Silicon Valley runs on Futurity
lessons from the 2023 tech recovery
Here’s my working hypothesis: Silicon Valley runs on futurity.
I’ve been trying to make sense of the tech recovery of 2023. It is broader than you would think. The entire sector had a catastrophic 2022. It wasn’t just crypto, Twitter, and Netflix. A year ago, I wrote about the fallout in a post titled “What Comes After the Tech Crash.”
AirBnB stock dropped 50% in 2022, from $172 to $85/share. It ended 2023 at $136/share, a 60% increase.
Alphabet (Google) stock dropped 39% in 2022, from $145 to $88. By the end of 2023, it was back to $139/share.
Amazon stock lost half its value in 2022, dropping from $170 to $84 per share. It ended 2023 back at $152/share.
Apple was hammered less in 2022. It only dropped 28.6%, from $181 to $130/share. It recovered in 2023 as well, and is now over $192/share.
Coinbase had an atrocious 2022. It shed 85% of its value, plunging from $251 to $35/share. In 2023, it recovered to $174/share. (That’s a 417% increase for a company that has spent the year in serious trouble with the SEC. Yikes.)
Doordash dropped in 2022 from $144/share to $49/share. Now it’s back up to $99/share.
Lyft lost 75% of its value, reducing from $44/share to $11/share. It has a much slighter recovery, improving only to $15/share (which is still a 35% increase if you made a New Years 2022/3 resolution to buy all the distressed big tech stocks).
Microsoft began 2022 at $336/share and ended the year at $240/share. That decline is about on par with Apple. It experienced an upswing in 2023, ending the year at $376/share.
Netflix, which appeared to set off the tech crash with its spring 2022 quarterly earnings report, dropped from $597/share to $295/share in 2022. It has now bounced back to $487/share, despite 2023 being dominated by, y’know a historic writer’s strike.
Tesla, I mean oh man, oof. The stock began 2022 at $400/share. It ended the year at $123, and Elon was staring at potential margin calls. Over the course of 2023 — a year full of stories of missed deadlines, vaporware products, vehicle recalls, and the Cybertruck — the stock closed at $248/share.
Uber dropped from $42 to $25/share in 2022. That’s about a 40% decline. It ended 2023 at $61.50, a ~140% increase.
Then there are the two big cryptocurrencies — Bitcoin and Ethereum. Bitcoin started 2022 at $47,733 and ended at $16,583. Ethereum plunged that year from $3,767 to $1202. Brutal. And yet, somehow, Bitcoin closed out 2023 at $42,645, Ethereum at $2,293.
One year ago, the tech crash seemed to be a long-overdue correction. The price of all these tech stocks had skyrocketed during the pandemic, when big tech and real estate were basically the only two safe investments. When the Fed raised interest rates, a lot of speculative tech investments dried up, and people started looking at the big tech companies and asking sensible questions like “wait why the hell does this company have a 50x price-to-earnings ratio? Their business model is a monthly membership fee. That isn’t a $200 billion business.”
The message, in other words, seemed to be: “digital technology isn’t magic anymore. We’re going to start evaluating big tech companies as though they are companies — on the basis of their expected profits, not their lofty ambitions.”
I’m not surprised that many of these companies rebounded in 2023. But the size of the rebound is shocking. It’s as though the tech crash of 2022 never happened. Like the message of 2023 was: “jk jk, you are magic and special and we just love how you inspire us to dream.”
And what makes this especially jarring is that, if we pause to look back at the major headlines that defined 2023’s year in tech… It was not a good year.
The Metaverse was pronounced dead. Venture Capitalists freaked out and killed their own bank. All the crypto kingpins went on trial and lost (except CZ. He entered a guilty plea and stepped down from Binance instead.) Elon Musk and Mark Zuckerberg toyed with a literal cage fight. Apple released the Vision Pro AR headset at a price point so high they clearly didn’t expect anyone to buy the thing. Self-driving cars hit a, uh, speedbump? (Is that we should call Cruise running over a person, dragging them 20 feet, and then getting shut down after lying to regulators about it?) Tesla’s big release was the Cyberfrukk. Elon spent the whole year Elon’ing Twitter. Crypto spent another year producing zero practical use cases besides gambling (as Brad Delong elegantly pointed out earlier this week).
Interest rates remained high, which directly translated to a terrible year for startups. It was a bad time to launch a new company, unless that company had a stash of GPUs and could make big AI promises.
Meanwhile, employees started unionizing. Which is great! But it’s specifically great in an “investors will hate it, but they’ll just have to deal; tech is about more than share prices” sort of way.
The sole bright spot was generative AI. Generative AI still has a ton of problems. It remains a revenue sink, not a revenue generator. And OpenAI, which was supposed to be the white knight of the industry, dissolved into weird-nerd-infighting at the end of 2023. And, particularly for the big platforms, the biggest near-term promise made by AI seems to be, “This will cannibalize your main revenue streams, and replace them with something that generates less money for you.”
But what it nonetheless provided for the industry was a sense of futurity. AI is full of futurity.
Futurity ain’t solid. You can’t hold it in your hands. It isn’t liquid either — there’s no money in it, at least not yet. It’s a gas. It fills a room. No one can quite see it, but everyone breathes it in.
Or, to lean on the slighly-overused parlance of the day: futurity is a vibe shift.
The 2022 crash was about vibes. It was also about high interest rates and recession fears and increased regulatory scrutiny etc, etc. But all those datapoints congealed into an overall mood. Signs of “the rot economy” and platform “enshittification” were everywhere. The big investors got spooked, and high frequency trading algorithms picked up on that mood shift and amplified it.
2023’s AI hype bubble fixed the vibes. Tech companies started to feel like the future again. They started to tell stories about the limitless revenues they would unlock someday. And the big investors liked that talk. And the trading algorithms amplified that too.
This, more than anything else, seems like the underlying point of the AI hype cycle of 2023. Sure, there is a material thing being produced here. And that thing might one day have real value. But investors and executives — the big players with market-shaping power — don’t actually live in the future. They just enjoy LARPing at it. They live in the present. And, in the present, what they require is a big story of potential future profits that can fuel absurd current valuations to sustain all the advantages that come from their extraordinary paper wealth.
I don’t see how else to make sense of it. 2022 was the year the 20-year tech bubble finally burst. 2023 was still bad for startups, and was full of bad headlines for the big platforms. And yet, in the markets, tech investors just took a deep collective breath and started inflating the next bubble, as though the previous year had never happened.
The essence of the Silicon Valley business model is futurity. It’s a genre of storytelling, a vibe. This is divorced from the business models of the individual companies. But, as we saw in 2022, those companies can lose a 25-50% of their market value while still recording huge profit margins.
I won’t pretend to know what this means for tech stocks in 2024. But I expect what we’ll see is that the fates of all these big companies are heavily correlated with one another. And it will all boil down to how well they can maintain the air of futurity. That’s what sustains the ridiculous valuations. It’s the thing that protects these companies from being judged like normal businesses. Futurity is Silicon Valley’s most valuable asset.
At the risk of proposing a Theory of Everything, I think - maybe! - part of what happened in 2023 was a continued cleavage of our culture/discourse/economy into, broadly, the materialist and post-materialist ("vibes") camps. We see this also in the confounding and uneven reception of macroeconomic data, the stickiness of glib misinformation in the face of contradictory actions and reality (see: student loan forgiveness, responsibility for). Probably the continued gutting of our information ecosystem at every level teamed with continued dissipation of faith in institutions has something to do with this!
But also, this would explain why there's a resurgent labor movement focused on and motivated by real material concerns happening at the same time as you have the return of goofy tech valuations - a widening *epistemic* divide.
GOOG is trading at a PE of 26; that's not insane for a company that owns a significant chunk of the online ads business, has a decent share of the cloud computing business (which will only grow as security issues continue and software gets more complicated to manage), and may even have usable self-driving car technology.
MSFT's PE is 35, but they've had better growth than Google, and (much) less of it is in the highly variable ad business. MSFT has the 2nd largest cloud business after AMZ, and they own the office productivity business.
AAPL has a PE of 29, and they own the consumer cloud the way that MSFT owns the business cloud. AND they sell hardware that's essentially impossible for mortal humans to clone.
A lot of these stocks took a hit when interest rates went up to ~5%. They're going back up because falling interest rates mean that soon you'll have to do something smarter with your cash than just buy T bills.
You discuss Crypto as if it has no value, but it is very valuable to criminals who can use it to launder money. They can also use the Etherium block chain, if I understand it correctly, to create their own coins, which they can then pump-and-dump.
The Metaverse was a huge money suck for FB; it was hopelessly bad idea. Zuckerberg pulling the plug on it just means more profit for FB. Kudos to Mark for pulling the ripcord *before* hitting the ground.
Generative AI is harder to evaluate. I don't really understand regular machine learning; I only understand back propagation for about 30 minutes after reading an explanation of it, before it fades away again, and Generative AI is more complicated than just back prop. It looks like it can generate mediocre answers to a fair number of queries, although it BS's when it doesn't know the answer, and sometimes even when it does. It sounds, more or less, like a fairly bad employee trying to hide the fact that they don't really know how their business works. Given how expensive these things are to operate, it probably makes more sense to hire the bad employee than go all-in on an AI system that can't even tell you how it made any decisions.
All in all, the only thing that looks bubblicious right now is generative AI -- deploying it to do real jobs from 9-5 would require a massive energy budget and still promises a crappy result at best.