In this issue:
One of the most powerful models in competitive analysis is the concept of commoditizing the complement. When a company dominates one part of the supply chain, the way it grows is by making it easy for new entrants to grow in adjacent parts of the supply chain, so the monopolist has a bigger market and more more negotiating leverage. This idea was articulated by Carl Shapiro and Hal Varian, popularized by Joel Spolsky, and explored at length by Gwern. Examples abound:
Google acquired a company that made analytics software, and changed it from an enterprise product to a free one. Now, Google Analytics is ubiquitous, and since Google's own ad products are close to the bottom of the funnel, the last-click attribution that Analytics promotes tends to increase spending on Google's own search ads.
Snap turned content creation into a tournament, which encourages new entrants; Snap's negotiating position is strongest when there's more fame to go around, but turnover among the famous is high.
Amazon wants to ensure that for any popular product category, there are enough sellers that the market price for sponsored search slots reflects the unit profit from selling the product. With one or two sellers, the sellers capture the economic surplus; if there are half a dozen, Amazon captures the surplus.
But it's important to note that commoditizing the complement is a continuous process that doesn't lead to a stable equilibrium. Every company that monopolizes a layer of its supply chains will leave a low-return dead zone nearby, because it absorbs the profitable parts of its supply chain and then commoditizes the rest. And commodity sellers slowly go under; if a business has mediocre returns on investment on average, and returns vary over time, it will eventually have a run of bad luck and go under.
And the result of that is that, over time, monopolies who commoditize the complement end up adjacent to one another in the supply chain, since there's nothing left. I expect this process to continue, and for more of the uncertainty in big tech to result from complex negotiations between sometime-adversary/sometime-collaborative big tech companies.
Bilateral monopoly relationships aren't the default, but they're not especially uncommon. For products that have high transportation costs and few end users, the buyer is a monopoly and the seller a monopsony. Lignite coal mines and power plants are a classic example. In media, cable TV companies have monopoly access to (some) homes, while content companies have monopolies on (some) content. And it happens every time a large company identifies a competitive threat from a small company. In 2012, Instagram was a monopoly seller of potential photo-sharing market share; Facebook was a monopsony buyer since no other company could justify paying the kind of premium that would convince Instagram to sell.
But the most interesting of these transactions is the agreement between Google and Apple on how to divide up the profits from Google's position as the default search engine on iOS. According to the DOJ complaint filed against Google last year, Apple earns $8bn-$12bn annually for setting Google as the default.
This deal is a uniquely important one: there are larger one-off transactions, like acquisitions; there are larger repeated ones, like treasury bond auctions. But the only transaction I can think of that is a) repeated, and b) contains so much economic uncertainty, is the Hacienda hedge.
The deal has economic uncertainty in each iteration, and over time. Holding Apple's and Google's capabilities constant, the question is how to divide up the benefits of 1) Apple's ability to deliver the attention of high-income iPhone owners, who are frequent users of their devices in part because of the extended ecosystem of iOS-only or iOS-friendly apps, compared to 2) Google's ability to monetize search traffic better than anyone on earth, and their need for continuous data on searcher behavior to maintain its lead. There are no obvious ways to determine who is adding the most value here. Both companies worked hard to make the Internet better for mobile, Apple in hardware and Google by making mobile compatibility and site loading speed important ranking factors before the direct benefits of mobile revenue really justified this. Both companies have worked hard to make their products more compulsively useful. And both companies have done a lot to monetize them to the utmost.
The second-order question is even more uncertain. If Apple were to make a different search engine the default, that search engine would not be the world's #1 search engine, but it probably would be the most-used search engine for English-speaking mobile users; English is the most-spoken language in the world (about 20% more speakers than Mandarin) and the most-learned language in the world by a much bigger margin (3.8x as many second-language speakers as the runner-up, Hindi). So this search engine would be quite valuable, albeit not as valuable as the iOS revenue stream Google currently gets. Apple has moved into businesses in which it lacks compelling economics in order to improve its strategic situation in the past; Apple Maps, for example, was a way to make the iPhone less reliant on Google's products. And Google has done exactly the same thing, using Google Fiber to bully ISPs into providing faster connections. In both cases, the decision was a poor risk-reward on its own, but led to a better negotiating position.
Google, too, has opportunities to exit the deal, mostly by reducing Apple's relevance. Any investment Google makes in better Android services, or any assistance they offer to handset manufacturers, pays direct dividends in higher search volume on Android handsets. But it also has the indirect benefit of gradually reducing the risk of losing the Apple deal. Google would be much happier if it were dependent on a collective of mostly commoditized hardware companies, rather than on a single company with a distinct vision for how phones should work—and of who should capture most of the economic benefits of their software.
It's a bit like a Cold War standoff: both companies have the option to strike first; both have the option of massive retaliation; each company is broadly aware of the other company's strategic options. And both companies have ample cash on hand and slowing growth in their core business. There are, broadly, four kinds of outcomes that make sense:
The least likely is outright conflict; Apple switches to serving its own search results, or partners with Bing on much more favorable terms. This would materially hurt both Apple's and Google's profits, at least in the short term, and would degrade the user experience. It's a hypothetical outcome both sides would like to avoid.
Apple and Google each invest escalating sums in making scenario 1 look less expensive and more beneficial to them; Apple continues to hire top Google search execs, so Google has to offer more attractive terms to retain them; Google does what it can to make Android a more attractive mobile OS.
Either Apple or Google gets ahead in the next form factor, and finds a way to generate search-like economics without sharing them.
Somebody else entirely does #3.
The second option is by far the most likely, but it's an unstable equilibrium. Both companies have an incentive to visibly invest creating in their ability to exit the deal or obsolete it through new products. Signaling willingness to walk away from a deal is a good negotiating tool, but it requires constantly escalating signaling to work. And at some point, the cost of the signals gets high enough that the deal itself can't hold. "Minsky moments" are mostly a financial phenomenon, but they can occur in any system where naturally increasing stability raises the expected returns of taking more risk.
These scenarios will only get more common over time. Consider the situation in chips; there are a few companies that operate at a scale where they can design chips for use entirely in their own hardware, rather than for sale to someone else—Amazon, Google, Alibaba, and Apple all have custom silicon. But there are only a few companies that can fab the most cutting-edge chips; TSM and Samsung can, to at least a limited extent, decide when someone else's new design actually gets put into practice. In e-commerce, Facebook sends a lot of paid traffic to direct-to-consumer brands, who rely heavily on Shopify to run their sites. Right now, Facebook and Shopify are cooperating, but at some point each company will have to decide how much value the other is really providing.
The policy question of what to do with powerful tech companies remains open. But antitrust cases always deal with snapshots in time, while the system they're responding to is dynamic. Breaking up big tech companies can work, but it's a race between the legal system and the game theory of tech companies doing it themselves.
Nic Carter has a good object-level explanation of what NFTs are, and why they're hard to understand. There's an extreme sense in which all the value in software is created by a) building a new zero-scarcity product, and then b) creating a way to impose scarcity on it. Part of the struggle with NFTs is that scarcity models are resistant to copying, since ease of copying eliminates the scarcity premium.
Meanwhile, Amy Castor has a more meta look at what drives NFTs: the headline-grabbing $69m Christie's auction for an NFT was won by a business partner of the artist, who operates a digital art fund. It's painting the tape, a way to get a widely-recognized high price for one asset, executed by someone who owns other assets in the same class.
That dubious auction result doesn't mean that NFTs will immediately collapse in value, or that they're not useful. Plenty of other markets have dubious behavior—certainly the art world has some purchases that make more sense as a way to move cash around than as an aesthetic judgment. But it does show that any new market is prone to abuse; marrying the speculative froth of cryptoassets to the KYC-light, loosely-regulated art business is a way to select for bad behaviors.
Bryce Roberts has a post-mortem of Indie.VC. Indie was an effort to make venture investments without the expectation of short-term liquidity; it offered companies the opportunity to buy back 90% of the investment at 3x the original valuation, and accepted that some companies would stay private for a long time. One of the major headwinds was GAAP accounting: when companies don't raise additional funds, their carrying value stays at the value of the investment. Roberts contrasts the returns in terms of carrying value to estimated returns based on comparable market values:
Until our investment converts onto the cap table, GAAP, and industry best practices, say we have to hold those investments at cost- no mark-to-market, no next round. Given we often hold these investments as options on equity for years also means we have to hold those investments at cost for years, even if the underlying companies are making tremendous progress... As of 12/31/20, the lifetime gross GAAP IRR for the Indie.vc strategy is 16%. In contrast, the gross "as converted" IRR is 48%.
In one sense, Indie.vc was tragically miss-timed: an illiquid stake in a valuable company that has no specific intention to provide liquidity to investors is basically a non-fungible token with a small chance of spontaneously converting into cash. With the right spin, Indie’s portfolio could have been valued by LPs at a premium rather than a discount to its real worth.
An interesting interview on the topic of local tech champions: as countries get more nationalist, and as they worry more about multinational tech companies eroding their sovereignty, it creates more possible winners. One case study:
Take Lazada, the privately-held Singaporean multinational technology company that focusses mainly on e-commerce and which was funded by Germany's Rocket Internet (ISIN DE000A12UKK6). For years, its core business model was flooding the Indonesian market with Chinese imported goods. Tokopedia, Bukalapak and some of the other local Indonesian e-commerce companies were frustrated because they were losing business if they tried to promote their local brands. This became an issue for the government of Indonesia. Over time, helped by support of the local Indonesian government, Tokopedia was able to successfully curate local products which appealed to local consumers and ended up taking a dominant position back from Lazada (although they are not technically #1).
While this increases the number of potentially investable Internet companies, it truncates the returns. Part of the valuation premium that the biggest companies get is that their marginal cost of expanding into one more market keeps dropping; if they're staying with just their home market, that upside doesn't apply. And close relationships with the government work both ways: a company that owes its position to protectionism will have to help out favored industries.
Tech Sees Like a State: Vaccination Edition
Facebook is releasing tools to make it easier to find vaccination sites. There are some great projects working on this, but in many areas awareness spreads by more ad hoc means. (A few days ago, a neighbor went door-to-door on our block, telling people that there were open vaccine spots a twenty minute drive away.) Surfacing timely information to specific people based on their personal traits and geographic location is something tech companies have gotten fairly good at, and they're increasingly able to apply this skillset to domains that don't directly impact their bottom line. Indirectly, of course, Facebook benefits both because of the PR value and because advertising is a levered bet on economic growth, so nudging the reopening a few days earlier is materially profitable to them.
Where Amazon Earns
Ben Evans speculates about the profitability of Amazon's ads business, which may produce more operating profit than AWS. It's an interesting question, but Amazon's reticent disclosure here is actually helpful: the ads business is a price-discrimination layer on top of e-commerce, so it's hard to meaningfully distinguish between retail and ads' contribution to total revenue. Like airlines and loyalty programs, the accounting profits skew to the asset-light side of the business, but the most cost-effective way to grow that business is to spend more on the lower-margin, more capital-intensive complement.