My guest today is Ben Thompson. Ben is the author of my favorite business strategy newsletter called Stratechery. He’s also the host of the exponent podcast, and now the Dithering, a podcast he recently launched with John Gruber. I think Ben is among the most interesting business analysts in the world, and I’ve learned from and directly applied many of his ideas. We cover many of the major concepts he’s introduced over the years, including his well know aggregation theory. I think that to understand how the internet has changed the business world for good, you must read Ben and follow his thinking. I’m excited to finally have him as a guest on the show. Please enjoy our conversation.
(01:26) – (First question) – Companies that are built for the next disruption
(1:32) – The End of the Beginning
(9:58) – Aggregation Theory and the Smiling Curve
(13:18) – Steps to creating an aggregator
(19:46) – Pattern of successful aggregators or luck?
(24:34) – How aggregators interact with suppliers and consumers
(30:49) – Taking on other aggregators
(34:09) – Platform vs aggregator in the scope of Shopify vs Amazon/Walmart
(40:55) – The Moat Map
(46:16) – Value chain thinking and profitable business models
(51:58) – Future of media and independent content creator’s vs bundles
(56:07) – Bundling independent creators
(1:00:37) – The infrastructure layer of technology and software companies
(1:02:35) – His thoughts on gaming platforms
(1:06:13) – The atoms vs the bits in the tech world
(1:12:18) – What he’s learned from covering Netflix
(1:13:46) – Kindest thing anyone has done for Ben
(1:15:56) – Stratechery Podcast
So Ben I thought a fun place to start would be with a recent post that you call The End of the Beginning. And we’re going to back way up after we talked about this concept to talk about aggregators and platforms, ecosystems and all these cool things you’ve written about, but love to start at the end there. Tell us a little bit about what you were thinking through in that post and why it might be important for these big incumbent technology companies.
Well, first off, thanks for having me on, even though I’m not your typical investor. I’ve been a big fan of the podcast. So that post, it’s interesting because there is an aspect to the post where it’s kind of obvious, but also it’s sort of very heretical in some corners of Silicon Valley in particular. So the idea there is that maybe tech isn’t so special after all, meaning that tech likes to think of itself as we’re sort of constantly disrupting ourselves and it’s like back in the day IBM was powerful and then Microsoft came along and then Microsoft was powerful and then Google came along and then Google was powerful and then Facebook came along, et cetera, et cetera, and Apple came along et cetera and the idea being that there’s always a sort of new power around the corner and the position of the largest companies is very fragile.
Even though they look totally dominant today in the future, they’re going to be disrupted. Disruption certainly fits very much into this sort of framework. The question that I have is it sort of looks at technology as the sort of discreet sort of series of events where you went from the mainframe to the desktop for example, and I’m obviously compressing significantly. There is many computers and a whole host of things sort of in the middle there, but then you went from desktop to mobile and you went from the desktop to the internet and all these are sort of presented as discrete events. It’s like we’re in one era and then that era goes on and then suddenly the era is disrupted by sort of a new era. You can actually look back and all these eras from another point of view are actually all on one continuum where you sort of have two dynamics going on and so on one side you have the shift of how you interact with computing, going from being a sort of batch process that’s sort of destination oriented where you literally had to go into the compute room and you went into the compute room with your stack of punch cards or whatever it might be, and you went in and you ran your job and then you came out and then you had your job finished and you went back and you checked it and then you made a new job.
Then you went in and did it and then you shifted to sort of being on your desktop was like, wow, the personal computer, because you have your own computer, you have to go to the, like the company batch center or whatever it might be. And you could sit there and you could do job mobile while your phones with you and people have talked about it. The phone is actually the personal computer, like we wasted a perfectly good name for the phone on the desktop computer, but you think about it, we said personal computer cause it was my computer but it was still a destination device. You had to go to your desk to use it. And the phone, what made it different is the phone is always with you and I would say that’s actually one broad process of going from destination batch oriented computing to continuous always with you computing and instead of those being sort of discrete steps, it’s a progression and where we’ve ended up in pretty ideal spot as far as that sort of paradigm is concerned, which is you have this device with you that has a large enough screen to do work done is small enough to be pocketable.
It’s always connected. You can always be connected with it and you think about it, what’s after that? Well there’s the watch but the watch is almost too small. It’s better for a few specific use cases, but it’s not better as a general purpose computer. And you’d see the same thing with like AR glasses for example. You could see them being better for specific use cases, but at some point you sort of cross a line where you’ve maximized the general purpose usefulness and now you’re going to specialized bits. You saw this with chips for example. Where it just sort of all came into a central processing unit and what’s been characterized the industry over the last decade has been an explosion of specialized chips. And I think you’ll see this sort of devices, you’ll see an explosion of devices and sensors, but the core piece, the hub of it.
That general purpose device that can scale in lots of directions is the smartphone. And we’ve kind of reached a logical endpoint of that sort of revolution. And so if you think about it in that perspective, that suggests that iOS and Android are actually far stronger and in a much better position than even Windows was. And Windows was thought to be the most sort of dominant monopoly ever. Why? Because there was a natural step a hardware enabled step beyond the PC that did everything the PC did but better it was more continuous. It was more usable in more places. Whereas what’s after the smartphone you started getting these specialized devices which may be better again for some use cases, but generally speaking they may not replace the general purpose nature. So that’s one side. One side is sort of the interactive part of computers. On the other side is sort of where data lives and sort of broad compute generally.
What’s interesting is we started out with an all in one room so when you went to that batch room to run your stack of cards, that was where the mainframe was and there was like a big computer that took up the whole room or multiple rooms and it ran all the computations. It spit it out. And so we started out with it being actually one sort of big piece. And then when you went to the desktop, yes, you had some compute locally but that connected to the internet and you sort of had the on premises server room and things would be handled there and there’d be data storage there, et cetera, et cetera. And now you back up into the mobile era at the same time you’re shifting to the cloud. As far as the backend storage goes. What has happened to the cloud? Well, it’s gone from being to one room, being on one campus already with VPNs to be one entire company to spanning the entire globe.
These companies brag about having how many data center reaches they have all over the world and what’s next after that? Are we talking like interstellar? But you think about it again, there’s more specialized applications. CloudFlare has taken on Amazon on work on the edge, but that’s maybe a more specific use case than just general widespread everywhere available compute. And so you’ve had the compute and data storage layer expanding from one room to being omnipresent in the cloud. You’ve had the interaction layer expanding for being a batch processing one room to in your pocket everywhere. And it’s kind of a natural end point. Anything past this that’s getting super specialized is not necessarily general purpose. So it follows then, the companies that are dominant at this specific moment in time at this end point are by far the best place for anything going forward. So I already mentioned iOS and Android on the handheld. In the cloud, it’s Amazon and Microsoft and Google to an extent.
And those are the companies that are best placed, not just because their current paradigm is probably going to last longer than people think, but also they’re best placed for these specialized applications. Because those specialized applications are not as large of an opportunity. It’s not going to have this sort of disruptive potential, that cloud relative to on premises server have or mobile relative to the PC had. And so in this view, the companies that we’re at, or this is the foundation sort of going forward and whatever happens is going to build on top of these companies. And in that article I gave an analogy to sort of the car industry where we had this explosion of hundreds and hundreds of car companies formed per decade in the first few decades of this new century. And then by 1930 it was over, the top three car companies were established, they were established basically all over the world and there’s barely any new car companies formed until Tesla like 67 years later. Because once things were in place, once the economies of scale were in place and once the sort of paradigm was set, it was set. And so that’s my point about it, sort of being obvious. This is sort of what happens to most industries, but tech has always thought it was special and that’s the sort of heretical part to say, well maybe you’re not so special after all. That was a very long answer to your question.
No, I love it. I mean it’s such important history and so far the stock market agrees with you. If you take price action in this most recent period as an indication of whether or not that will be true. It seems as though the market thinks it will be. It’s continued to assign really, really not crazy rich, but rich valuations to these companies and they’re near their all time highs.
It makes sense because I think the way to think about the pandemic generally and its impact on the world is not that it’s gonna provoke a sudden shift in direction or rather because we were already in this period of transition because of the internet and the internet impacting all sorts of businesses and the way people work, et cetera, et cetera. With the pandemic I think did was accelerate trends that were happening anyways and if this thesis is correct that the trend is these companies become even more foundational and critical to everyday life, then the pandemic ought to help their valuations because it’s accelerating that shift or that completing that shift as it were. And so from that perspective, it definitely makes sense.
We’ll come back around to some of the interesting issues around whether or not these companies should be regulated, some aspects of them should be considered utilities and so on. But I want to back up to some of the root level ideas that you’ve introduced into the world around technology strategy over the years. Starting with the big one, which is Aggregation Theory and I thought a neat entry point into aggregation theory would be this concept of the smile curve. Could you describe what the smile curve is and then we’ll use that as a jump off point to go a couple of layers down and aggregation theory.
The smiling curve is actually from here in Taiwan where I live. The former CEO of Acer, Stan Shih, the founder based, or actually what he observed a long time ago was that we weight the PC industry. On one side there is sort of the brand and service and marketing side of things and on the other side is sort of the R and D things that are secured by patents and excellence in manufacturing, et cetera, and then fabrication is in the middle and the value of those components looks like a smiling curve where it’s high on the end. The brand and marketing and all those sorts of things are high. The specialized technology secured by patents or foundries. Superior manufacturing technologies is very high, but fabrication actually putting stuff together, is very low. A company like Acer, ironically enough tended to sit in that fabrication space where you win in the fabrication space by having a superior cost structure, but you’re not going to capture a lot of value out of the value chain and you see this, for example, the iPhone is that a company like Foxconn is relative to their perceived importance has actually a relatively low market cap.
Whereas in this article, when I originally introduced this, I compared them to Largan Precision, which creates the lens assembly for the iPhone, a tiny part of the phone, but a much higher sort of relative market cap to their sales because it’s a highly differentiated piece that is sort of a sustainable protectable position and obviously TSMC being on the is a good example of this. Intel traditionally was on this side is why they charge so much for the processors. And then the other side when Stan was writing about this, he was thinking about, OEM, Dell basically or Compaq. When they sort of shifted away all their fabrication to Asia and they were trying to capture that high end and using [inaudible] up to the chain here, but particularly Windows and something on that higher end side of the PC value chain. And so this is this idea where value flows to either highly differentiated inputs on one side or sort of the end part where you’re capturing a lot of consumer attention on the other, and to be in the middle,
putting pieces together is not a particularly great place to be. I actually first introduced this concept, I talked about the Acer example and wiring and precision, but I was actually worried about publishing where the idea in publishing was increasingly that the value is going to flow to either side, either the side that captures consumer attention, which was Facebook and Google and these companies where users go to them first and then follow links to publishers, or it’s going to go to individual content producers like myself where you could actually be highly differentiated and you could be able to capture value in that way. And the publishers were the fabricators where they were sort of in the middle trying to use the differentiated inputs to sell to the users at the end, but actually all of the power and value was accruing to the ends and they were losing sort of their position in the value chain, which was different than the old days where the local publisher had a local monopoly and their power was predicated on controlling distribution. But when there’s worldwide distribution then that sort of goes away.
This might be a ridiculous question, but I’ll ask it anyway.
Partly ridiculous because the major aggregators may already be created in the internet era. Sort of back to our opening discussion, but the question would be how would one go about creating an aggregator? What are the steps to building something like this? And I guess of course we need you to begin with a simple definition of your sort of three rules for what an aggregator is. The one that I’ve seen you write before are direct relationship with the customer, zero marginal costs and lowering customer acquisition costs as it scales.
The idea of an aggregator is something I’ve been developing for a long time, but the most important thing is first and foremost is that it’s different than a platform. This is a big frustration I have, we can get into it perhaps later, particularly when it comes to questions of regulation, is that people think about big internet companies and they’re frozen with this picture of Microsoft where you have a platform with an API and drivers where so OEMs plug in on the bottom and applications sit on top and it’s sort of a facilitator for the entire ecosystem and platforms have tremendous value because again they’re a facilitator for an ecosystem. They’re also dangerous because they have direct controls or components in that sort of value chain and so you can squeeze the OEMs on one side. You can squeeze developers on the other. I think an example today would be Apple and their sort of control of the app store is a great example of leveraging the power of a platform in perhaps problematic ways.
An aggregator on the other hand, is very different in that they sort of, they leverage abundance because there’s so much stuff out there. Consumers are overwhelmed and they need a way to sort through it. So this idea, classic example is Google, where there’s so much stuff on the internet. How do you find stuff on the internet? Well you go to Google and Google helps you find it and when you go to Google and you find something, well then that’s great. You go there and everyone benefits. You benefit, Google benefits, and the site they directed to you benefits. And now all those sites are motivated, sort of work better on Google. They want to work harder to show up better behind Google search results. You know what’s funny, the SEO industry is in many respects, an industry where people pay money to be better Google suppliers. And so you think of Google’s perspective, they’re not spending a dime, they’re just putting specifications out there and people are jumping say, Oh, how can I serve you better, Google?
And they serve Google better. And then users get better results, et cetera, et cetera. You get sort of a virtuous cycle and then boom, Google can drop ads in front of you. And so what enables us to get to sort of those threes or characteristics you mentioned one is the sort of direct relationship with users. Google is directly connected to users. They go to Google to search for something or they go to the search bar in their browser. Facebook, same thing. You go to Facebook first. It’s a place that you go to directly and that’s what is the sort of core at capturing value and you do that because you provide something users like. This is what regulators have to get through their heads is no one’s forcing people to go to Google. No one is forcing people to go to Facebook. They’re going there by choice.
They could not go there. They could go to Bing. Or a better example, they could type in a URL. This is where complaints of companies like Yelp for example, fall very flat for me. It’s like at the end of the day, it wasn’t that long ago where people were saying, Oh, Google is doomed, and I wrote some things that weren’t always this a problem for Google because people can just get apps and do sort of search directly, right? They’re going to lose local because people would just go straight to Yelp. Well, it turns out that Google responded very aggressively to that. And is that a bad thing because they didn’t stop users from going to Yelp. It’s not like Yelp was banned from Android. That’s a platform issue. If Android suddenly made it so that the Yelp app didn’t work on Android, that’s an obvious abuse of platform power.
People using Google instead of using Yelp, is a problem of you not liking what customers chose to do, but there’s nothing you can really do about that. Right. And so some of these regulations end up being like pushing on a string where it’s like we started with Europe and Google shopping where they’re like, Oh, Google disadvantaged these shopping comparison sites. Well you know what? If you go to Google and you search for shopping comparison site, they give you a list of shopping comparison sites. If you search for shoes, they give you shoes and somehow this bothers regulators that by giving customers what they wanted, which were shoes, they didn’t serve up shopping comparison sites. It’s like, well I mean your problem is that customers are not searching out other shopping comparison sites, which is a marketing problem for those companies. It’s hard to see all that’s sort of Google’s responsibility.
And again, that’s not to say Google hasn’t committed abuses, it’s just to say that if you don’t understand the nature of their power, which is customer choice driven, your regulations aren’t going to work. And we’ve sort of seen that again and again. So number one is relationship with users. Number two is the sort of zero marginal cost of serving users. And what this means is because Google, when they get good, they can scale infinitely, right? There’s no natural geographic limitation or production based limitation on how many people they can serve. This matters for both a marginal cost perspective and sort of transactional cost perspective. It’s a lot of just accounting. Stratechery, for example, could I scale infinitely given that I’m doing a credit card transaction with everyone? There was probably some sort of gating factor. With Google it’s also frictionless that they can sort of scale infinitely. Facebook is the same sort of thing.
Number three is this idea where you get decreasing acquisition costs over time and so you think about Google, the more people use Google, the more data there is in the system, the more feedback loops there are, that makes sort of the search results better and that attracts more users and so even the marginal user is the products getting better over time. I actually think a better example of this that I like to use is Netflix. Netflix is a little interesting because they’re dealing with money is involved in a much more direct extent. So the scalability factors are much more limited relative to say a Google or Facebook where it’s all frictionless. But you have this concept where the marginal user for Netflix at the beginning, was having to sign up for a service with a few thousand shows and now the marginal users going to Netflix is signing up for a service with tens of thousands or hundreds of who knows how much content is on Netflix.
Why? Because the content they seek out is evergreen. It’s always available. I can go watch Orange is the New Black, one of the original shows and it’s valuable to me today. So the marginal value of Netflix is increasing as their customer base increases as opposed to a lot of services where the more customers there are, it degrades in quality and becomes even harder to acquire them and then they have to spend marketing costs to acquire them. And that’s just where companies fall apart. So many companies based their sort of projections and calculations on the cost of acquiring a customer at the beginning. The problem is at the beginning you’re serving your ideal customer, the one that really wants your product, and so they’re going to look over the problems with your product, et cetera, et cetera. They’re going to be easy to acquire and usually your marginal customer gets more and more difficult to acquire. You have to spend more money, and what happens is Facebook and Google actually end up taking all your profit over time.
How much of the aggregators that we have that we can point to, do you think strategically thought about this concept early in their lives, whether it was the founder or the CEO? I’d be curious to hear your thoughts on how these things tend to get going. If there’s a common pattern between say Facebook or Google or some of the other aggregators that you’ve written about Airbnb being a level two aggregator. I’m just curious if you think there’s some sort of magic spark that makes it possible for a company to potentially grow into one of these things.
Yeah, I mean, I think that that is giving me, in this concept too much credit. There’s certainly a bit of a descriptive nature to this. I think someone who’s written about these concepts a lot is Bill Gurley. Particularly this idea of sort of owning demand is way more important than owning supply. What I hopefully I’ve contributed is noting that this concept is more generalizable than maybe it first seemed and it’s actually pretty descriptive. And the other thing that I think I’ve really focused on is why this is different than a platform. And this is actually something that I’ve talked to Bill about a fair bit is pointing out why this distinction is actually really, really important. And so I think that’s probably my biggest contribution is teasing out why those are different and why that matters for how to think about these companies. But you know to actually think about this, probably the best example of, I think a company that’s thinking about this explicitly is Spotify and sort of their shift into podcasts.
So the recent news about them taking Joe Rogan and he’s going to be sort of exclusive to Spotify. A lot of people assume, Oh, that’s just Spotify trying to get more subscriptions and I think they’ll be happy with subscriptions they get, but this is pretty clearly an advertising play and that’s why, for example, Spotify recently shifted their accounting of costs for content completely to the advertising division. Like I mean people keep pushing back on this. Well it’s like, well you don’t change your accounting if you’re not clear about where this is going. But most people are not into the ins and outs of accounting to say the least. Although your audience I’m sure loves it. What I see Spotify trying to do is trying to get a critical mass of listeners to podcasts and meanwhile their technology is much better suited for monetizing a medium that’s very under monetized.
How do you monetize a podcast now? Well one, you could have a host read an ad which is very hard to manage and scale and so you see the companies doing it are ones with very large lifetime values because it’s sort of worth the time domain registrars and stuff like that. Or you have something, we’re using the dumbest sort of dynamic insertion which is you insert on download and so you see the IP address. This person is in Los Angeles area, so we’ll put something somewhat related to Los Angeles in the download. You know nothing else about the customer. It’s the dumbest sort of targeting possible and I think people try to say this is similar to Spotify, not at all. Spotify is trying to deliver a Facebook type of advertising experience. They have a relationship with you. You already have a few hundred million people with Spotify in general.
A lot of them are paying customers so they have zip codes, demographic information. They have your name, they have your email address because you had to register with them. The email is sort of the key to unlocking identity broadly and they have a much more finely tuned understanding of their customers, number one and number two because Spotify streams everything, they don’t need to worry about this download issue because as you’re listening to a podcast, they’ll be able to put in an ad specific to you in the time and place that you are and it will be different for everyone. And again, we’ll see if it works, but you can see a world in which this monetizes way better than other podcasts ads, just like Facebook ads monetize way better than your general sort of banner ad on the internet. That’s not applicable to anyone. If that happens, then suddenly their CPMs are so much higher than everyone else that future podcasters, Spotify won’t need to pay them.
They’ll say, look, we have all these customers, we monetize better because we understand these customers better. You need to come onto our platform and use our superior, we will monetize you better. And then suddenly they become the center of this where they have a critical mass of users who they understand better and they pull podcasters on and start getting this sort of a virtuous cycle and becomes sort of the center of podcasting. We’ll see how that works out. It’s interesting because in one respect, podcasting is super fragmented and that’s a great place for an aggregator to come in and sort of clean up the mess. Google, came into a world of all these websites everywhere that was just hard to manage and they just organized it and made it approachable. I think a challenge for Spotify is because the ecosystem is relatively has been around for awhile.
There’s a lot of people that are set in their ways and you see that even Spotify to date, their growth in the area has mostly been converting new podcast listeners, making people who didn’t listen to podcasts before start listening to them, the big question is can they get people to actually shift their habits to use Spotify and it’s going to be a little tougher because that’s what they have to do is exclusive content. If you do exclusive content, then your values start being captured on the smiling curve idea by sort of the differentiated creator themselves, and so it’s going to be a little more difficult than the path that Google or Facebook went, but it’s certainly what they’re trying to do.
You’ve raised such an interesting example with Spotify. I think about sometimes Netflix in a similar vein where it lets us explore this concept of the suppliers in the marketplace. So we’ve already talked about these concepts in a few different company examples. You’ve also written about how there’s sort of three different levels of aggregator depending on the type of supply. The magic of course is the Google and the Facebook where people, like you said, paying money to make themselves better suppliers, let alone you’re not having to pay them anything, but things like Netflix and Spotify, obviously they’re paying suppliers to come on the platform. And I guess I’d be curious what you call Netflix’s originals, I guess first party suppliers or like this.
But how do you think about the role of supply and from a competitive standpoint, how aggregators interact with providers or suppliers?
The idea that on the internet controlling demand is more important than controlling supply is sort of the big picture takeaway, but to your point, the relative power of suppliers determines how big and profitable the company that controls demand can be. Take Spotify’s other business music as an example. In this case because the suppliers sort of have a monopoly on music for all intents and purposes, what’s so brilliant about the music industry model is because they’re doing sort of almost a venture idea where they give you money up front to do your recording, but then they own it. They’ll take a lot of losses on a lot of bands that don’t break through or singers that don’t break through, but because the ones that do, the moment that song is recorded is now part of the back catalog and the power of music companies come from the back catalog.
You’re like, Oh, why don’t do artists just go to Spotify? It’s like, well they still need help sort of getting off the ground and if Spotify, so Spotify is certainly working in the long run, I think to make that so that something they can help new artists with, but as long as new artists are going somewhere else, the back catalog of these companies is getting more valuable over time and the more valuable that back catalog is, the morning negotiating power they have relative to Spotify, thus they can take a percentage of subscriptions as opposed to, well that’s the other thing why the podcast about subscriptions doesn’t make sense. Spotify pays based on the money they make, not necessarily the share and yes, they’re trying to change those parameters over time, but the hold that the industry has on Spotify, because Apple can step up and offer the exact same songs is a very strong one.
And so Spotify is a very weak aggregator in the case of music. Whereas podcasts, because the supply is much more disparate, everyone’s independent, there’s not sort of that unifying force tying a bunch of them together. That’s actually a place where Spotify can exert much more potential control and power in the space. If Apple actually cared or had any interest in doing this, they could have rolled up the podcast space ages ago because they had the central directory. iTunes doesn’t host podcasts. A lot of people don’t realize this, but it’s sort of the directory where everything is and the more important thing is the Apple podcast player is the biggest podcast player. Had Apple chosen to leverage that particularly a couple of years ago. They could have built this whole thing, built an ad network, but all these sorts of things. It’s just not what Apple sort of does and so they sort of just let it be, but that’s where the power sort of would have been because they controlled the map.
They control the end user. Netflix is interesting here because Netflix has shifted over time to integrate down the value chain and exert more and more control over suppliers, so it used to be people don’t realize that the Netflix original strategy has actually also shifted over time where originally they were just buying shows. House of Cards was, I believe it was Sony. It was also sold internationally. So I remember it being on global TV here. It was only on Netflix, I believe in the U S. Sony took on all the costs. They produced the show and they sold it to Netflix. What Netflix shifted to over the last several years, and this is why Netflix has had such crazy cash burn, is they are producing shows from the get go, which means they own the entire thing. They own all the worldwide rights, they own all the residuals, and it’s something that creators have grumbled about.
A lot of them don’t get residuals on Netflix. So when Netflix keeps all the upside, they pay more upfront for that right. But in the long run, you remembered the Orange is the New Black concept. It lives on Netflix forever. It provides customer acquisition benefits to Netflix forever. And so Netflix wants to keep all that forever upside on their own books. Well, if you’re putting in all the money up front for a show, taking all the risks, you’re going to spend a lot of cash. And that’s why for about three years, Netflix’s free cashflow was massively negative because they were pulling forward all these costs that whereas like House of Cards, they paid Sony over time as it was delivered. They’re putting in all those costs up front so that in the long run they only have revenue upside and you don’t have cost in the future.
I mean obviously from an accounting perspective they’re going to deal with the cost then, but from a cashflow perspective, the cashflow is all pulled forward. The idea there though, is, they’re actually exerting more control over supply so they can garnish more upside from controlling demand. People really fundamentally, I think don’t get that specific point, there’s such a focus on their free cashflow without understanding why is it that their free cash flow for these few years was unusually high in a way it wasn’t before. In a way they projected won’t be going forward. Well because if you’re shifting your business model from cash flowing out over X number of years, to all the cash flowing out in one year and then you garnish it in the long run, you’re going, if you’re shifting your sort of working capital model, you’re going to have a huge cashflow.
But once you adjust, then it will look more normal over time. I think that’s what you’re seeing with Netflix forecasting that their cashflow is going to improve. It’s because they’ve now shifted their model and sort of flipped it on its head and we’ll be able to capture that upside without these sort of extreme sort of outflows in the meantime. And meanwhile they’re doing it at a time where interest rates are minuscule. So it was a good sort of taking advantage of the moment and now they’re, I think, much better place sort of going forward. And by the way you see this in the pandemic, a lot of the networks and a lot of the traditional folks are trying to stretch out their content because they don’t have enough content. They only have content three to six months ahead of time. It’s like just in time content. Whereas Netflix, because they’ve shifted to this working way ahead of time sort of model, they have content for the next year and so they’re not shifting their timing at all. Obviously they’ll might have to make adjustments to production going forward, but they’re sort of ended up being better placed because they’ve shifted to a higher inventory model, if that makes sense.
We’ve talked about two really fascinating examples where the categories are sort of video for Netflix and as Daniel’s talked about with Spotify, not music but audio for Spotify. How do you think it’s possible to compete against one of these aggregators when scale economies tip in their favor? So the Netflix example, the amount of cash that is being spent and burned to produce this stuff is kind of staggering and its growth rate is staggering and obviously they’ve got an enormous scale economies advantage. Do you think that it’s possible to attack them in that category of video?
I think it’s very stupid to attack them head on. And so the problem is that, just think about that scale advantage. If Netflix spends X amount on a show, they can spread it out over way more subscribers than you can. Which means that your sort of cost per subscriber for a new show is going to be massively higher than Netflix’s and that’s just going to be a fact of life for all these folks sort of going forward. And that’s a tough place to be in. The answer though is you don’t necessarily have to take the same business model approach as Netflix. So Disney, the reason what is so compelling about the streaming model and the low price that they’ve done it is they have other ways to monetize customers. They can monetize through theme products or cruise ships. This is all pre pandemic thinking, but one suspects
that would be the case in the long run. We shall see, but the idea is that Disney actually establishing a direct connection with customers really for the first time. I mean beyond sort of maybe their cruise ships and stuff like that, to understand their customers have their email addresses, to have their credit cards. To be able to communicate with them directly. It’s so valuable given the other ways Disney can monetize those customers that you have to think about Disney plus, not just as being a potential moneymaker, but as a phenomenal sort of top of the funnel tool. A lead generator for the entire business. And so in that case they can compete with Netflix because their payoff for buying content or not selling it to Netflix as it were, is monetized not just by their streaming price, which is where they compete with Netflix head on, but also by all the other parts of their business.
And so it’s a much more sort of rational approach. Another company where they’re just trying to also build a subscriber business, compete with Netflix. You’re trying to compete on Netflix’s grounds in a manner where Netflix has a massive advantage and it’s going to be a very sort of difficult position to be. Now, you couldn’t brute force it. I mean if Apple really wants to spend several billion dollars a year, that could be a problem for Netflix. But it entails Apple accepting pretty significant massive losses for quite a long time by definition because they are going to spend less efficiently just because they have a smaller user base. That’s just why Netflix is difficult to compete with in this regard, again the scale matters. You need a different business model, different way to monetize or just limitless cash and willingness to suspend it. I think the outcome here is the next couple years will be a little tricky for Netflix because there’s all these competitors popping up, but I think in some respects it’s setting Netflix up for the next stage, which is all these companies realizing this is a terrible business for them and they’re actually better at being suppliers and then they go back to selling to Netflix and then Netflix ends up being sort of the future cable bundle for basically anything that’s not live and everyone just sells to Netflix and I think that is actually is the logical place where this ends up.
So maybe a rough few years as Netflix has to compete. But I think they’ll win that competition. And then there’ll be very well placed
Leaving media now to talk about two other fascinating companies as it pertains to this lens of sort of platform and aggregator, which are Amazon and Shopify. So the famous example of Shopify’s call to arms is arming the rebels against maybe the, the empire that is Amazon. I’m curious how you think these concepts of platform and aggregator apply to those two companies and whether or not Shopify specifically will trend towards becoming an aggregator. When thus far it’s really has been about arming merchants with better and better platform tools.
It is a good example. And the third company I would discuss here is sort of Walmart and their particularly jet sort of focused sort of eCommerce ambitions. What Walmart has figured out is, it’s a great example. To the extent they can do things differently than Amazon, they can succeed. So grocery, for example, Walmart’s doing much better than Amazon because they already have existing grocery infrastructure and it turns out a big problem with groceries, perishables where you need sort of a regular supply chain that’s different than an eCommerce supply chain. And Amazon’s really struggled to sort of figure these aspects out because they’re totally optimized for sort of a different sort of use case. And whereas Walmart has done well, Walmart meanwhile has done terribly at traditional e-commerce because they’re not optimized for that and Amazon is. And so Walmart then tried to build up their capabilities to take on Amazon and it was a total disaster because Amazon, they’re trying to beat Amazon at Amazon’s own game and Amazon’s already at scale.
They already have all the capabilities and all the customers go to Amazon already. How does Microsoft [Walmart] compete? They offer lower prices, they try to marketing now you’re killing yourself with customer acquisition cost and you’re sort of getting hit from both sides. You have a worst cost structure and you’re handing money to Facebook or Google to acquire customers. So it was a very sort of tough go of it. Meanwhile, Shopify is very different. If you’re a merchant, you go to Amazon, you’re sort of giving everything to Amazon. It’s Amazon labeling, it’s Amazon’s website, it’s Amazon boxes. You’re literally paying Amazon to hold your inventory for you like it’s a pretty tough gig. Why do you do it? Because the customers are at Amazon. People go to Amazon, they searched for something. If you want to access to customers, you need to be there. It’s a great example of the aggregator concept
Where controlling demand lets you control supply. Shopify goes in the opposite direction in that it enables suppliers to have their own direct relationship with customers. They can set up their own store. They can get customers to come in and acquire things without having to have the middleman there and this is a concept. This is how you compete. I am online. I’m publishing things online. How do I succeed in a world of Google and Facebook? I go around Google and Facebook. I establish a direct connection with my customers. I send them emails. Email is an open standard. Just like the web is an open standard. That lets me sort of send my content to them directly and that lets me build a nice little business that goes around the aggregators as a word. That’s what Shopify enables as well and this is the way to compete with Amazon because you’re competing with Amazon in a way that can’t respond to it’s an orthogonal sort of competition cause your entire point is to support and feature suppliers.
The left side of the smiling curve to go back to our initial idea that are differentiated on their own now there’s lots of challenges on that side of the supply chain which are particularly because the customer acquisition issue, Facebook ends up eating up a lot of the profit there. But if you’re actually not differentiated and you don’t get customers via word of mouth or spreading and you have to go through Facebook advertising. Facebook is going to end up eating a lot of that up, but just the fact of the internet because competition is so high, a lack of differentiation is going to kill you either on one side or the other. That’s not necessarily a indictment of Facebook or Shopify. It’s just a reality that the competition with the entire world is your addressable market is higher than it has ever been. So will Shopify become like Amazon? I think that would be a mistake.
I think people don’t associate, the entire reason why people use Shopify is because people don’t know it’s Shopify, they know it’s, Oh, this shop with cool shoes that I like or the shop with cool knickknacks or whatever it might be. Whatever thing is on Shopify, and the brilliance of the Shopify model is people like to point out of the churn rate? Or we don’t what the churn rate is or how many people start shops on Shopify and flame out. It doesn’t matter. For Shopify, their platform is scalable because of technology in a way that a million people can start a Shopify shop and 950,000 of them can go out of business and Shopify now has 50,000 new customers who are generating that money. That’s a great place to be. Why? Because the cost of scaling to serve people because of technology, the marginal cost is zero, so it’s actually high churn on Shopify is a positive signal.
Not a negative signal and this is something I think a lot of people that aren’t familiar with the internet have a hard time wrapping their heads around this. Wait, you want high churn? Absolutely. That means you’re getting way more customers in at the top of the funnel. Now it’s going to be challenging for Shopify. I’m actually writing today, I think we’ll probably post before you post this, but about the Facebook shopping thing where they’re partnering with Shopify. Well the problem is that the gravity on the internet always flows to whoever owns consumer attention and so I actually think this is a natural place for the value chain to go, but it does show the weakness of Shopify, which is, I actually don’t write about valuations. Shopify is an Amazon competitor. It doesn’t follow that they’re going to in the long run ask for an Amazon valuation.
It’s a different model that can be a very, very profitable, very successful one, one that arguably has a better and greater impact on the world than sort of the aggregator model, but you’re not going to have the same sort of big picture scalability. If you think about where Shopify could really make a difference, to cement their position, it’s getting into things like logistics. Now, this last year sort of trying to get a three PL network and you say, well, it’s like Windows. Windows had the OEMs at the bottom and application developers on top. Well Shopify should have merchants on the top, all the infrastructure to support e-commerce on the bottom. Any one three PL provider, logistics provider can’t interface with a tiny mom and pop shop selling cookies on the internet, but Shopify can intermediate that so they provide scale in both directions such that you can make those connections in a way that’s very, very useful to both sides and also a very, very deep moat because you have that connection that you can’t form one on one.
An aggregator, I can always go around and form a one on one connection. It’s a very sort of the aggregator type of moat and this goes back to the regulation question an aggregator type of moat is very different than a platform moat. A platform type moat is all about APIs and drivers in a literal sense for an operating system, but in a broader sense. It’s about making these connections that can’t be made without the platform. An aggregator, I can always go around and connect with my consumers directly they succeed because of scale and sort of ease of use and it’s the obvious thing to do. To beat that you have to have something that’s easier and better or so highly differentiated that customers will go to the effort to go around it.
There’s an adjacent concept to all of this which you call the moat map. We’ve talked about parts of this. We’ve started really picking apart suppliers as differentiated versus commoditized, but there’s a second element here around network effects, which is a concept that’s nice that it’s taken us this long to get there. Usually it’s the central point of conversation in technology discussions. Describe what you mean by the moat map and where some of these companies might sit and also what some examples are like Uber that may not sit in the right spot on this conceptual map.
I actually thought it was an interesting concept. It was a hard one to sort of communicate, which probably means I haven’t thought it through sufficiently, but there’s this idea of basically what I just referred to where a platform there network effects are sort of externalized and so Microsoft built a network between developers on one side and users on the other. And basically OEMs on a third side, so a three sided network and you could argue all our like systems integrators and all those sorts of things on four sides. It’s like a four sided network. It was a brilliant business. It used to be a brilliant business. Microsoft has leveraged it into sort of what they’re doing today and they’re trying to build a OS in the cloud. A lot of this build last week was about this and the things that they open source, and the tools they made are all about making it possible for people to build on a business,
applications that connect to users that work on Microsoft services, et cetera, et cetera. And in the future where iOS and Android are the OEMs of the future, and Microsoft services sit on top of them, so that’s what Microsoft does. They’re very good at that. Apple’s app store is a similar sort of thing, network effect between developers on one side and users on the other, and so in this case the network effect is externalized. You go the opposite direction where the network effect of having more users becomes more internalized to the product. Facebook is an extreme example. Facebook, the suppliers are the users. The users go on, they write content on Facebook, they upload photos, they look at content and photos from other users and they’re all locked into the network and Facebook owns that entire thing and to the extent that others come in, it’s still defined by that Facebook sort of owning that sort of network there.
Same thing with Google. I talk about feedback, the data feedback that Google gets from people on the internet or on Google services that makes the service better and learning and they go on about machine learning and AI. All this is about Google harvesting the usage of users to enhance their services in sort of a feedback loop in those are network effects. It’s a type of network effects but it’s much more internal to the product. It’s not about tying together different pieces on the outside and I think that you get this idea where in that world where everything’s internal, the suppliers are totally commoditized. The websites on Google – Google doesn’t care which website it is, all they want is a particular information or fact total commoditization of websites. Facebook does the same thing. A story from the New York Times is given the exact same prominence as a picture of my nephew.
They’re just all in the same feed total commodization of content and I think that’s a characteristic of companies that have sort of internalized network effects where it’s all self-contained and they’re just diminishing what a supplier is. That differentiation, they’re squashing that left side of the curve. We go back to, we talked about before. It’s the opposite in the case of a platform, if you have external network effects where people come to your platform because they want a particular application, then it’s in your interest for that application to be highly differentiated. Having Photoshop on your platform, you want Photoshop to be more differentiated, not less because you want people to want to use that application and thus use your platform. And so the motivation and the way that the moat works for an externalized company with externalized network effects, they want highly differentiated suppliers because those suppliers are dependent on the platform.
So the more differentiated suppliers are, the stronger their own platform becomes. And you can see, again, this gets at why a platform, their motivations, the way they work, the way they’re defensible, the way they need to be regulated are just fundamentally different than aggregators on the other side, it just works totally different and it drives me up the wall we try to do it sort of the same way. So you mentioned Uber. I think a challenge with Uber is their network effects are externalized cause they have drivers and then they have sort of passengers, but there’s no differentiation. They’re trying to commoditize the drivers, and so they’re kind of in this weird spot where it’s sort of misaligned with the way these platforms typically work. They don’t keep people completely internalized, but also there’s no differentiation. And so that’s why they end up, they’re kind of oppositional to everyone.
Uber is very, very user friendly, but they have a very oppositional relationship with drivers and the drivers are, they’re always multi-homing. They want to be able to Lyft or maybe they’re all doing delivery with other services, et cetera, et cetera. And so when you get out of that sort of alignment, it gets much more challenging. And I think you’ve seen this with, when Airbnb was really humming, they had a much more sort of, I think, positive relationship with their hosts where the hosts were commoditized to an extent where you go on Airbnb and you saw sort of who was out there, but it was totally a sort of like internalized Airbnb. Again, it’s all matters of, it’s all gradient as a word, a spectrum, but I think the more that you’re highly tuned to one side or the other, the better your business is going to be and the less you need to. For example, like Uber, at least in theory, they should spend all their money on acquiring users and that should bring suppliers onto Uber to serve those users. It ends up that Uber actually spends a ton of money trying to recruit drivers. That’s a signal that they’re not quite an aggregator. There’s something that’s not quite right here and so part of this article is trying to figure out what is it that’s awful about this business model.
One thing we haven’t talked about, we’ve talked a lot about how the lack of friction on the internet just changes the entire dynamic of the world. We haven’t talked a ton about just pure constraints. There are lots of business strategy books written about wanting to control a choke point in a value chain. I’d be curious to hear your thinking as kind of related to everything else that we’ve talked about, about value chain thinking and sort of this difference between modularity and integration that you’ve written so much about and how that relates to good profitable business models.
Well, the big problem is you’re right, people talked about sort of controlling a point of the value chain, but it turns out in the analog world, that point was almost always physical distribution. So you think about you had the people watching football on TV where there’s a limited number of broadcast licenses. There’s a limited number of professional football teams, so you have sort of scarcity there. You have scarcity in the license, and the distribution over sort of the football game, you have commercials. There is a constraint on the number of commercials that could be showed based on pauses in the game there are, yes, football has a lot of stoppages, but there still has to be a game at some point. Or your sit-com show is 22 minutes long. There’s eight minutes of commercials. That’s it. There’s no more commercials after that. There’s 24 hours in a day.
That’s a constraint. You had to go to the store where there’s a constraint on land, how many big stores can there be in sort of your area. You go to the store, there’s a constraint on the amount of shelf space that there is. How many things can go on the shelves. There’s constraint on how much stuff can you actually physically carry in your cart or in your car to take home? And then of course there’s a constraint on money. How much can you afford to spend. Constraints up and down that entire sort of value chain. In that value chain that I just described was basically all the dominant sort of US Western companies in the era. You had the media companies, sports is a big linchpin for that. You had the car companies, you had the big retailers, you had CPG companies. All of them were defined
by physical constraints. All of those constraints are going away on the internet. So you can watch content on Netflix and Netflix goes on and on and on and on. There’s no constraint of content. Everyone can watch different content because you’re not constrained by time. There’s not, Oh, we have a limited block. There’s no constraint of broadcast channels. There’s an infinite number of things you can watch at any one time. You don’t have to fit in three channels or even on cable, a hundred channels or 500 channels. It’s infinite channels and you can start and stop at any one time, Oh yeah, we have 500 channels, but the movie is going to start at 8:00 PM no, the movie can start at 8PM for you at 8:30 PM for me and 8:25 PM for the other person. It’s just the loss of constraint of time. You have no more constraint when it comes to shelf space.
Amazon has infinite shelf space. It can hold as many products as possible. That’s why they have the merchant program because the merchants let them expand their inventory massively at zero cost to Amazon. It’s an incredible model. You have no more constraint on how much you can buy because you can sort of bring stuff in. There’s no more external advertising. Facebook and Google, the internet, generally, there’s an infinite amount of space to advertise. Now, realistically, Facebook has chosen to limit how much advertising is in a feed, but then stories comes along. Suddenly Facebook inventory explodes. Remember, people when Google first IPO’d in their early years would always focus on, Oh, the cost per click is decreasing. Our cost per add is decreasing. This is such a bad sign. It was totally analog thinking. No, the cost decreasing is a positive signal because it shows they’re increasing inventory faster than it’s being consumed, which means they’re actually increasing their longterm upside and same thing with Facebook.
Facebook stories come along. They have that earnings call a couple of summers ago it was like, Oh, our cost is going to go way down. These ads monetize as well and their stock plummeted and it was like, this is actually the best news about Facebook in quite a while because they were previously talking about we’re going to have a constraint on the number of inventory we have because we want it to be saturated or over- give users too many ads. Well, they certainly created tons and tons of new inventory. That’s a great single for them. That’s a very positive indicator for them. The price dropping per add is good news, not bad news. Again, this requires the shift to thinking about abundance, not thinking about scarcity and so all these parts of the chain, so users, users, now everything’s ad supported. It turns out the constraint of spending is actually lower too because users can consume basically an infinite amount of media whereas they had to pay for all their media previously and so in all these cases, the entire thing that under girded all parts of sort of our economy were based on scarcity.
In almost every case that scarcity has disappeared. The way you win on the internet is you be the starting place where people go to to start with Google, Facebook, these aggregators or you’re highly differentiated and you leverage the fact that you have zero distribution to reach anyone. Also, the internet allows drastically lower cost structures and you see this in publishing. It blew my mind last week where the Atlantic laid off a bunch of people and they have 80 writers on their masthead. And they had a staff of 350 well, it made sense in the previous world where you have all these support functions that actually made the money because what actually made money in publishing previously was not writing. It was delivery trucks and distribution deals, printing presses. That’s what actually made the money. On the internet though, you actually make the money through Atlantic subscriptions.
It turns out those writers, they’re no longer a cost center. They’re a profit center and they’re your only profit center that you have and they need to dominate sort of your roster of what’s on there and you have to get way more efficient on the backend and how you support them because everything else is a cost center. It’s a shift in mindset about what is profitable and what drives sort of meaningful profits in the long run, sustainable profits in the long run and has to get away from distribution and into how do you deliver sustainable differentiation.
Obviously you’re doing a version of this, but I’d be curious to hear your thoughts on the future of the media business model more generally speaking. You’ve started to hint at it there. That differentiated talent that people to get around the aggregator will seek out the absolute best writer or podcast or video producer, whatever on a given topic. Do you think then that that’s the model that it’s individual creators that go direct to consumer? Does it start to rebundle again as things always seem to do, what do you think about the future of media on the internet?
It has to be a lot of experimentation and trying for sure. I mean it is what I do. Basically being I write analysis about tech and media and sort of try to win on that regard. That’s honestly a really hard model. I’m not sure if it’s actually the best way to think about things, but differentiation does not necessarily have to be just analysis. Differentiation could be, I’m going to write about this specific topic that no one else is covering. You have to think about it as being a horizontal opportunity which lets you go very deep in sort of a specific area. Whereas previous publications were more broad based. Future publications will be very, very narrow and they’ll take advantage of the fact that their addressable market is not just New York or it’s not just the United States. Their addressable market is the entire world and I just need to get X number of subscribers wherever they might be and I will succeed.
So one example, I’m actually, I’m pretty bullish on and I’m waiting for someone to sort of crack this code. I don’t know that it’s been deeply tried yet, but is this idea of local journalism. I’m from Madison, Wisconsin. If I want to find out what’s going on in Madison, I could go to the Wisconsin state journal and it is an atrocious experience. There’s ads everywhere. It’s hard to find out what’s going on. There’s just wire stories that national news stories, et cetera, et cetera. And why? Why is it like that? Because they are a newspaper that has been around for a long time and their assumption is thatwe need to publish something every day and we need to get content out there that we could put ads next to it. And the motivations on this sort of version one media copies of the internet were the same thing. More content so we can get more ads.
It turns out though what I actually want. I just kind of want to know what’s going on. I usually go back there every summer. I want the news what’s going on. They’re building a new brilliant new convention center, a new hotel’s going up. What I would actually really like is someone on the ground to send me out an email every day saying, here’s what happened today. City council did this, et cetera, et cetera. And if nothing happened, then send an email that said nothing happened. That’s actually way more valuable me to get something that says nothing happened because I’m not looking for content to read. There’s so much content to read on the internet. I don’t need more content. What I’m looking is to feel informed and you deliver the experience of feeling informed oftentimes by telling someone nothing here, go spend your time somewhere else and you’ve actually checked that
that sort of mental box. It’s just a total shift in mindset about what you’re delivering. Trying to sell content on the internet is a loser’s game. It’s bits on a screen, zero marginal costs. What you need to deliver is a service, a service of continually making someone feel informed about something or continually delivering analysis they find valuable. A service is something that in the consumer’s mind is an ongoing sort of thing. In that case, the emails that I send four days a week are artifacts of the service I’m providing of my service to you is I’m going to think about the internet and media and the way things are changing and current news and I’m going to figure out and continually shift my thinking. And oh, I’m going to deliver to you a sort of summary of my thought process four days a week. That’s the service that I provide.
Do you think that’s valuable? You can pay me directly. You don’t need to go through Apple. You don’t need to go through Facebook. You don’t need to go to Google? We have a one-to-one relationship that by the way, thanks to the internet does scale very, very, very well. I do the same amount of work that I did a few years ago. But, I make more money cause I have more subscribers. That’s me leveraging the internet, leveraging technology to make a model that was not possible previously possible and Oh by the way, I have subscribers in like 85 countries or something because the whole world is my market. So many media companies view the internet as killing their business and it did. The ones that succeed in the future though, we’ll figure out how the internet makes their business possible in the first place.
It seems like there’s room in media for, I’m thinking about writing here, so I pay your annual fee and then I bought another one on sub stack, which is an interesting platform that enables writers as creators to create these sort of subscription business models. But then I very quickly got fatigued at constantly paying cause the subscription rates are pretty high for sub stack writers call it 50 a hundred $200 a year after four or five of these things I’m like geez, I’m paying more for this. Then this is rising my list of cost pretty quickly. It seems like a great place for there to be a bundle. A recollection of different writers that I could peruse. What are your thoughts there? Do you think that there are, let’s say for writing as an example room for someone to go try to build all of the demand, which doesn’t seem to have happened thus far?
For sure. There’s an aspect of this model that’s a bit consumer unfriendly to the exact point that you said, you have to manage them all separately. You have to pay for them all separately. It doesn’t scale very much with one customer. So first off, I would say the answer to that is there’s a lot of people in the world and the answer to publications is to find their specific niche such that the customers that are willing to go through the hassle and pay, and there’s something that it was your second, third, fourth subscription, then you weren’t their target customer anyway. And I think this is an aspect where people underestimate how big the internet is because I hear this subscription fatigue thing a lot and there’s sort of this assumption that there’s only about 10,000 potential subscribers in the world and how much are they going to pay again and again.
It’s like, no, there’s actually billions of potential subscribers in the world and a lot of them have different interests, will care about different things, so I’m a little skeptical of the big picture subscription fatigue issue as far as this model being a problem because I think it underestimates how big the internet is. That said on an individual basis, for sure it’s a thing. It would be better if there was a bundle you could pay X amount and get all these great independent writers and I’m a big believer in bundles. The economics makes sense for everyone involved. The problem is how do I get from here to there because if you wanted to make a bundle of say tech analysis, I would hope to think you would want Statechery to be involved in that. Well for a bundle to make sense for me, the person that would not subscribe to Stratechery directly, but would subscribed to the bundle, the number of those customers and their incremental value has to outweigh the value I’m giving up from my current subscribers by contributing them to the bundle.
Given that I’m in a strong position, I was fortunate to sort of be early to this space. That’s going to be a very difficult proposition, which means you probably have to pay me a bunch of money and then your economics starts to fall apart. You see this with Spotify, they’re paying Joe Rogan a lot of money because they want to him to bring his customers into what is their bundle for podcasting. For all intents and purposes, and so that’s just a real challenge. What you see with bundles is despite the fact that they make sense because they’re so hard to form, they usually end up forming for infrastructure reasons. A classic example here is the cable bundle, which did not form because people thought it’d be a great way to make money it formed because there’s people in like West Virginia or something that wanted to watch broadcast TV and their homes couldn’t get a signal cause they were like hollow in a mountain, and so they got together and they put up a big antenna and ran cable from that antenna to their homes so they could watch the big three broadcast networks.
And that’s something that more and more communities did this and at some point someone was like, wait, we have cabled everyone’s home. We could put extra stuff on this cable and we could actually show things that aren’t on the broadcast networks. And thus, you had community access television, which came along CATV into that transition to cable. And then HBO came along and was like, well what if we put satellites dishes at all these antennas such that we can broadcast one signal that is not over the air, but it’s actually something different. It goes to these satellites and then it plugs into the cables. It goes everyone’s house and boom. The cable industry was born in order to, actually, I think the T and T did it first and that’s where the cable industry came from that you ended up with a bundle, but the bundle came from the fact that you needed the cable and on the internet where everyone has distribution, it’s a lot harder to figure out what is the cable, what’s the thing that holds everyone together? Maybe sub stack given the fact that everyone’s on sub stack, they’re probably better placed to maybe make a sub stack bundle. I think that’s something they hinted at. That certainly would make sense in the long run, but that goes to the point that you need a way to sort of already have everyone in place before any one of those providers gains too much economic power such that you have to buy them off to get them in and your economics sort of fall apart.
We haven’t talked at all about what I would call a layer below platforms. I think infrastructure is probably the right word. A company like Stripe or Plaid that is providing tools to all sorts of different companies. Any thoughts here from a strategic standpoint? I’ve seen you write less about these companies, but I’d love to hear your thought on this sort of infrastructure layer of technology.
Yeah, well I think that that’s more of a platform layer is the way to think about it. I mean platforms don’t need to interact with the end user. AWS, most people are using programs that run on AWS and they have no idea that they’re using AWS. There’s a story a couple years ago, it’s someone saying I’m going to try to not use Amazon and they like IP blocked all of Amazon’s addresses and none of their applications worked. That’s a great example of AWS is a platform and that a platform does not need to interact with customers at all. It can just be the layer that stuff sits on top of. I think there’s an aspect where that’s a way to think about like Stripe. Stripe is a platform. Another way to think about it is it is just infrastructure and infrastructure requires massive amounts of investment to build, which provides sort of own natural moat because if you want to build sort of a competitor, Stripe has such a huge cost advantage relative to you trying to catch up.
That’s going to be very hard sort of to deal with. And obviously Stripe is sort of taking that and trying to layer up into merchant services into financial services, et cetera, et cetera. And I’m pretty bullish on that opportunity to do so. It’s a approach that is less about, it’s not actually at all about acquiring demand, it’s about having that superior cost structure that comes from scale and that is what gives you your moat. So it’s actually kind of living at the bottom of the smiling curve to go back to the example from before, but your dominant there because you have a superior cost structure and you have scale that lets you sort of provide services at a way that no one can compete with and it lets you spend on things like superior developer tools and superior documentation and all the things that Stripe is so great at because your costs for production of those things is lower than anyone else because you have such a large customer base.
A company that I haven’t seen you write a ton about, which I’d love your opinion on is Epic Games. They seem like both infrastructure and platform components. They’ve created amazing media. If you consider games to be media interactive media and they also seem very intent on behaving strangely and sort of reducing their own cashflow in a variety of interesting ways. They recently increase their floor from 50 grand a year before they start charging you to a million dollars a year as one example. Any thoughts on Epic as an infrastructure layer or a platform company or what it might represent in the competitive field that we’ve discussed?
I’ve written about them a bit. I think the gaming industry is fascinating and is in many respect sort of a leading indicator of where things are going. The whole sort of free to play versus free to win sort of concept is fascinating. So free to play is you can use a game and then you’re sort of hit a wall, you’re doing an in app purchase, to sort of unlock it. That’s been a dominant sort of model on mobile and there’s a reason why it’s going to make a lot of money. Whereas free to win is you can be fully competitive in the game. There’s no advantage accrued from pain. The only reason you would pay is because you want a different avatar or you want different clothing or you want sort of decorative functions that don’t or different fortnight famously different dance, the celebratory dance, which have nothing to do with your ability to win.
But do I have a lot of ability to do with your status and your sense in the game. And this really taps in the idea that why are these games better thought of as different worlds? I mean no one bats an eye at someone buying a luxury bag to sort of signify their status in the real world. And if you’re online and in these virtual worlds, why is it any different to buy a special dance to signify the same sort of thing? It’s not any different at all. Again, there’s sort of an insistence on the real world is different than the virtual world. The world is defined by where we are and where are sort of, we are mentally and a lot of people are spending a lot of time in these spaces and it makes total sense that we would monetize them in the same way you monetize in the real world.
So just broadly speaking, I think that’s one way to think about it. Another way is the benefits that come from being a platform that come from the broad base availability. And so from Epic’s perspective, I would imagine it comes to their engine or those lines is if you have it in your game, they make all their money, all their money, the huge amount of money from the huge winners. You think about like a VC model, you don’t really care about ones that you have a nice outcome. Yeah, it’s okay to get your money back, but you’re not making your reputation or raising your next fund based on someone doubling your money over five years. That’s actually arguably a worse rate of return than just investing in the big five tech companies in the market. You make your money off of the huge return and I think this is the case.
I think on the internet generally there’s this sort of returns to scale these exponential sort of returns and I think that’s probably a similar mindset that Epic is taking where actually we only really want to collect money from the games that go huge and the best way to maximize the number of games that go huge with our engine is to get our engine in more games, which means charging less upfront. Making it more attractive it’s the Shopify thing. The more companies that use Shopify and fail is a positive indicator for Shopify because that means they’re getting more rolls of the dice at that one that hits it big. Same thing with Epic. The more rolls of the dice, the better. Increasing the rolls of the dice is in a world of abundance, it’s all about increasing your rolls of the dice. It’s not about your failure rate because failing is cheap. It’s often zero. It doesn’t cost Epic anything. If someone uses an Epic engine and doesn’t pay Epic a dime, it doesn’t cost anything. It’s all virtual. There’s no marginal costs. So in a world of zero marginal costs, you want to increase your rolls of the dice as much as possible.
We’ve talked so much about the sort of bits and the rules of bits instead of the world of atoms and the transition between those two things. I’d love to kind of come all the way full circle now to the future and this notion that has become popular of it being a time to build that you wrote about Marc Andreessen’s kind of now instantly famous call to action saying that this is a key time for us to make big advances that we haven’t or maybe have failed to in the last couple of decades, but it seems like a lot of that might be much more in the real world, than the internet world. I’m curious how your thinking shifts over into that part of things, into the atoms rather than the bits.
It’s certainly a tough question. And sort of striking how the responsibility for this is immediately thrust upon the tech world. The entire point of technology in many respects is this idea of zero marginal costs. I mean the whole idea of Silicon Valley is chips, which are sand. The idea of a chip is that you spend a ton of money developing a chip, but once you start producing chips, the chips are basically zero marginal costs in all of the high price of the chips is either capturing profit or it’s sort of catching your R and D dollars and your infrastructure dollars and in that model we need a lot of money up front and then sort of infinite upside on the backside. That’s what venture capitalist is. Venture capital was formed around the economics of chip production. It turns out the economics of chip production are very similar to
the economics of software production, which is you spend a lot of money building out a product or a service. Then once people use it, it scales infinitely. For Epic to build a game engine cost tons and tons and tons and tons of money. For an additional user to use that game engine costs zero. This is the fundamental core economic principle of technology, and it’s the reason why so many people get the industry wrong because they don’t grasp that issue. Well it follows though that that is a very poor model for real world infrastructure spending, which often entails ongoing costs, putting things in. Now maybe there’s an asset and you build a road once and that over time there was a form of venture capital around canals and turnpikes and things along those lines. It’s not as extreme as sort of the modern venture capital model, so that’s one area where there is room for innovation is is there room for a type of investment model that is less sort of grand slam driven and is a more of a guaranteed return even though that absolute return might be slightly less.
I think there’s definitely need for new kind of sort of investment model to drive that sort of thing and maybe tech VCs can sort of help drive this, but I think VC is a very small part of the overall world to say that VC has to take this responsibility is kind of weird when your typical sort of fidelity fund manages as much money as the entire US VC industry combined. There’s a very weird focus on this specific sector that doesn’t actually manage that much money. Other things tech can do though is the work from home distributed work thing is very, very interesting and it’s interesting not just for the implications for the companies themselves. I think a lot of companies haven’t fully thought this through and there’s going to be lots of hiccups and they’re projecting experiences and epidemic to what’s going to be normal all the time.
And that might not work out well. But this idea of the more that tech spreads out and is less San Francisco centric and the more there’s driving competition between cities and between politicians and spreading the influence, I think we have a very positive sort of impact. Why can San Francisco be so poorly run? Well, all these tech companies are there. Well, there’s no competition. They’re not going anywhere. And so there’s just a general state of annoyance with each other and no actual change. Well, if it turns out we see this with an Amazon second headquarters and it’s framed sort of negatively for understandable reasons given Amazon’s wealth that, Oh, they’re setting these cities against each other. Well, yes, in a very narrow sense. Anything has good or bad sides. That was a pretty bad articulation of it, but this idea that companies should compete for workers by having better infrastructure, by having better capabilities is something that I think is an important one.
The third thing though is there is absolutely a political aspect of this and tech probably needs to be more involved in politics. There’s to what extent are a lot of the most innovative people in tech because they’re naturally interested in tech or because to build a service online doesn’t encounter any regulation and new challenges. There’s not going through 5 million committees to build sort of a road or whatever it might be. There’s certainly a big issue there and just how hard it is to get stuff done in the real world. Why would you spend time doing that if you could just do stuff online and not face any obstacles and that’s going to be a more challenging one to sort of fix? Certainly one that when you think about broadly, there’s also I think more room for industrial policy. I think the sort of globalization that was unlocked in part by the internet in a big part by the internet, but not just the internet.
Also things like 747 is that could fly over the ocean and carry a lot of freights or containers being a standardized shipping thing. All those sorts of things drove towards this world of extreme efficiency that it turns out if something hits or there’s an issue ends up one being a problem as we’re seeing currently, but also shifts where the power is and this idea that opening up to China would allow liberalism to flow from the West to China, well turns out it’s a two way street and Chinese authoritarianism can flow out where they are demanding a NBA GM be fired because something he said on Twitter, which has supposedly abandoned China, what are we doing here? China wants to ban Twitter. That’s fine. Now they’re reaching into the US to say what we can or cannot do and that requires a collective response, which is a government response, which may be industrial policy to encourage production if not in the US maybe in the West broadly.
It is funny because it’s like the Trump administration on one hand, good to stand up to China. On the other hand, it would sort of be much more cogent to meanwhile be forming deals with your allies with like Mexico for example. We should be trying to make Mexico into the next China. There’s clearly not sort of a holistic thought about this, but it is certainly I think good generally that we’re taking steps that direction. But there’s gonna be lots of pieces and it’s not just Tech’s responsibility, but tech can probably do more as well.
Over the years covering so many different companies. What company have you learned the most by investigating?
I mean, it’s a tough question to answer. So I’ll go with the one that popped in my head right away, which is Netflix and the way that they’ve shifted their strategy over time and in every step they’ve made shifts that increase their embrace of internet assumptions if that’s the way I can put it.
So starting out with DVDs and no more late fees. Well that is an idea of unhooking from infinite inventory and also sort of time’s no longer a constraint and then shifting to buying the catalogs from other folks. Well, when they bought that catalog from Starz, famously, what was interesting is there was I think 11,000 movies in that catalog. The effective number of movies on Starz was one, which is whatever was showing on the Starz channel. Whereas the effective catalog size on Netflix was 11,000 so they actually immediately had a superior product on a new dimension that never even occurred to Starz because they were thinking internet assumptions. You could actually stream things. And then the shift to sort of acquiring first buying shows that they owned, but then producing their own shows. All of these are predicated on this idea of zero marginal costs, the power of scale, being able to stream to the whole world with no increased costs. In every step,
they’ve been ahead of their competitors and their suppliers. They’ve gotten superior deals because they understand how the internet works in the way that’s fundamentally flipped. So many assumptions on their head.
My closing question for everybody is to ask you for the kindest thing that anyone’s ever done for you.
The one that sort of pops in the head, I think I’ve told this story on other podcasts before was when I first started Stratechery I’d been doing it for sort of a month or two before I sort of reached out broadly cause I wanted to have a sort of bit of a catalog there. I’m a big believer that for a publication, the most important story you write is the second story someone clicks on in that, yeah, anyone can write one good article, but they go to your site and are like, Oh that’s a good article, and they start clicking around.
If the second article they click is also really good, they’re like, Oh, well maybe there’s actually some degree of consistency here where I should to sort of be connected with this. So I sort of had this in mind. I wanted to happen. I just want to send my sign to someone with one article. Oh, please read my blog. Well, there’s one article here. Is this actually worth my time? So anyhow, a few months in I emailed a guy named John Gruber who does Daring Fireball, mostly focused on Apple and someone I’ve respected a lot and he’d been, and he had been running, been independent since the mid two thousands and always looked at as, that’s what I want to do. I want to be an independent writer on the internet. Be great. Reach out to them and say, Hey, I read news site. You should check it out.
Listed a few articles. Didn’t hear anything back. Maybe a month later I got an email from him pointing out how I misused a word. He gave me the etymology of the word. Well, I used to make all you make the same mistake. I actually made a shortcut, so I’d always make auto-correct itself, et cetera, et cetera, and that was the end of the email. But this is knowing people who know who John Gruber is. This is very unsurprising email to get from, but I figured, Oh, he’s probably going to link to a piece. It’d be very exciting. Well, he didn’t just link to a piece. He actually wrote a full article in the beginning. He was like, Oh, this is the best new blog I’ve seen in years. Read this article, read this article. Glowing praise, and at the end he says, well, for the first time I disagree with Thompson.
Then he spends like a thousand words saying why I was wrong about something, but that endorsed it. At the time I had about 500 Twitter followers just to use as a metric. I went 500 to 1500 overnight. Since then I have 140,000 more, but it’s been like a just a slow and steady progression. That was as far as the step change goes, that was by far the biggest and that is really what kickstarted sort of the site. And I’ve always been sort of deeply appreciative and grateful for that. It’s funny now six, seven years on, we actually just started a new podcast. So I do Stratechery, obviously which you’re aware of. Four days a week you’re going to consume it. Either via an email, which is the way it’s always been or via a podcast which was a new offering. Me and John are actually doing a new podcast two days a week, 15 minutes per episode just to sort of explore what does it mean to monetize a podcast. And so it’s subscription only, which is a new idea, but it’s kind of cool that he did sort of this great favor for me and now seven years later we sort of get to do something as equals. It’s very gratifying in that regard.
I love the bookends of the story. It’s such a neat concept and so many of these answers are about helping somebody early on in their career or their project or whatever it is. And I love that example. Thank you so much for your time today, Ben. I’ve learned a huge amount from your writing over the years. I’ve been trying to get you to do this for years. I’m glad that we waited because there’s so much more to talk about. I really appreciate your time 12 hours ahead on the other side of the world. I’ve loved the conversation and I love your writing. Thank you.
Well, thank you. I appreciate it. And I’m a big fan of the podcast and it’s a real honor to be here.