:

And there is truth to that, I do want to mention that. There is truth to that, there is switching costs. It's not like you can just rip it off the next day, if you don't like the price.

Liberty:

Yeah, it's built in a lot of deep software, like the SDKs and the [inaudible 01:22:02]... You can do it, it's not literally built into the code, but it's not just a switch that you flip either.

MBI:

But I think that's the key debate for DataDog. I feel like if, in 2030, if we see DataDog is still spending less on sales and marketing and more in R&D, I think this company will be much [inaudible 01:22:21] business, much, much more in a strong shape, than any of their competitors. Right now, they have some competitors, like Splunk has similar platform, they also are focusing on cloud and if you look at their cloud-focused revenue, they're basically growing at a similar growth rate... They used to grow at a similar growth rate in the last three years, I think this year DataDog is growing faster than Splunk's cloud.

So again, maybe it's because of the faster product iteration and the innovation that we are seeing on DataDog versus, let's say, some of the competitors. And that difference can grow larger and larger over time. In my opinion, it's super important for that sort of narrative to sustain... Yes, we don't have to spend as much on sales and marketing because we are differentiating through product, and our customers, it's a very bottom up approach, we don't have to necessarily sell it to the CIOs of the world to buy our products.

Liberty:

I have so many things to say. First to be fair to Jeff Lawson and to Twilio, the problem with them seems to be more that they don't have much leverage over the cost to the telecoms. It's a variable cost to them, so they don't scale as well as... So maybe the developer led, product led kind of thing, is still a good idea and maybe DataDog is in a better position to do that more profitably than Twilio.

MBI:

For Twilio's skills, I was actually specifically mentioning Uber.

Liberty:

You have a big customer with lots of leverage and then poof, they disappear.

MBI:

And also WhatsApp. And WhatsApp is still a big customer for them. I just want to add this one point, that I'm interested and curious to hear what you think... So one of the things I'm thinking about, DataDog may have a very bottom up approach today, in selling their, let's say... Infrastructure monitoring, API or log management products, but obviously they're now also getting into security and who knows what else, in two years, three years, five years down the line. The question that I wonder, I don't have a good answer to it, I'd love to hear your perspective, whether it can remain bottom up as they broaden their ecosystem, broaden their platform.

Maybe, the infrastructure monitoring products or APIs or even log management, doesn't require... Bottom up is totally fine for this kind of products, but is that also relevant for cloud security? Is that also relevant for XYZ, that's going to come up in the next three to five years? That's something I wondered, I don't have a good answer to. I don't have a good sense to how to think about it... But yeah, I'd be curious if you have any thoughts on those points?

Liberty:

I think that's a very, very good question. I feel like the answer is maybe some kind of hybrid. Maybe that's not satisfactory, but I feel like it's possible that you get in with the observability stuff by being best of breed, you do better there than almost anyone, and engineers, they go get you. And maybe for the other products, then you get into your distribution power, you get the bundle, because it's always a pull and a pull between, if you're building something, you're like, okay, I want a best of breed of everything, but then you have 17 different vendors and you have lots of agents and SDKs and APIs and it's a big mess... So most people seem to end up somewhere in between, where they're like, okay, Microsoft, there's going to be 10 of our things and then for five others we're going to go best of breed.

We're going to have Zscaler or CrowdStrike there, or Palo Alto or Splunk or Elastic or maybe DataDog... Maybe DataDog gets to be one of those, and then over time they can bundle other stuff with it, and people are like, "Well, we could go get our security from someone else, but we can get it super cheaply with DataDog and the agent is already there, we have to flip a switch and it's on..." So I feel like a bunch of their other products don't have to be quite as good, quite as best of breed, because they're just going to be bundled with.

So maybe DataDog over time can be kind of a mini Microsoft. Microsoft has big distribution power on one side, it makes everything simple, everybody knows they go to one place, they get it all. And maybe for some other smaller subsection of the stack, people go to DataDog and they're like, okay, we're going to get four things from you, 10 things from you... Maybe that's the way to get there. I also wonder if one of their advantages on cost is that they have a bunch of employees over in Europe, they're not all based in Silicon Valley. Maybe it's the [inaudible 01:26:38] advantage... I don't know if it's a long-term thing, if everybody else is going to do more geographical arbitrage now that more stuff is done remote. I know that Cloudflare is a very remote company, that's part of their advantage too-

MBI:

Even before the pandemic, right?

Liberty:

Yeah, yeah. [inaudible 01:26:54] too. Some of these companies are almost born remote, and they were kind of ready for the pandemic without missing a beat. But the advantage there can also be, I'm offering someone living in a very low-cost environment, a very nice salary. But maybe that salary doesn't have to be quite as high as if you were going to the office in Palo Alto or something. So maybe there's still some talent arbitrage there where you can tell people, you can still live where you want if you don't the big city or the expensive parts of the country... But I don't know if that's sustainable as inventive.

MBI:

I think you're definitely onto something. I think those two have to be part of the answers to the talent cost problem. One is AI, whether it can mitigate the need for the number of engineers that you may have to hire-

Liberty:

I think lots of people listening to this, if they're not programmers, if they're not in that world, they may imagine that coding is like you're sitting all day, thinking deep thoughts, doing algebra and just doing logic on screen and building the app... But a huge portion of coding is, I'm going to go and Google and try to find something on Stack Overflow and cut and paste some code from someone... It doesn't work, you change a few things, you go on Discord and chat with someone, and then... A bunch of it is template stuff and boilerplate stuff and you're trying to figure out...

So if you can do that much more efficiently with something like Copilot or Ghostwriter or these types of software, that frees up, especially the good engineer's time, for doing higher-value stuff. It's not so much that all of a sudden you don't need the engineers-

MBI:
Right, no.

Liberty:

...it may be that the engineers you have produce more value because they're spending less time on the stuff that adds no value. Some stuff is creation and some stuff is just typing. And if you have less of the make-work typing stuff, well maybe you get more of the rest. I hope so.

MBI:

Yeah. So the other aspect that you mentioned is talent in the remote, outside the US talent. That has to be also part of the answer. Obviously, I think the US doesn't have really a monopoly on talent. Talent is much more geographically dispersed and distributed than opportunities are, right?

Liberty:

There are no Americans on this call today.

MBI:

Well Americans, I am in the US though, so...

Liberty:

Yeah, yeah, yeah. No, I meant originally.

MBI:

Originally, yeah. Right. But yeah, I totally think that can be part of an answer. But I didn't see the evidence for DataDog's case though. Yes, they have a lot of employees in France, but when I was looking at their operating expense per overhead, it's still like $300,000. If you look at [inaudible 01:29:25]'s income statement, it's blindly obvious that they have some sort of advantage.

Liberty:

That's a good point.

MBI:

So that's one. And also [inaudible 01:29:34] is also, they don't hire, they really built the company much more efficiently. We can talk about the questions, we can talk about potential concerns... Most of the answers seem to be TBD. I think the onus is on many of these management of these companies, to actually show us a glimpse of what can be. Most of that is just assumed that these companies will scale it to a much more profitable business, now there are obviously a lot less believers out there. So it is a time for them to at least show a glimpse or at least some sort of... I can't believe, I don't know who exactly many of these management of these companies surround themselves with, isn't there anybody to tell them that non-GAAP margins are just not going to work? That's not how the math works.

I can't believe that people are still kind of boasting on, oh we are 25% FCF margin or 20% non-GAAP margin... Yes, I understand, you wanted to say that you wanted to preach that to analysts in 2020/21, but it's clear that even if... Sales analysts is never going to tell it on your face, because if they tell you they'll be blacklisted, they will probably not get the invitation for the call, all sorts of things... You can't expect that level of bold statements from sales analysts, but there should be people in those companies, in investor relations department, who are talking to many buy-side people, probably many other investors, who should be telling them how exactly the math works. If you're managing your long-term target to be 20/22% operating margin, then give a very specific guideline, what is this? Not long-term service compensation?

Is it 5%, 10%? What is it? You can't talk about long-term margin unless you also give some sort of guidance. Some companies talk about dilution, that's not in your control. You can't really manage dilution. If the company trades at 50 times, 70 times revenue, multiple, your dilution must be very minimal. You'll not be able to dilute your shareholders 5% even if you try, because the stock is still high. Now it's opposite.

So now you'll probably dilute much higher than in the last couple of years. So you can't really target dilution, it's very hard. It's not under your control. What you can control is basically, give some sort of guidance that this is the kind of [inaudible 01:31:56] with compensation, or maybe what's some of employee base you need... Those are the kind of things that I think investors need to hear. And I suspect, because I don't own personally DataDog or Cloudflare, but I do own some companies that have similar set of concerns and dynamics. And I'll be frank, I'm somewhat scared that if these companies, if this management is not surrounded by people who can tell them how exactly the math works, then that's not going to be good news for me.

Liberty:

Yeah, I hope the very best CEOs, they're going to adapt and change with the times. But I feel like a lot of it is mimetic. Every company is doing what every other company is doing, they're all doing the same thing. And that's why I put almost no stock into these long-term projections. They don't know... Cloudflare was a CDN. Right? If they had given us 10 year projections 10 years ago, what would that have been? And so even now, especially with companies that have very high development velocity like DataDog and Cloudflare, trying to predict out five years, 10 years... You can try to... General percentage of revenue and that, but if you try to predict what's the end markets that are going to be in, what are the tabs going to be? What are the products are going to be, they're probably going to invent a bunch of stuff, buy some M&A, added some modules that way... Some of those could be huge hits and become a huge part of the company.

You look at some companies where... I was running Invidia recently, the data center stuff was zero six years ago. Now it's the majority of their revenue. So some companies are very fast-moving like that. If you have a gravel pit somewhere, you can probably predict out five years, 10 years, you're still going to be doing the same thing. But with these companies, it's so hard. So it's more about having good management that's going to be able to, every day along the way make good decisions, and all those good decisions they add up to a good end result. Right?

MBI:

I totally agree and understand, but you still need to have the correct framework.

Liberty:

Oh, for sure. For sure.

MBI:

So if you have a wrong framework, if you're going in the wrong direction, then no matter what you see on the road, it doesn't matter, because your company, you'll end up in a different destination. If you are under the impression that as a CEO your job is to basically post 30% FFC margin or 20% up in a non-GAAP operating margin, you'll deliver that. But if you don't have the right framework, your stock will completely just move around based on what the Fed is doing, not necessarily what your business is doing. So as an investor, obviously I want to own businesses, I know Fed will always have an influence and role on what's going on, but I definitely don't want to buy companies that just moves under the whims of what General Powell is thinking or doing. So that's what [inaudible 01:34:42] and I'm saying, this is not a specific criticism to these companies. There are a couple of companies that I own, that have similar concerns, and if the management has the wrong framework, I think shareholders may suffer in the long run.

Liberty:

I totally agree. I guess I'm just not sure that what they're telling us in their presentations is anything like what they're thinking about the business internally.

MBI:

I hope you're right.

Liberty:

That's the thing, if they are, that's probably a problem. But I kind of doubt it. Maybe I'm optimistic, but I hope that they're thinking about it differently and they're all kind of saying the same thing, that's what [inaudible 01:35:17] want to hear, and for now that's good. But as a major business I would be very, very surprised if we're still giving out, I don't know... 20-something % SBC and they had 5% GAAP margin. I don't know, we'll see.

I think in both, Cloudflare and Datadog's case, it's also going to be a test of, everybody loved founders a little while ago and now everybody's like, yeah, get the founders out. They control everything and they're too rigid. These are two still founder-led companies. The founders own decent chunks of both. They have voting control in DataDog. I don't think they do in Cloudflare. I don't think Cloudflare has the old class. So that's also going to be a test of this kind of thing, is this... Everything that's attributed to founders, long-term thinking and bolder bets and better at keeping the culture, and making the hard decisions when things on the ground changes... These are going to be interesting to watch as showcases of founder-led companies, because there's fewer of them around these days.

MBI:

No, absolutely. I know the tide has changed, but I'm still a very founder-friendly investor. I still maintain a lot of the bull market philosophy that people used to have about founders. I would be very hard pressed to think Olivier, Matthew Prince, Michelle Zatlyn, or Alexis, are just going to waste money by just over-hiring and all that... And these people own this business. They have significant ownership. They have a lot more reporting power for DataDog, but a significant percentage of their personal wealth is probably these companies, so their incentives are highly aligned. So I share in some of your optimism... Yes, like I said, the last few years were a crazy time, crazy period... And you can have wrong framework, that's fine.

Liberty:

For sure.

MBI:

And if you talk to an investor and they're all saying give us 50% growth, and that's how we're investing, that's how you are hiring, that's how you're going... I get it. You can have wrong framework. But I share your optimism, it is highly likely that they'll be able to figure out the right framework over time, as these things... Personally, I think a lot of people, perhaps including us, are still... I don't know, hoping, that things are going to be normalized and it's not going to be as bad. But let's say if the recession happens in 2023 and it's actually worse than fear, I think the message will be there to these management teams. These are not dumb people.

If Matthew Prince takes 15 minutes of his day, he can figure it out, how exactly are shareholder values created, how non-GAAP margins are just not a right framework... This shouldn't take more than 10, 15 minutes for founders to figure it out. My only concern is, I don't know, I'm not a billionaire... I'm not running this $10, $20 billion companies, who knows how I would behave if I were surrounded by all these yes men and women telling me how great I am, how great a company I built... I don't know. But it shouldn't be hard, that's what I'm saying. And I totally understand your sense of optimism. It shouldn't be hard to figure out that non-GAAP margins are not the thing that you should be aiming for.

Liberty:

The question always is the carrot and the stick. The extremes have been so high and low recently, that does that distort management's thinking? So last year when prices were super high, is that getting them to drink their own Kool-Aid and be like, oh we're geniuses, we can do anything, we can hire anyone... And now, oh okay, a company is a complex thing. There's a bunch of pushes and pulling macro noise and effects and employees... And so these days they miss out the expectations from the street by 1% and the stock is down like 20%... Is that going to give them PTSD in the other direction and be too conservative, right? Too, oh now we're going to cut back on a bunch of stuff that was creating great value... But let's have a thought experiment for a second.

Let's imagine that these are private businesses. They've never been put in the public markets. We're only looking at their financials because... I don't know, we're on the cap table and we get reports, and we're looking at the revenue growths and the gross margins and all that stuff... Employees, number of customers, NRR... We're seeing all the same numbers, for the past five years of DataDog and Cloudflare. It's like, wow, these are doing super well. These are some of the best businesses in the world. So the stock movements on top of those fundamentals can affect psychology a lot for investors and probably for management too.

MBI:

No, I want to make one point absolutely clear. I'm not asking Matthew Prince or Olivier, or any of the companies that I own, to be profitable next year, next quarter or even two years from now. I just want them to have the right framework and give them the message, edit how they are thinking. I don't know how to even interpret that, if they think that they want to get to... If they have a different model but they're not communicating with shareholders... No, I'm not sure if that's the best way to give your shareholders the right message. The other thing is, if I were Matthew Prince or Olivier, I want to have the right kind of shareholders. I don't want to have the kind of people who are just looking with the wrong framework or we're just going to sell the moment I go down from 50% top-line growth to 40% or 30%. So you need to have... If you think about all the great companies, shareholder base is important. It gives-

Liberty:

I think they're probably learning that right now.

MBI:

Yeah.

Liberty:

Because they all IPOed recently, they're not Buffet that has been in the public market for 50 years, and they probably are learning that if you attract the Wall Street bets crowd or something, that's a problem because you're paying your employees in stock and you don't want your employees to be all depressed or something because they're looking at their RSUs going down and their stock options, or you're having trouble hiring great engineers because your stock keeps going down and they're like, oh, I don't want to go work for a sinking ship... You can create a bunch of perceptions that don't stay in the public market, they end up having a real-world effect on hiring and retention and that kind of stuff...

MBI:

For sure.

Liberty:

I'm betting that they're like... I don't know. It may change how the messaging and that kind of stuff in the future, they're probably learning the public market with a big trial by fire.

MBI:

Yeah, probably. Hopefully they will come to the right framework in not so distant future.

Liberty:

I hope so. I think that's most of what I had. Did you have anything else to add on these two companies?

MBI:

No, I think we are supposed to talk about only Cloudflare and DataDog, but we ended up talking about probably the broader space too. A lot of these concerns... These are nothing unique concerns for these two companies. In some ways these are more better concerns for the broader software space, not necessarily an idiosyncratic concerns. And like I said, I think there will probably be a couple of companies who will prove that these concerns are invalid for them. But yeah, so far, to me, it just looks like it's going to be super hard for the whole software space in general, to all have 20, 30% GAAP EBIT margin, let's say-\

Liberty:

Oh, for sure.

MBI:

...in five to seven years down the line. But I guess that's the job. I think in some sense, you and me probably shouldn't complain too much. I think that's the job, to figure out which companies end up with those kind of margins. If everyone ends up 20, 30% operating margin in five to seven years, then the job would be easy. Everyone's so profitable. The measure of capitalism is very different. The measure of capitalism dictates that most companies will not graduate to any level of compelling profitability.

Liberty:

Yeah. If software is a great business, well great businesses attract lots of competition and competition competes away margins and profits, right? So you need something in between to keep the competition away. And so that's what people forgot, I think, in the past few years. People went from not understanding software, maybe 10 years ago or something... I'm talking people in general. There were always people who knew, but... And then because software changed too, because Microsoft was known to be a great business for a long time, but people were like, well, they own the OS, they own the platform... There are not that many like that. Then people had to figure out the Cloud, for a long time AWS was totally misunderstood. And then over time people started to get really excited about all this SaaS and Cloud stuff, and I think it went way too far. And painting the whole category with a broad brush... And all these are going to be winners, right?

MBI:

Yeah.

Liberty:

It's not that simple. But I think that's why I like that we're not only looking at the income statement here, we're talking about the products, the management, the cultures, the hiring, the R&D, the end markets... Understanding that kind of stuff, I think is where you're going to find that, okay, this company is different from this other one. They have the same gross margins and they have the same growth rates, but this one has probably a lot better chances of having good terminal value and stickiness and of creating new products and all that... And this other one, well, competition's going to come in and nothing's going to stop them from a race to the bottom. So that's the part I like about investing. It's trying to understand the systems level, how the machine works, not just these numbers versus these numbers.

MBI:

No, so true. It's true. I think the fact that it's so challenging makes it so much fun.

Liberty:

That's a good place to end it. Have a good day my friend.

MBI:

Same to you. Take care.

Liberty:

Bye-Bye.

Liberty’s Highlights

Discussion about this post