The Money Has to Come from Somewhere
On crypto's blitzscaling problem
(I want to begin by doing a bit of level-setting. I spend a lot of time reading the work of crypto critics like Stephen Diehl, Molly White, Jacob Silverman, and David Gerard. I’ve also become an avid listener to the Cartoon Avatars podcast, and have watched Dan Olson’s video essay, “Line Goes Up” more than once. I’ve learned a ton from these sources, and a highly recommend them to you.
If you already follow this crowd, then you can safely skip this essay. My argument is going to be very recognizable to you. If you don’t, then I’d encourage you to check out their work.)
There’s a passage in Max Chafkin’s book, The Contrarian: Peter Thiel and Silicon Valley’s Pursuit of Power that I keep circling back to. It’s about the early, sketchy days of PayPal.
PayPal further juiced its growth by appropriating [Elon Musk’s company] X’s aggressive tactics, giving $10 to every new user and another $10 for every user they referred. To sweeten the deal even further, PayPal allowed users to immediately withdraw any money they earned (…) It was an expensive feature, costing PayPal an average of $20 for every new user it added.
In November PayPal’s user count was a few thousand. By January, the [number] had risen to 100,000, and just three months later, it was up to 1 million. That was a more or less unprecedented rate of growth, even in Silicon Valley, but it meant that PayPal had spent something like $20 million on referral fees out of the $28 million raised so far. Early employees tell stories of walking in and seeing that thousands of users had signed up overnight—and feeling a sense of awe and terror. (pgs 59-60, emphasis added)
This is “blitzscaling.” It has become standard practice in Silicon Valley. You gauge an early-stage company’s success by its growth. Profitability will come later.
Seeing the numbers laid out plainly like this, what I get stuck on is just how bonkers-simple this strategy is:
Step 1: Raise ~$20 million from venture capitalists who like the cut of your jib.
Step 2: Pay people $20 apiece to sign up for a free account on your website.
Step 3: Congrats, you’ve got 1 million users. You’re a huge success! Bask in the accolades.
Step 4: Raise another $20M from VCs on the back of all that demonstrated user demand. Dole that money out and sign up another million users. Lather, rinse, repeat.
This… seems like it should’t work. Right???
Of course, it did work for PayPal. But you have to take another step down the value chain in order to see why. PayPal’s actual business thesis was something like this: “Hey, it’s 1998/99 and payments online are obviously gonna become A Thing. They are going to be a lot easier for merchants and for customers. And payment processors get to charge a fee on every transaction. If we grow faster than anyone else, we could become the de facto payment processor. Then, as the online commercial universe expands, we get filthy rich.”
When you put it that way, it makes a lot more sense.
The part that I find fascinating is how, in the rush to embrace “disruption” and treat startups as a unique engine of innovation and growth, the second half of that business thesis has been completely memory-holed. Venture Capitalists have mythologized the growth hacks. They value growth over profitability, forgetting the most important rule: eventually, the money has to come from somewhere.
If you forget to ask where the money is going to come from, the focus on early growth can mask some truly bonkers, never-going-to-be-profitable-are-you-kidding-me ideas.
(Like, oh just for instance, Axie Infinity. But I’m getting ahead of myself.)
Reading Mike Isaac’s book, Super-Pumped: The Battle for Uber, it’s striking just how heavily Uber relied on, and was rewarded for, blitzscaling. Uber “disrupted” the taxi industry. But it did so less by offering a better competing product than by lighting giant piles of money on fire in order to incentivize drivers and riders to start using Ubers instead of taxis.
Travis Kalanick’s crucial insight was that VCs would keep giving him more money so long as he kept demonstrating growth. Uber enticed new drivers by giving them free iPhones. It enticed new riders by setting ride prices so low that the company lost money on every ride. Disrupting an established industry gets a lot easier when your competitors have to turn a profit and you can spend unlimited cash and just be judged on growth.
We shouldn’t get too sentimental about the incumbent taxi industry. It was bad for riders on basically every level, and had no incentive to improve. But the point that bears emphasis is that taxis are a low-margin business. We don’t need to feel wistful for the olden days of bad taxi service to point out that Uber’s business strategy never made an ounce of sense! Cory Doctorow has made this point much more thoroughly:
The blitzscale approach worked for PayPal because of all the money PayPal could reasonably hope to scoop up later. For Uber, it only makes sense if the company can eventually make so much money off of ride-hailing services that it can turn a profit while also making up the $31 billion of debt that it built up through blitzscaling. That money has to come from somewhere. And that somewhere is bound to be a mix of overcharging riders and underpaying drivers.
We’re seeing the results in 2022. Uber has gotten more expensive and less dependable. The company has never turned a profit, except through little bits of magical accounting tricks. It’s entirely possible that it never will.
And the same is quite possibly true of Doordash, Grubhub, and the rest of the delivery-on-demand sector. When those companies were being judged based solely on growth, they looked like industry-transforming juggernauts. But (shocker), food delivery is also a low-margin business. As we start judging them on profitability, the foundations of the whole sector turn out to be made of soggy cardboard.
Many of the brand-name companies of the 2010s, in other words, succeeded because of an investor-class that had learned the wrong lessons from the success of companies like PayPal. Blitzscaling only works in limited circumstances.
Now comes the fun part. This is where I get to dunk on Axie Infinity, the video game that, for much of the past year, was every Web3 supporter’s go-to example of the radical new internet they were building.
There are basically three ways to make money from video games:
You can be a video game company. You can actually, y’know, create, market, and sell video games. There are a bunch of revenue models for video game companies — they can charge for the game itself (old school model), and/or they can charge a monthly subscription fee (World of Warcraft model), and/or they can charge for in-game customization goods (Fortnite model), and/or they can charge for in-game power-ups (gacha model).
You can be a YouTube/Twitch streamer. These are people who play popular games in front of large audiences who enjoy watching other people play video games for some reason. They make money through supporter donations, and online advertising. I guess the biggest ones probably have endorsement deals or something, too. (I do not understand the appeal. I am middle-aged and The Culture has passed me by.)
You can be a gold-farmer. These are people who work long hours playing the most popular games, earning resources or leveling up characters that they can then sell to richer people who have more money than time. Some games allow gold-farming, others forbid it (but grey markets emerge if the game is popular enough), and still others allow it for awhile, then crack down and forbid it if the player base complains.
Gold-farming sucks. The whole point of gold-farming is that there are countries where minimum wage working conditions are so bad that people can earn a better living through serial completion of rote tasks in a video game that lazy (or, um, busy) Americans like to play. The existence of gold-farming is itself a biting social commentary on late capitalism. It is not a feature; It is a bug.
Web3 boosters decided to make it a feature. They decided the best way to fix video games is to make everyone a gold-farmer. Hence, Axie Infinity and endless think-pieces about the future of “play-to-earn” gaming.
Axie Infinity is basically an off-brand Pokemon game where you farm, train, and battle your Axies. The Axies are NFTs. The in-game currencies are cryptocurrencies. You can use the currencies to breed more, upgraded Axies. You can buy and sell both the currencies and the Axies on crypto exchanges.
There are two other things you ought to know about Axie Infinity:
(1) You can’t download it from Apple’s App Store (because crypto).
(2) You need three Axies to start playing the game, which costs about $1,000.
[Let’s stop right here for a moment. Would you ever spend $1,000 to try out a single video game that you’ve never played before? Do you know anyone who would ever spend $1,000 to try out a single video game they’ve never played before? No? Good. Congrats on not being a rich idiot, and on not having rich-idiot friends.]
But wait! You own the Axies! They’re on the blockchain and everything. So you can sell them later. It’s an investment! And “guilds” can help defray the cost of getting started. They’ll front you the starter set, and then you get to be their… um… employee, giving them a portion of your earnings as the game grows. (Uh, that hasn’t turned out so great.)
The reason this obviously-stupid idea is worth writing about is that, for about a year, it was treated widely as not-obviously-stupid-at-all. It was hailed as a breakthrough use-case for Web3. Credulous articles were written about how play-to-earn sensation Axie Infinity was letting people in the Philippines make far more playing a game than they would in their normal jobs.
Packy McCormick helped inflate the hype balloon with this breathless post. What really blew him away was Axie’s “eye-popping growth”:
Delphi projects that Axie will end the year having generated $1.1 billion in revenue, mostly from fees generated through its NFT marketplace and from breeding Axies to meet the demand from hundreds of thousands of new players each month. There are more than 600k players currently, around the world, as many as 60% of whom are in the Philippines.
But while those numbers are mouth-watering, they’re actually not the biggest or fastest ever in gaming. The rare gaming mega-hit gets really big, really fast. Many fade away just as quickly.
What makes Axie Infinity unique is that this wild growth is just the first step in a much larger plan that includes vertical integration, aggregation, decentralization, and literal worldbuilding.
Impressed by all that growth and convinced they had finally found a real Web3 use-case, Chris Dixon and Andreessen Horowitz (a16z) followed up by leading a $152 million investment round into Sky Mavis, the company behind Axie Infinity.
Liron Shapira has a great Twitter thread discussing what inevitably happened next. It’s worth reading in full.
The economics of Axie Infinity only work if more and more people are joining the game, buying Axies, and paying for tokens. That’s a tough value proposition, since the game isn’t, y’know, actually fun. Gold-farming in World of Warcraft only works so long as tons of people want to play it, and some of them have more money than they have time. YouTube/Twitch streaming works so long as tons of people like playing a game, and enjoy watching someone else play it better/at a higher level/while making jokes.
But you can brute-force the economics of Axie Infinity (and likely any token-based economy, since tokens are unregulated securities) by having investors artificially inflate the value of the tokens. That can, at least for awhile, create a hype cycle where the cash infusion raises the value of the tokens —> enriching the existing player base —> leading to credulous coverage of the product as the future-of-gaming —> bringing in more suckers (er I mean players) downloading the game and forking over a grand/signing up as an indentured servant to a “guild” —> boosting the value of the tokens even further. It’s the speculative financialization of gaming, with only the faintest hint of an underlying game that people would want to play for fun instead of money.
Axie’s plan was for this to last forever, or at least until they could turn their crap game into something that could compete with Fortnite and Roblox and the rest of the market. But everyone cannot make a profit off of play-to-earn games. You need a player base that is spending money because they actually enjoy playing the thing.
Of course it all collapsed.
First there was a hack, because big money ruins everything, and when you announce to the world “look how rich we’re getting! And it’s all held in an immutable database! Also we’re still working out tons of kinks in the system!” you are also holding up a “rob me” sign.
Major crypto investors (a16z again, plus a few others) stepped in to cover the $625 million shortfall, because stories like this can spill over into the rest of their investment portfolio. So the hack wasn’t the end of Axie. Instead, what killed it was that the player base stopped growing and the underlying ponzi scheme collapsed.
And of course people stopped playing. Of course they did. It’s a garden-variety game that you have to work hard to download and then pay a grand to try out. You can buy a Playstation 5 and a Meta Quest 2, plus a bunch of games for each, or you can buy an Axie Infinity starter set. The only people who were ever going to get into Axie Infinity were the ones expecting it to become a profitable side-hustle/full-time job. The crypto boosters and the VCs could pump the growth and make it all seem viable for a little while. In doing so, their token holdings increased in value, and they had the opportunity to turn a profit before the hustle fell apart.
This is, essentially, a modified blitzscale. They focus on early growth, ignore long-term profitability, and lock up a profit by riding the speculative wave and getting out on top. — they’re equating early growth with long-term profitability, and hoping no one notices that the numbers just don’t add up.
And it’s obviously a dumb blitzscale because online gaming isn’t anything like online payment processing!
I’m not sure whether Packy McCormick and Chris Dixon couldn’t see this long-term revenue problem a mile away, or if they could see it but saw a route to profiting off of the blitzscale they were driving. Both options are bad, but they’re different kinds of bad. Either they believe so fervently in the gospel of blitzscaling that they don’t stop to ask "where will the money ultimately come from,” or they’ve realized that an all-speculation economy works well for people who have the inside track.
Antonio Garcia Martinez has a recent post where he presents his strongest, earnest argument for why the economics of Web3 will flip everything from Web2.0 on its head and eventually work out:
In Web 2, you first find a viral consumer use case and then, often much later, find some way to pay for it. Meanwhile, you float the company with speculative capital from VCs. As late as the time of Facebook’s IPO, the ads system was a mess and did not generate the sort of revenue growth that justified the company’s massive valuation. The second half of my memoir Chaos Monkeys documents just that mad (and successful) scramble fix the company’s broken monetization.
In Web 3, companies achieve liquidity early via ‘tokenomics’ and other novel mechanisms to raise capital from users (often highly speculative in nature), and build the necessary technical infrastructure to create an alternative internet: decentralized servers, identity-management ‘wallets’, token exchanges allowing easy movement among various ecosystems, etc. The viral consumer use case (if there is one) comes much later, much to the derisive trolling by the latest wave of crypto-skeptics.
That sounds very clever. It’s actually too clever by half. The money still has to come from somewhere. In Web 2.0, plenty of companies blitzscaled their way to unicorn valuations without ever having a thesis for how it would become profitable. The success-cases of the late 1990s calcified into a mythology about grow-first, profit-later that was at best only sometimes supported by the evidence. The speculative nature of financialized Web3 means that early growth is the business model. Early-stage investors create hype bubbles, cause a spike in the unregistered securities they are pumping, and then can sell quick for a sweet profit.
In the past year of Web3 hype, what technical infrastructure has actually been meaningfully built by all these Web3 companies? The legacy of blitzscaling has given way to a system where throwing-everything-at-early-growth is the beginning, middle, and end of the story. None of these companies have an answer to where the money is supposed to eventually come from.
One thing I’ll be watching for in the next few months is whether the change in stock valuations translates into a rethinking of some of the investor-myths that structure the reward system in Silicon Valley.
Blitzscaling is one of those myths. (There are others. Tech accelerationism is the biggest of them all. But that’ll have to wait for a future post.) Some of the most valuable tech firms in the world succeeded through blitzscaling strategies. Some of the most hollow tech firms in the world relied just as heavily on early-growth-at-all-costs, though.
It’s time for some serious rethinking about what works and why.