The boom-and-bust cycle of bubbles is dangerous, destructive, and only really benefits the wealthiest and most-disconnected people alive. While there might be ways to get rich by virtue of piling your money in at the earliest point and exiting before everything explodes, that’s something only achievable if you’re either incredibly lucky, or you have some inside knowledge that others lack.
Bubbles are by definition unstable, illogical, and manipulated by a combination of captured media, wealthy (and brainless) financiers, and cynical grifters that know how to pull the different levels to keep the numbers going up. And they’re always, without fail, destructive.
At the root of everything lies a cynicism that vibrates through every part of society. Try and call a bubble early and you’re told that “the market can stay irrational longer than you can stay solvent,” or that “being early is the same thing as being wrong,” with any attempt to push back on narratives met with some sort of half-hearted statement about how “everybody is putting money into this” and the implication that you should simply give up and accept it.
As the bubble inflates and approaches breaking point, the very same people that told you that it wasn’t a bubble will say that “it’s obviously a bubble, but there will be winners at the end,” never quite explaining who they might be, how they might “win,” or really making any attempt to make a call about anything.
Bubble boosterism is, at its heart, an act of cowardice. It makes no attempt to explain why or how anyone wins, what that might look like, or when anything might happen other than repeating whatever the powerful says and then declaring victory based on whatever shred of proof they can find.
These people also frame skepticism and cynicism as somehow villainous — that critiquing the rich and powerful is some sort of malignant act that works counter to a productive society — when boosterism is far, far worse. Boosters (correctly!) suggest that LLMs are now a dangerous new tool in the armory of the average cybercriminal (none of which requires LLMs to be autonomous, sentient, or otherwise “going rogue”), but refuse to admit their role in the ascent of the companies that built them.
Had major media outlets not glazed OpenAI, Anthropic, and other AI companies in the earliest days, they wouldn’t have been able to raise the funds or justify the capacity to build models that are specifically trained for hacking. None of this had to be built, nor does its existence disprove anything about my greater thesis that AI’s ROI and capabilities have been and are overstated.
This industry — like every bubble — has taken advantage of a few core weaknesses in the media:
- Its willingness to believe anything with a large number attached.
- Its willingness to believe anything with a big company behind it.
- Its willingness to believe anyone with enough “experience.”
- Its inability to do anything other than follow consensus.
- The fact that we have multiple members of the media that desperately want to be friends with tech executives, and/or to see their companies succeed.
The assumption — despite the opposite being the case again and again and again — is that the powerful will always pull off whatever it is they promised, with everyone somehow forgetting that Mark Zuckerberg changed his company to Meta and burned $70 billion on the Metaverse before killing it two years later, and that Satya Nadella said in 2021 that he “could not overstate how much of a breakthrough” it was.
What has made AI so profoundly different to the metaverse is that there was something to speculate on, and a consumer-facing product for journalists to half-glance at before assuming that all of the other promises were true.
It all “made sense” if you didn’t really want to think about it. Hyperscaler stocks skyrocketed as they invested tens of billions of dollars in GPUs, banks loaned them hundreds of billions of dollars to do so, and AI startups raised hundreds of billions of dollars. It didn’t matter if nobody would talk about their revenues or that OpenAI would lose $5 billion in 2024 because theoretically, at some point, all of this might be worth a lot of money, maybe.
Asking sensible questions like “how much money are hyperscalers making?” gets you attacked and harassed on a daily basis, with random people screaming that “cloud revenues are skyrocketing because of AI” despite hyperscalers not actually telling you how much money AI makes, because hyperscalers know that all they need to say is that “cloud growth is so big because of AI” for everybody to look the other way. Questioning Anthropic’s profitability or failing to be sufficiently excited about its IPO is considered bad faith, because in a bubble, it’s considered “bad faith” to ask questions or be remotely skeptical. Those who dare to turn away the pro-AI (which can mean both “AI is the biggest thing ever” and “AI is going to literally destroy everything tomorrow”) slop are framed as outsiders, their intentions questioned a hundred times more than anyone who sits there saying “I love whatever the rich and powerful have planned for you and me so much!”
This is, consciously or otherwise, a systemic response. To question a bubble is to question whatever new fad the markets and the media have decided is “in,” and doing that is scary because so many people benefit from a bubble’s inflation. People in the media get a new thing to write about (or be an “expert” in), banks get a new thing to invest in, tech reporters get new funding rounds to “cover” and executives to obsess about, venture capitalists get a place to park their money, and governments have a vague new place to point at to say how the economy will continue to grow indefinitely. It doesn’t matter to anyone on the way up how much damage there is to follow, mostly because thinking about bad things requires you to learn about why they’re bad, and take an actual position, which requires courage, when it’s so much easier to just follow everyone else.
The bubble is yet to burst, but it’s clear that it’s very much on the way to doing so.
And today’s premium newsletter is about how we stop one inflating again. I’m going to have a little fun with this, mixing regulatory ideas and societal moves that could be made to kill the momentum of any further attempts by any industry to drum up speculative valuations.
This will be a grab-bag of ideas I’ve been working on for a number of years — a fun newsletter that you can dig into over Thanksgiving, where I’ll be taking a week off both the free and the premium — and I look forward to hearing your feedback.
Let’s get going.