Simone Cicero

Hello everybody and welcome back to the Boundaryless Conversations Podcast. On this podcast, we meet with pioneers, thinkers, and doers and we talk about the future of business models, organizations, markets, and society in our rapidly changing world. Today I’m joined by my usual co-host Shruthi. Hello Shruthi.

 

Shruthi Prakash

Hello everybody.

 

 

Simone Cicero

Thank you. Thank you for being with us today. And we’re also joined by a three-time returning guest. He’s actually the first guest coming three times on the podcast. Nothing less than a legend of platform ecosystem stinking. Dear friend, we have known each other for more than 10 years now.

 

Sangeet Paul Choudary. Sangeet is a globally recognized thought leader in platform economics and digital markets. He authored the two bestselling books, Platform Scale and Platform Revolution, and he’s also working on a new book that is distilling in his newsletter that you will get pointed at at the end.

 

He has been featured on leading platforms such as Harvard Business Review, The Economist, and the World Economic Forum. He’s also the founder of Platformation Labs, advising global firms and governments on platform strategies for more than 10 years. Hello, Sangeet, it’s super incredible to have you here back for the third time.

 

Sangeet 

Thank you, Simone. So excited to be here.

 

Simone Cicero 

Thank you so much. OK, so let’s jump into the topic right away. So the idea of this conversation between us started now several months ago as you were working on new ideas for a new book that you’re working on. And we’re all excited to wait for the release. And the main, probably the core, the main idea, let’s say, behind the book is this idea of what you call sandwich economics, which is a very fun name. 

 

And I think you have this talent for choosing book names. But besides the funny name, I would really love for you to just introduce our audience to this idea and why you moved from the platform ecosystem scene to sandwich economics why this thing applies in this very broad way across industries, and also maybe to what extent it’s differently plays out in different industries. So give us an idea of what sandwich economics is and why it’s important.

 

Sangeet

Yes, Simone. So the idea of sandwich economics is essentially zooming out of thinking about what’s happening today in terms of platforms and ecosystems to really thinking about how is competition fundamentally changing. What is a new form of competition that’s emerging because of these players? Very often when we think about platforms, we think about the distinction between platforms and what I call pipelines. So platforms versus traditional product-industrial firms. And we look at it within that context only. 

 

When we think about ecosystems, we think of the framing of, well, they’re the focal firm at the center, and then there are complementers and other players engaging across them. You could use the platform. You do not necessarily need to use the platform. You have other ways of coordinating ecosystems, contracts, licensing, patents, et cetera. 

 

But in all these things, my view is that we’re missing a larger picture of how competition is fundamentally changing. And that is what I think about in the idea of sandwich economics. My key point with sandwich economics is that we’re entering an era of competition where firms actively strategize, a few firms actively strategize to dominate not just one, but at least two, possibly three or four layers of the value chain, move all the profits and power into those layers, and then they sandwich the rest of the value chain between these layers. 

 

So when we think about platforms, for instance, very often we’re thinking about dominance at one layer. We’re thinking of, let’s say, Uber, for instance, creating a dominant position between the driver and the traveler. But when we think about sandwich economics, we’re thinking about actively strategizing a future view of the value chain where you dominate not just one layer but you dominate multiple layers. The critical condition is you dominate at least two layers at the industry scale and then you you find creative vehicles through which you are able to co-opt other dominant players into that system. 

 

I’ll give an example to illustrate this because I want to talk about the economics of how this plays out as well, but it probably comes to life better if I give a story first. So think of the telecom 

 

industry. The telecom industry has been a vertically integrated industry for the, you know, since we’ve known it, right? And then we had two phases of horizontal domination at a particular layer. So traditional telecom companies were customer services, operations, and then infrastructure. 

 

And then we moved into these two phases of shift where first we had the key source of value, which is the application that drives performance moving away from the telco to the handset manufacturer. And then eventually to just two companies, Apple and Google, who because of 3G connectivity and cloud hosting, were able to aggregate application performance at scale. So initially, apps got unbundled from telcos and moved into the phone. That’s when Nokia became powerful. Then, apps got unbundled from the phone, moved to the cloud, got re-bundled at the operating system and the app store. So that created a dominant layer. 

 

Then we had a second dominant layer getting created on top, which fundamentally disrupted the economics of Telcos because the economics of Telcos was invested in infrastructure. And once you’ve got those fixed costs at scale, make money on the SMS and data traffic. And we had WhatsApp, Skype, and others coming in and creating this layer on top, particularly WhatsApp, which sucked away all the profits, and SMS deleted profits from the Telcos. So what we saw was we saw this inadvertent sandwiching between WhatsApp and Apple and Google at the other end in terms of where the apps were stored. 

 

So to some extent, the profits moved away from telcos, but I still would not call it a strategic sandwiching approach. 

 

Moving forward a few years, what happened in India in the mid-2010s is the perfect example of how sandwich economics plays out. So in the mid-2010s, Reliance, which is the largest conglomerate in India and has roots in oil and gas, decided it wanted to enter the telecom industry. They invested close to $40 billion in rolling out 4G fiber connectivity, and fiber networks across the country. And they did that with debt. And they were able to create a nationwide new connectivity infrastructure. So that’s single-layer dominance with a 4G spectrum. All players are still at 3G. They are now doing 4G. 4G fundamentally allows you to run calls on data as well. 

 

And so they then came out with the cheapest 4G smartphone globally, if you will, which they provided at a very cheap price across India and amassed more than 400 million users in a very short period of time. So what ends up happening then? You basically collapse the cost of voice. You make it extremely cheap to be a telecom user. So the rest of the industry moves into a price where there’s rapid consolidation. And very soon, we’ve gone from 10 plus players to a top three or four players, which remain at this point in time. And of all of them, Reliance really is the most dominant one. 

 

Now, that’s not interesting enough. That’s traditional classic Monopoly games. Upward the economics, the move competition, dominate a particular layer. That’s great. 

 

But what happened in 2020 was super interesting. Reliance then said this is cool. Let’s now think about sandwich economics. Let’s think about the entire value chain. We can’t build the world’s best operating system. We can’t build the world’s best aggregator on the consumer side. We can’t build the best consumer services. Let’s create an investment vehicle and ask Google and Facebook to invest. And so Facebook now owns the aggregation layer at the top with WhatsApp. Google has the Android operating system. Qualcomm and Intel have invested. They’ve got all the power with patents for 5G. 

 

So essentially, Reliance has created this investment vehicle that allows consolidation of all the power and profits in that value chain with itself. And what then ends up happening is now you’ve created this mechanism where Reliance has 400 billion users on its phone. All of them can be moved on to WhatsApp, which can be a layer that commerce can also be enabled with the new, you know, UPI and other infrastructure in India coming in. 

 

And with all of that, you can start sandwiching entire value chains. I mean, a great example of that is what’s happening in agriculture in India, which is one of the biggest spaces in India, where Reliance has a comprehensive detail infrastructure. It’s got WhatsApp through which it can do a lot of, you know, with a lot of market activity already happening at some of these village marketplaces in India. And so it’s kind of created that sandwich which can squeeze power than profits from multiple value chains now, not just the telecom value chain anymore, but anything in the digital economy can be kind of squeezed in between these layers that they’ve kind of set up. 

 

And I’m happy to talk through the economics of how this plays out, but that’s a really good example of a company that proactively thought about it, and more interestingly, did not ever create a platform itself. Reliance is not a platform player, but it’s a sandwich economics player.

 

Simone Cicero 

Is it just to clarify for our audience, for those who are listening, so how is this different from true, let’s say traditional ways to capture value markets, which is vertical integration. So of course, you know, when you vertical integrate to capture most of the margins or the tendency that we have seen more recently from platforms, as you said, they started mainly doing aggregation of demand supply and then we suddenly saw a kind of progressive rebundling of other pieces. So for example, FinTech integration or capital or demand fulfilling actually the demand, so sorry, playing the supply side role essentially, or fulfilling as I said, or other capital intensive elements of the value chain that they progressively re-bundled and provided to the customer. 

 

So how is this sandwich economics different from the traditional playbook of integrating vertically so that you can capture more margins?

 

Sangeet 

Yeah, actually, that’s a great point. So if you think about competition law in general, a lot of competition law prohibits horizontal mergers, right? Because you’re essentially consolidating power at a single layer. But they don’t necessarily, I mean, traditional competition law never looks at vertical mergers very badly, because they think it’s actually going to provide greater, it’s going to reduce cost because you’re going to gain the benefits of vertical integration. So. That’s not really seen as anti-competitive. 

 

Now what’s different this time around is traditionally vertical integration was about integrating across different parts of the value chain to actually at the end of it, deliver the better outcome in terms of the product to the consumer and do so at lower costs. What’s different this time is that you are doing first, you’re doing selective vertical integration, which does not necessarily lead to a better product at the end. You’re selectively vertically integrating only across those specific positions that are going to concentrate all the value. Secondly, these layers are horizontally dominant at the industry scale. We’ve not had that in traditional industrial settings where you had vertical integration and you had four to five players dominating the industry, but they were siloed vertically and they offered the best products within there through that vertical integration. What’s happening over here is you could actually end up with much worse products because the players creating the products are getting sandwiched in between.

 

So the benefits of vertical integration go away, but the power of vertical integration stays. And that’s fundamentally different from the bargaining that we had agreed to in the industrial economy, where we were willing to give the power of vertical integration to companies who are willing to give us the benefits of vertical integration in exchange for that. Now, I mentioned we use selective vertical integration. So you could still gain some of the benefits of vertical integration, but that again happens in a slightly anti-competitive way, because these players who organize either dominant market-making layers or dominant infrastructure layers, can get into product development as well. 

 

Amazon getting in as a retailer, this is one example. But that is again, very anti, it is done in an anti-competitive way where you are constantly removing competition within your specific product vertical by virtue of the fact that you have market-wide visibility through your infrastructure position and you have market-wide control over distribution through your aggregation position. 

 

So because of all of these factors, my one-line summary would be we are moving in a direction, and this is why I say it’s a fundamentally new form of competition. We’re moving in a direction where the benefits of vertical integration go away, but the power of vertical integration accrues to a few of these players dramatically. So that’s one piece. The second piece that’s important over here is that platforms do eventually, I mean, some platforms move into sandwich economics. But as I mentioned with the Reliance example you don’t need to be a platform to move into sandwich economics what’s important here is that some platforms do move into sandwich economics, but they very often are you know, they take a core position, and then evolve into additional layers over time.

 

What is different about you know starting with a sandwich economics lens is that you start with the position of the value chain You say this is where value is going to gather and you start moving proactively at gaining those positions upfront. And while some of today’s platforms have not done it in the core markets, they are increasingly doing it in other markets. So if you look at Google trying to get into healthcare, it’s following a similar strategy where it’s creating joint ventures with different forms of diagnostic system manufacturers to create consumer health devices for the home and connect that back into Google Assistant and connect that back into Google Cloud, get all the data back in, connect it back into telehealth systems so that eventually you have a layer across in-home care for elderly and for chronic patients. But at the same time, they’re using DeepMind at the back end to create an infrastructure for physicians. And eventually what happens with this is I know health care is a tricky industry. Regulation might come in and prevent too much sandwiching from happening. But if it were to play out without sufficient backlash, you’re going to concentrate more of the diagnostic power into your infrastructure and provision that back through the Google Assistant at home to the consumer. And in effect, you’re going to take away a lot of the value of sitting in between and being the actual physician in a variety of roles. 

 

So these have long-term consequences. If you start taking the salary, and the power in terms of learning effects into your infrastructure, does that give sufficient margins to some of these companies to invest in R &D, the ones who are in between and providing the services? Or does the whole future R &D for health care now sit with a few companies as well? So in the long run, hollowing out of the middle creates a lack of competition creates lower interest in or lowers the returns on innovation. A lot of those factors are playing out. That I believe is fundamentally where sandwich economics is different. And I’ll make a final point over here. 

 

I would say that Amazon, I want to make a distinction between two things. We could say that Amazon is powerful because of network effects. And that’s more you know, more merchants coming in, more consumers, more consumers coming in, bring more merchants. But if you look at the data, till five to seven years back, the average take rate on Amazon was 15%. And that was the take rate that they could charge on a network effect. Today, the take rate on Amazon on average is more than 45%. Whereas the network effect hasn’t really improved. What has happened is sandwiching. So what Amazon requires you to meet certain levels of sales in order to get the default position on the buy box. Now the only way to get that position, and that level of sales is to also invest in advertising with Amazon. So suddenly your take date has gone from 15 % to 30 % because you’re giving an additional percentage into advertising and that is, on average, that’s what it’s amounting to. Again, another way that Amazon does sandwiching is to enforce vertical integration, where it says, if you want to sell products to Prime users, in order to be Prime eligible, you need to use Amazon Logistics. And so now you’ve shifted it from 30 % to 45%, because you’re now forcing them to not just use the marketplace, but also advertise and use logistics. So what Sandwich Economics does is it looks at what the key assets I have over here, access to a prime user, access to the buy box. These are specific assets. These are not market assets. These are specific assets created by Amazon on top of the market, and they are brought together to sandwich more power than profits away. So it’s no longer network effects. Network effects can give Amazon only 15%. This is what I would call, at best, I would call it power effects. You’re essentially charging the extra 30 % purely because of the ownership of these assets. Amazon Prime, buy box, Amazon logistics. They are not market assets, they are private assets put up on the market.

 

Shruthi Prakash 

This is the yeah, I’ll go before Simone on this one. I’ve written down some points, right, which I just wanted to sort of highlight as well. Some words. One was the fact that it is a dominant market integration and then anti-competitive, enforced vertical integration, power effects, which are maybe not balanced, and so on. Right. So a lot of these, at least to me, sound like have some level of negative connotation.

 

And you also mentioned about how the competition is changing. There are new ways of competition. So just wanted to sort of touch upon that to see whether, let’s say the values are sort of, you know, outwardly sort of decentralized, but inherently centralizing power in newer ways. And when you say, let’s say new ways of competition therein, what does it really mean? 

 

Sangeet 

So, you know, when we think about centralization and decentralization, I’ll probably take that point first. The Internet, originally, was supposed to be an infrastructure for decentralization because it was built on top of client-server architecture working through protocols. And what we see today, you know, forget platforms, but what we see today is that the dominant players on the internet, they call themselves market makers, but they are not just market makers. A good way to understand is this, right? And I call this the dual agent problem. If you are out searching for a house, if you want to buy a house, right? You go to the market and you find a buyer’s agent. Similarly, the person selling the house has a seller’s agent, and then there’s the market in between. That market is regulated by the land authority. It has other actors. But fundamentally, there’s a difference between an agent on the seller side, an agent on the buyer side, and a market in between. 

 

A perfect market-making mechanism would balance the buyer’s agent and seller’s agent incentives separately so that the seller agent is optimizing for the seller, buyer’s agent is optimizing for the buyer. Now, one of the biggest distractions of our time is that we call Uber a marketplace. Uber is fundamentally a buyer’s agent, a market, and a seller’s agent because Uber is helping you find a ride, it’s helping the taxi driver find a job and so there is no way it can positively solve for both those players together. It’s a bit like going to a real estate scenario where the buyer’s agent and the seller’s agent are the same person. And you might say it’s an efficient market because there’s throughput, but there’s no way you’re solving for their interests individually. The reason the real estate market, despite all its inefficiencies, is a better organization of a market is because if your agent is working on your behalf, there is some way that if the eventual transaction happens, the transaction has the world in your favor as well. 

 

So what we see on the internet today is what I would call a dual agent problem. And that is why it’s essentially, if you look at the inherent political architecture of platforms, we call it a chicken and egg problem. Why do we have a chicken and egg problem? Because of the dual agent problem. Because you can’t optimize for both sides together. So that’s fine when you’re taking off, but when you pick grow at scale, there are only two outcomes that happen. At any point, you’re either optimized for the buyer or the seller, or you’ll optimize for yourself, which is what’s happening right now. A lot of platforms, once they gain power beyond a certain point, they just start extracting because they’ve reached that level of success. And the only way around it, if I go back to your point about decentralization the only way to gain the benefits of decentralization is to unbundle the buyer’s agent from the seller’s agent. That’s also why it’s impossible to regulate some of these platforms because you can’t regulate somebody who’s both a buyer’s agent and a seller’s agent. You can’t necessarily prove that this is anti-competitive because you don’t know what the… It’s a bit like saying Amazon is not increasing prices for consumers. So how can it be anti-competitive? Well, it’s acting like a buyer’s agent. So you can’t think of it as a traditional producer.

 

You have to think of it in terms of the effect that it’s having on the seller side and vice versa. So I think my key point is that the decentralization narrative definitely doesn’t play out in the platform economy, not because these platforms have just become highly scaled up. But even if you scale them down, as long as you don’t unbundle the buyer’s agent from the seller’s agent, that does not really play out perfectly. 

 

The second aspect to this is that the platforms, you know, do another, I call these things the false bundles of the platform economy, right? They say that this bundle is good, but this bundle is not necessarily good. Bundling buyer and seller agents together is not necessarily a good thing. It will always lead to suboptimal outcomes for at least one side. The second thing that platforms bundle together is market-making mechanisms with active accreditation. So if you think of the financial services industry in the US, if you want to take a loan, the person giving your credit rating and the person giving you the loan is never the same person, right? So the credit rating is handled by different authorities that you lead it differently and the loan comes from the bank. 

 

If you go to Airbnb, the person finding you an accommodation and the person giving you the rating is actually the same player. It’s Airbnb. It’s the same body that is managing both things. So that creates an inherent conflict of interest because if I can tweak your accreditation around so that it increases throughput in the marketplace and it helps me gain a better valuation. I’m incentivized to do that. So these are the false bargains of the so-called, you know, concentrated platform economy. And this is sort of the other side of the sandwich economics thesis. I mean, I’m kind of looking at both things together. Sandwich economics, what’s driving that? And what are the fundamental structures of the internet and alternate structures of the internet that we should be thinking about if we were to move away from sandwich economics? But because you talked about decentralization, I just wanted to bring this out. The issue is not about Web2 versus Web3 and all of those kinds of things. The issue is all these false bundles in the platform economy.

 

Simone Cicero

So I have two things in mind at the moment. And maybe you can tell me what you want to address first, right? So there is one kind of soul of myself in the conversation telling, what are the patterns that companies and organizations can use to achieve this type of domination? So for example, in earlier conversations we had, you mentioned commoditization of complements or other approaches to rebundling that allow platforms and operators like that to kind of sandwich out to the others. What is the playbook of sandwich economics? And then on the other side, I have a question in me, which is, this sounds very much like a concept that Cory Doctorow has brought up recently, this idea of enshittification, right? So these dominant positions end up in making, as you said, optimizing for the owner of the system. And then everybody else has a bad experience. And I was thinking of two things. So how’s this happening? Competition was supposed to make things better, not worse. And secondly, who is the player that can subvert these dynamics?

 

So for example, recently you spoke about digital public infrastructure as a potential way to counteract these dynamics. So I leave it to you to maybe see how you want to approach these two questions.

 

Sangeet

Yeah.

 

Sangeet

Yeah, absolutely. I mean, these two questions could make a book. So, you know, let me take a little bit of time to structure. Yes. So I think let’s do it this way. I want to start by putting a framework first and then let’s apply that framework. Let’s use that framework to look at what Cory Doctorow says because what he says is one instance of sandwich economics. It isn’t all of sandwich economics, but it is one instance. When you look at sandwich economics in the creative industry, that’s what he’s talking about. 

 

But what he talks about does not explain sandwich economics in health care. It does not explain sandwich economics in energy. So the creative industries are a specific instance. They use specific technological and legal constructs. But the underlying economics are the same. And so I’ll use that. So let me start it this way. 

 

When digital technologies first come in, and hopefully today’s world, we’re not debating whether digital technologies are coming in. They’re coming in every part of the industry, no matter what the resistance, right? When digital technologies come in, they collapse transaction costs. So what ends up happening is that in any industry where activities were bundled together, vertically integrated together because of transaction costs, those start disintegrating. 

 

So the vertically integrated firm, even if we don’t have a fully vertically integrated firm, there’s always going to be pressure on any point of integration when data technologies come in. So the first phase in any industry is unbundling. There are two problems with unbundling. One is from a value creation perspective, and one is from a value capture perspective. The value creation problem is that unbundling leads to better point products, but it leads to poorer solutions because the coordination costs increase. In order for the end consumer at any layer of the value chain to find value, they have to re-bundle things on their own and the coordination cost of that is too high. Hence there’s an opportunity to resolve that coordination cost. The second thing is a value capture issue that unbundling does not capture value because in the digital economy, most value is captured by creating bundles where you subsidize something and you monetize something else. Literally, everything we use, there’s an aspect of which that is being subsidized. And so in order to benefit from cross-subsidization, you need bundles. 

 

So unbundling never lasts long enough. What then ends up happening is rebundling. But this time, it’s not the vertically integrated bundle out of production convenience. This time, it’s a horizontally integrated bundle because we’re solving the coordination cost for the user. So we know, that the core reason you can capture value is because you can solve the coordination cost for the user and then cross-subsidize components of that bundle for the user. 

 

So that is where we have the notion of end-to-end solutions. We have the notion of solving the whole customer journey. But that is where platforms come in as a locus of organizing this rebundling. Now, in order to do that rebundling, you need to have a key control point around which that rebundling happens. A control point is not necessarily a bad thing. For example, Cory Doctorow talks about choke point capitalism. When control points move from value creation to value extraction, they become choke points. But control points in themselves are not bad because in any system, you need some level of coherence in the solution and that coherence is created through ownership of control points in any system. It’s not about technology, it’s any system. And so control points are important. It’s just that in the digital economy, control points are built around what is scarce. So attention is scarce, data is scarce, and the ability to organize innovation at scale is scarce.

 

The ability to own licenses and IP at scale is scarce. The ability to own intelligence at scale is scarce. So these become natural points of control in the value chain. And so around these points of control, there are three key vectors of the bundling that come in, which I call aggregators, integrators, and infrastructure. I’ve written extensively about these. Aggregators come closer to the user. They aggregate demand at scale, and they aggregate user data at scale. So they create a bottleneck to serve the market. Facebook, Google, Amazon, all of these started out as aggregate. 

 

Then you have the infrastructure. The cloud computing players, but all the AI foundation models today, create production infrastructure for industries that scale. 

 

And then you have integrators in between who integrate a fragmented industry and create a single locus of innovation. So a Stripe, a Twilio, what Stripe did to payments or Twilio did to communication are examples of integrators, banking as a service platform, API integrators. Those are the integration points. Value is not just an integration, but by integration, you create the incentives for innovation because everybody comes to you, so new innovation will find a market.

 

Simone Cicero 

So they are more or less the ones that can kind of establish, I would say, an ontology.

 

Sangeet 

Yes, exactly. And that is exactly what allows them to create a horizontal dominant position as well because they organize highly fragmented structures around themselves. And so this is where we’ve come to a point of rebundling. We’re still at value creation. Rebundling is value creation. Where it starts getting into value capture is once you’ve re-bundled, you become more valuable by commoditizing your complements. And that’s a unique thing that applies to ecosystems and to the digital economy in a way that it does not necessarily apply to a buyer-seller relationship in a value chain. In a buyer-seller relationship in a value chain, you can extract greater value just through negotiation, through predatory contracting. But commoditization of complements is not necessarily predatory contracting. It’s a way to create value for somebody in that whole ecosystem.

 

And in order for it to do that, you commoditize complements. Let me take a simple example to illustrate this point because there’s a nuance here. Think of Amazon. Amazon uses two very different tools in order to become dominant. One of those tools is only helping Amazon and is not helping anybody else. The other tool actually helps the user. So one tool that Amazon uses to grow horizontally dominant is the price matching algorithm – It goes around the internet and if it finds a cheaper listing for the same product, it essentially asks the merchant to match that on Amazon. That’s just predatory contracting. There’s no value to anybody except Amazon. The other tool is the buy box. If you think of the buy box, as a user, you’re solving their search costs because marketplaces are not just about bringing buyers and sellers together. It’s also about collapsing search costs. And if multiple sellers sell the same thing, then the way to collapse it is to create a buy box. So the buy box – If there are 100 sellers for a particular product, the buy box will default to the best seller. Now, if it were to do that only from the point of view of the consumer, that’s great. And so that’s an example of commoditization of complements, because you’re now forcing sellers to compete with each other. They’re getting more and more commoditized. They are no longer that they have the agency to charge what they want to. And so the consumer benefits. But now you’ve created a tool to constantly commoditize the seller. Now you could say, well, it’s great. You give the best price, but you know what?

 

We are also looking at sales throughput and in order to get sales throughput you need to advertise with us. Now you’re starting to commoditize the complement through many many many additional tools around that single control point. And so the buy box as a control point started as a market efficiency solution. It reduces search cost but then it becomes a choke point because it starts over extracting value by virtue of its position over there.

 

So that’s the idea of sandwiching. Once you start converting these control points into choke points, you can then start commoditizing complements. You can start eliminating your competition using a price-matching algorithm, et cetera. So you’re simultaneously expanding horizontally, and you’re expanding vertically by moving profits in your direction without moving the risks or the activities or the costs. So that’s sandwiching. And then you have the final stage of this process, which is what I call stacking. You now stack. Now that you started sandwiching, you now do it at multiple layers. Either you do it yourself or if you are Shopify and you’re not Amazon, then you partner with TikTok and say, hey, you take care of that layer, I’ll take care of this layer, let’s partner together. So different mechanisms for sandwiching. Some companies may set up a venture fund and they may say, we are very good at the infrastructure, we can’t manage the aggregator positions, and we’d invest in many different aggregators. Many examples of that as well. And again, depending on which industry you play in. There are certain industries where you may not have a dominant aggregator layer. 

 

A simple example is to think of Ant Financial. When it started out in China, it started out as a platform business and it started out around the payments capability. But when it started moving globally, it realized that it could not gain a payments license in every country. So it applied sandwich economics. It essentially went to each country and invested in the leading payments player over there on the condition that in exchange for the investment, they would move their entire operations onto Alibaba Cloud. So that then creates sandwiching power. You own the, you have an investment in the dominant aggregator in the country. You’re creating, you’re moving your infrastructure into that country. And even though you can’t create a global payment system, you can create a global infrastructure underpinning all these payment systems. And that allows you to then start creating all kinds of, you know, control positions across the value chain. A lot of things happened in China post that because of which this strategy has played out slower, you know, Jack Ma had to pull back, but this is where they were going to 2020. And that’s a classic example of sandwich economics. So going back to your point, you know, how does this relate to what Cory Doctorow says? So Cory Doctorow specifically talks about, and I won’t necessarily talk about the enshittification part, which is actually closer to the dual agent problem because he says,  first platforms are good to their users, then they are not even good to their users. First, they’re good to their users, then they’re good to their advertisers, and then they’re only good to themselves. That’s a dual-agent problem. You are trying to be an agent to both, and so at some point, when you become too big, you end up becoming it to yourself. And then you finally cease, and then they die. I don’t know if I fully agree with that, because it’s not as simple as just because you get filled with low quality, you end up dying that easily. I mean, so far we’ve not seen that happen very easily. So my point about what he then says in choke point capitalism is that that’s a version of how sandwich economics plays out in the creative industries alone. And that’s a good example because creative industries are where this is played out at scale. But in creative industries, there are only two major choke points. 

 

One is access to the user, and the second is exploitation of essentially licenses. 

 

One is access to the user, and the second is ownership and exploitation of, you know, access and exploitation of essentially licenses and any mechanism of ownership. 

 

So an example being, in the music industry, you’re getting sandwiched between Spotify, access to the user, and the three music labels at the back end which own all the IP. In the publishing industry, you’re kind of getting sandwiched. 

 

Well, you could argue that in physical books, Amazon actually adversely impacted the publishers, and hollowed out the publishing industry. But in digital books, the publishers got their way back. Because if you really look at the terms and conditions of an e-book, you never buy an e-book. You’re only given a license to an e-book. Because you’re only given a license to an e-book, you don’t have any rights in terms of expanding or doing anything with it moving forward. And so that creates a significant choke point in how content gets distributed and to what extent the value of the content can accrue to its owners. So in creative industries, it’s a little different. You don’t necessarily have the integrator positions to the same extent or the infrastructure positions to the same extent, but you have this ownership of IP as a key sandwiching capability in the creative industries. And what’s ironic, especially in the music industry, is that one of those sandwiching layers was not created through innovation. It was not even created through investment. The three big music labels ended up consolidating that layer only because the others went bankrupt. So neither did they invest in getting that amount of power, nor did they actually innovate by finding the right musicians. So they’re essentially extracting without having really taken the risk to gain that.

 

Shruthi Prakash 

So I was listening to what you’re saying, right? I think one of the points that even Simone had asked previously. So when let’s say these ecosystems have multiple entities holding different levels of control points and so on – how is it that you see, let’s say the balance of power dynamics sort of holding out true today and especially when there is collaboration, interoperability and so on between these different, let’s say value sectors. 

 

And how do you see this sort of changing with time? Do you see, let’s say, new players coming in or the involvement of public bodies like in India? So how do you sort of see this being challenged?

 

Simone Cicero 

Where are the antibodies?

 

Sangeet

Yeah. OK. OK. There are three sources of antibodies, right? So there are the regulators, the competitors, the users – these are going to be these three sources of antibodies. And when I say competitors, I also mean complementers. They may not be direct competition. In fact, for some of the sandwiches, there may not be competition left anymore. So it’s all of these three players create alternatives. Now, in order to think about this, one thing we need to understand is that interoperability, or the lack of it is not a technological issue. It’s a legal issue.

 

The framework for interoperability already exists. It’s just that you’ve bundled something that should be open with something that you can naturally control through IP laws, because of which you can then defend the lack of interoperability. And the other thing that I just want to say is that, just like I said, you know,  we are in an era of selective vertical integration. We’re also in an era of selective interoperability. A basic strategy, if you were to work with a company building anything in the digital economy, the basic strategy would tell you to advise them to not be interoperable with competitors but to be interoperable with complementers, because that’s how you increase value without reducing value. So it is all about selective interoperability. Now, I think the idea of interoperability is not strategically understood until you understand the idea of commoditizing complements. And this is part of the issue I take with, you know since you mentioned Cory Doctorow, I love his writing, nothing against him, but he proposes interoperability and portability as a solution and that we should move in that direction. But it sounds good as a solution until you think about the fact that you need a strategic reason to move in that direction. You need a profit-preserving narrative to move in that direction. And the profit preserving narrative comes from the commoditization of complements. You will want to interoperate to the extent that you are also able to commoditize complements. I’ll take a simple example over here. If you look at it, this will eventually take us to digital public infrastructure because that’s coming at it from the other side. But if you look at industries that are highly concentrated, say, energy, where there are a few big energy players in a certain geography, or real estate, where there are a few big real estate players in a certain area and have most of the land data, or automotive, where there are a few big automotive players, right? Logistics. In all these industries, you benefit from interoperability. So logistics, for instance, the more you can interoperate, the larger the network you can create across the country.

 

And in these scenarios, if you want to really benefit from interoperability, you need to have a clear view on what is the control point that still allows you to sit in that ecosystem. And what are you commoditizing at scale? And in a way that it does not, that some of the parts remain profit-deserving. So let me take an example over here. Maybe let’s take logistics as an example and then come back to automotive if that makes sense as well.

 

But in the US, you have today a really big logistics player called Amazon Logistics. Now, before Amazon went into logistics itself, FedEx was sort of enjoying a nice sandwiching relationship with Amazon, where Amazon owned the aggregator layer, and FedEx and UPS provided most of the logistics powerhouse at the back end, alongside the postal service and others. Then Amazon went into Amazon Logistics, so FedEx suddenly realized it could not compete with that. So then FedEx needs to now think about its own sandwich. So I’ll start with the sandwich and then talk about how interoperability actually helps to sandwich and then go into thinking about interoperability as an antidote. So FedEx, what did FedEx do as a response to Amazon doing this? The first thing it did was it needed a partnership or needed access to demand site data because you can’t do e-commerce fulfillment unless you have deep demand site data on the shopping cart.

 

And apart from Amazon, nobody had that level of domination. So FedEx ended up buying a company called ShopRunner. ShopRunner is another horizontally dominant play, which essentially works with any brand out there, and it allows that brand to provide two-day delivery. So if as a user, you sign up on a brand’s website, ShopRunner will show up alongside the items that can be delivered within two days to your location. So ShopRunner essentially offers two-day delivery to all brands and so gets access to all of the shopping carts. So ShopRunner is the only player that can aggregate shopping carts at scale. And so it gets all of that data on the demand side. And then FedEx has all the logistics on this side. And then FedEx has increasingly been talking to smaller retailers who are working with ShopRunner to bring all of their warehousing assets together so that FedEx can combine its logistics capabilities with their warehousing assets, with ShopRunner’s data to create a nationwide alternative to Amazon. Now you would call that interoperability, but it is selective interoperability because it is built around the control points of ShopRunner’s data, which FedEx owns uniquely, and FedEx’s own nationwide logistics infrastructure. 

 

So, that’s an example where I would say, yes, there’s collaboration, there’s interoperability, but it is still selective interoperability. And the warehouses are getting commoditized. You would say it’s getting commoditized because now retailer A’s warehouses can be used by retailer B and retailer C because they’re now part of the same network. So some assets are getting commoditized, but it’s fundamentally commoditization of complements. The warehouses complement the logistics service, to the actual fulfillment. And so eventually all detailers will win. So in order for us to think about whether interoperability is possible, we need a clear framing of what is the value story and whether complements being commoditized. Interoperability will come to the extent that the complements get commoditized towards that value story. 

 

Now this is what happens, for example, in automotive as well. Because if you look at automotive, all the big automotive players will always compete with each other, but once in a while, they’ll come together to agree on a car vehicle data standard. And that they are essentially commoditizing a standard actually so that the rest of the industry can organize around it. That’s again, selective interoperability. Now, if you come at it from the other end, that’s what digital public infrastructure is. It essentially says we’ll commoditize something at scale and you can build many different value stories on top of it. So eventually my key point over here is that you cannot talk about interoperability unless you understand commoditization and its contribution to profit preserving value story. Because otherwise, you’re just trying to make the world a better place. But it sounds good, but you need a profit -preserving value story alongside it, unless somebody like a regulator is trying to break you up. And over time, I’ve become less and less enthusiastic about the regulator’s ability to break things up. I think even regulators need to understand the commoditization of complements much better to know where the, and even things like dual agent problems much better to think, to create the right framings over here.

 

Simone Cicero

So in terms of evolution, what do you see as a kind of next stage of this? So what is the picture of the digital market that we can think about in, say, five years or more?

 

Sangeet 

I think sandwich economics is going to stay here for some time. And everybody needs to understand sandwich economics, not just because they may necessarily make the sandwich, but because they’ll be part of somebody’s sandwich. So whether it’s good or bad is a secondary question. I think the answer to any question, like whether it’s good or bad, is it depends. It depends on what you’re optimizing for. It depends on who you’re answering that question on behalf of. And so my attempt with this is to clarify how sandwiching works. What are the tools of sandwiching? What are the tools of power extraction? How are they different from going back to your point from traditional vertical integration and so on eventually whether you’re the regulator or a firm or even a user think of it this way – Users are also getting sandwiched in ways that we don’t necessarily understand. 

 

I’ll give you a simple example – In traditional industrial scenarios there was a production ecosystem that would create a product, and then the product would go to the user and the user would consume them. There was nobody in the consumption ecosystem in the sense that once I took a toaster back home, it stayed with me. Nobody came and participated in that. The consumption ecosystem was very fragmented. How does sandwiching happen with users over here? One way is thinking of if you buy an Apple laptop, you buy a MacBook. MacBook says that as long as you keep getting it repaired with an Apple Center, we are fine. If not, then we’re not fine. So essentially, MacBook is saying, I’m going to sandwich you between the sale of the product and the ongoing maintenance of the product. So in a lot of traditional markets, you cannot tie the aftermarket services back to the sale of the product. But increasingly, that’s happening because we have digitally connected products. Similarly now, BMW says, you know what? You can buy the car. But Germany is super cold.

 

You cannot you have to pay a monthly subscription to buy the heating from us. So you’re going to buy a car but there’s a software lock on the heating capability and that’s a way to Sandwich you between purchase of the car and ongoing usage of the car and so users are increasingly going to get sandwiched and it’s a natural consequence of a connected economy. There are many more ways to extend the value chain and extend your dominance at one position into another position. There was no way for manufacturers to become equally dominant in aftermarket servicing, but now they can. 

 

And so that’s the reason why understanding sandwich economics is important. Because very often we say, well, I’m an Apple fanboy, it’s very nice, I only want to get it repaired at Apple. But do you really want to just do that? Or do you want to allow an at-scale competitor to come in and organize the aftermarket servicing? So that’s why I believe it’s important to understand the mechanics.

 

Simone Cicero 

So I feel like you’re also hinting that consumers, or in general users, will have a much broader role in setting the shape of the market in the future, more than just the competitors and the players and the regulators.

 

Sangeet 

I absolutely think so because I think users have an increasingly greater ability to do that. At the same time, technologies are moving in a direction where users can increasingly get more commoditized and more controlled as well. So you will have the ability to do that in a connected system. And again, you know, part of, I have a whole chapter that I’m planning to write on how users can gain the advantage back. So I absolutely believe so.

 

Simone Cicero

So very quick question before we jump into the breadcrumbs, if you don’t mind, and this may be a teaser for our next conversation I was fascinated by one of your latest posts on Substack recently, as you hinted that with the advent of AI agents.

 

we are seeing AIs much more capable of completing, and fulfilling, let’s say, full organizational roles and not just helping us with productivity and tasks. Why I feel it’s connected with what you’re saying, to some extent, me, and this is something I still need to understand and research, but for me, this sounds very radically at the same time, but intriguing and very kind of transforming, transformative for what an organization is. And at the end of the day, what you mentioned today during the conversation is organizations playing on the market, taking strategic decisions, and so on, creating value, especially in the knowledge economy and also beyond, of course, there are some aspects of the economy which are much more dependent on resources and production capabilities and so on. But as a recap, what I’m asking is, is there any connection that you already see between the shape of markets, the shape of societies, and the emergence of AGI, essentially, as a potential partnering, organizing, producing value, and interacting on the market?

 

Sangeet 

Yeah, absolutely. I absolutely do. I think the fundamental difference between humans and machines is that machines went through standardization and humans went through differentiation. All our life, we are trying to be unique, right? All our life, we are trying to do something that is uniquely us. Whenever we’ve been treated like machines, whenever we’ve been asked to work in a factory to turn us around, we’ve never lived to our potential and we’ve hated those jobs. So machines went to standardization and humans went to differentiation. And algorithms actually love standardizing. If you look at it just very tactically, if you look at marketplaces out there, Uber is like an extreme tailor-stick marketplace, if you will, it has converted a human into a fully substitutable driving resource. And so the human part of it does not matter anymore. All that matters is who’s the closest driving resource available. Now, the reason I think this connects back to artificial intelligence is twofold. One is that artificial intelligence in general, like any technology, commoditizes the skills, right? So it enables those with lower skills to perform higher. And that’s a good thing with technology. But where AI is different is that it’s also constantly learning. So the more that augments you, the more you learn about how to do what you do, and the more it takes away your ability to differentiate yourself over time. And the more it eats into your skill premium and democratizes it. Now, even that might not be bad in itself because you’re now saying, you know, we’ve democratized education, we’ve democratized all of society gets lifted up, that’s great. But then the problem is that when you combine the democratization of skills, and hence commoditization of skills with centralization of market making and work organizing mechanisms, this is where AI agents come in and marketplaces come in, right? If everybody had Google Maps, and everybody could drive. And that just meant more drivers. That would be fine. But when you combine that with private ownership of the marketplace, and Uber can now set the price, and Uber can determine your minimum acceptance rate, and so on, then it becomes a mechanism to sandwich you between the dominance of Google Maps and the dominance of Uber as a way to get the ride into you. And so eventually, the more you get commoditized, the more you can be managed by metrics using an algorithm.

 

So the same thing ends up happening to a lot more managerial or, you know, knowledge economy work if AI can do what maps did to the London cabbies, right? So if AI can commoditize what we are uniquely good at and make us more substitutable standardized pieces of work, then it can most actively allocate you across a market of demand. And when I say markets, markets exist everywhere today. A lot of HR software inside a company works like a market. You are connected to the right opportunity internally to the right incentives internally, the same way that markets work externally. So we’re combining commoditization technologies with private ownership of market-making capabilities. AI agents take this a step further. They essentially say we won’t just take you and allocate you to specific points of demand.

 

We are goal-seeking so we can also manage some of this work. We can manage it with AI agents, we can manage it with human input, but we are taking on the goal-seeking managerial aspect of it. So I think my key aspect point, and I know we are short on time, so I don’t want to get into how agents work. But my key point is that the organizational capabilities of AI agents and the fact that all the value accrues to the people writing that AI agent and the market-making capabilities of today’s algorithmic marketplace and the fact that all the value accrues not to the Uber driver, but to the data scientist at Uber who makes a million dollars a year. What ends up happening is this democratization of knowledge that AI is going to do is going to lead to a skew of all the value capture moving to the ones who are writing the agents or writing the market-making technologies. So that’s the key concern. And that’s where real sandwiching happens again because you are working on top of AI, making it better. And you’re working under the algorithm which is exposing you to work or organizing your work and you’re getting sandwiched in between the two.

 

Simone Cicero

Yes, that really looks like a challenging situation we find ourselves in with modern markets and emerging technologies. Shruthi, I think we have a few minutes if you want to dive into the last point that you raised in the background.

 

Shruthi Prakash 

Yeah. Yeah. I mean, I’m just curious where, let’s say, how the shifts are going to happen from, let’s say, or where the word falls, right? Is it democratization, or commoditization? Because I can see, let’s say, customers or users in some way, consumers being commoditized, compliments being commoditized, skills being democratized, and so on. So there are a lot of parallel sort of drawings. So I’m curious to see how that sort of plays out, you think.

 

Sangeet 

Yeah, yeah, I think, you know, personally, I feel centralization versus decentralization, democratization versus not, commoditization versus not as distracting narratives. The key narrative is value creation and value capture – You know the people who capture the most value keep going back to talk about how much value they are creating but if the two are not in sync and especially if you can today capture value without taking on the risks of value creation then there’s a problem. So I think what a lot of us need to understand is that most of the risks in today’s economy are taken by the people who are not capturing the value. It’s taken by the people who are below the algorithm and once the market infrastructure has been set and has been organized at scale. Yes, the marketplace creator does take the risk of starting from scratch, raising funding, and all of those things. But once they’ve set it up at scale, at economy -wide scale, the risks are constantly being radiated to those who are below the algorithm. So I think there has to be a clear set of framings to think about how we reallocate value between where it is created, where it is captured, how do we not just democratize value creation, but democratize the incentives for innovation and entrepreneurship. And when I say entrepreneurship, I don’t mean Silicon Valley entrepreneurship. I mean the ability to just say, well, I’ve got all these AI agents. I’m going to run my own business. At the end of it, I should not be worried about some market-making mechanism, determining my access to clients, and so on. There has to be a better way to regulate some of these pieces together.

 

So I think to me, it comes back to value creation and value capture. And the fact that when you are at scale, there’s a lot of value being created in your ecosystem where the incentive to innovate is entirely on the ecosystem. The risk of innovation is entirely on the ecosystem, but the value of that innovation entirely comes to you. And there are many glaring examples of how this happens. 

 

Amazon looks at brands and once they take off, it invests in them or it gets into their category. In all these cases, the risk of innovation has been managed by somebody else. The reason these players have such high margins is not just because they have an economy-wide scale, but because they no longer need to invest in R &D at an economy-wide scale. The economy is doing it for them, and they’re just extracting the cream of the top. So I think that’s really the key idea of reconciling value creation and value capture.

 

Simone Cicero

So before Shruthi helps us to touch upon the breadcrumbs, I think I had this idea in mind after this conversation that I want to kind of highlight and share that maybe to really create an alternative to this perspective, it would likely entail a different definition of value. So it looks like if we don’t take value critically, the direction is written. We’re going towards this type of automation and sandwiching and commoditization of the user and so on. But maybe as you appeal to entrepreneurship, which is a way to build value in a very, I would say, self-defined way, a self-set set of values. So I think this idea of building different theories of value, may be a way to escape this kind of self -fulfilling dynamics that you have described in the conversation. So that was just an idea that I had in mind.

 

Sangeet

No, I think so. I think we are entering an age where defining the problem is as important as acting on a solution. Because we’ve always had a bias for solutions because problems have been well-defined. But today, a lot of regulation and tech is not happening because the problems are not well defined, not because of a lack of action. There’s a lot of insufficient regulation happening. For example, what we’re seeing with the EU, right, just because the problems are not well defined. And if you use the wrong framing, you end up with the wrong solution. So I absolutely believe that in today’s world, thinking about what is value, thinking about even what is extraction versus just justified value capture, but thinking about more broadly problem definitions versus acting on a solution. And that is why. Theory is more important than it ever was because if you don’t know the right problem, you can act on all sorts of solutions and it will not make the world better.

 

Simone Cicero

Yes, yes, I really resonate with this idea of theory becoming much more important. I often have this parallel when we talk about, for example, AI and the capability that AI has to make everything, patch everything together as a universal duct tape. I feel like it’s our responsibility to decide what we want to agree on intentionally. So what kind of ontologies and meanings do we build together as a group?

 

And so that we don’t get caught in this automatic, let’s say, adaptation to generally defined values that don’t resonate with us. So I really appreciate these ending reflections. So Shruthi, maybe you can drive us through the closing.

 

Shruthi Prakash 

Yeah, I mean, as Simone said, right, we have the breadcrumbs section as we call it. So just wanted to hear from you if you have any suggestions on books, podcasts, movies, or any other recommendations that our listeners can gain from.

 

Sangeet 

Yeah, I think, well, I mean, my recommendations will change every week because there’s just so much good material out there and there’s just so much to consume. It’s incredible. But I would especially recommend just looking at different lenses to think about technology. There’s a good book called Power and Progress, which has an interesting lens about how most technological value accrues not towards progress, but towards power. I think that resonates very well with sandwich economics as well. An interesting book I read about techno feudalism recently. I don’t fully agree with the whole thesis, but that’s part of the fun that you are actually challenged to think about it in a different way that doesn’t come to you naturally. So it’s an interesting framing of how capitalism has been built on productive capacities and feudalism was built around just ownership of assets, and in that case, land. And essentially, platforms are creating those kinds of assets with their market-making mechanisms. And Amazon’s BuyBox is one example of that. So I don’t entirely disagree. So it’s very interesting framing as well. 

 

So yeah, I mean, these are certainly some interesting books. I wish there was more real insight that we could see into how decisions are being made at some of these companies. But a lot of the books that come out of these companies are written by journalists who tend to more eulogize what’s happening rather than think. Even if you understand how strategic decisions are being made, you can translate that to understanding how that relates to tower relations and how the company is thinking about power. So I think that’s a big gap in the market in terms of understanding the architects of what is happening. And to a large extent, the architects of these systems, and this is why I make the distinction between, it’s not just about acting on solutions, but also framing problems. The architects of the system don’t fully understand what they’re building. To them, it’s just code very often. If you give a 22-year-old straight out of college to write an algorithm that determines the future livelihood of a single mother who’s supporting three kids, I don’t think that person is, I would say, mature enough to understand the multifaceted nature of that problem. So I think that’s where a lot of the issues lie. And that’s the gap in the literature that we have today.

 

Simone Cicero 

Thank you. How can our listeners stay on top of your work and follow your work? I mean, of course, your newsletter, maybe you want to be a bit more specific.

 

Sangeet

Yeah, no, thanks for bringing that up. Do follow my newsletter. It’s platforms.substack.com. I am going to be writing about all of this over the coming months. To me, a book is just a mechanism to bring an idea out, but I believe in constantly creating a stream of ideas, getting out, and a book comes out at a certain point. Some people prefer the book. Some people prefer the constant stream. So I’ll be building this in public on that newsletter.

 

I’m also looking to start a series of reflections on sandwich economics and analysis-style commentary podcasts. And that should come out in the next month as well. So yeah, I will share those links so that you can pass them on to your readers.

 

Simone Cicero

We can’t wait for your podcast to come up, really, and book, of course, as well. So thank you so much. I hope you enjoyed the conversation and brought up some new ideas.

 

Sangeet 

Absolutely, I did. Thank you so much. This was great.

 

Simone Cicero 

Thank you so much, Shruthi, for your contribution as well.

 

Shruthi Prakash

Thank you, and thank Sangeet. It was great having you.

 

Sangeet 

Thank you. Thank you.

 

Simone Cicero 

And yeah, for our listeners, as always, you will find the full transcript and all the references that Sangit has been sharing during the conversation on our website. So boundarless.io/resources/podcast. You will find everything there. And until we speak again, of course, remember to think boundaryless.