I’m hittin’ the “bubble” theme pretty hard lately. Some of you are probably tired of it. Ask me if I care.
No, but seriously: I don’t do bubble propaganda unless I think it’s warranted. Right now, it’s warranted.
There was a time — call it 2017 or so, during the earliest days of this site’s existence — when I felt the need to replicate the click-bait model everyone else who publishes macro-market commentary for public consumption leans on to generate reader interest. I abandoned that model a very, very long time ago, first because it’s dishonest and second because I don’t have any desire to reach reader “critical mass.”
An empty soul once said, “Obviously, every publisher’s mission is to maximize revenue and page views.” Not only is that not obvious, it’s not true. Only a mercenary would adopt such a strategic imperative for a publishing business, and if taken to its (il)logical extreme, such a mandate can and will result in the firehouse-style dissemination of Big Foot sightings, Area 51 hot takes and Nessie headlines, or their subject matter equivalents within a given publisher’s genre.
Bubble propaganda and bearish prognosticating are the macro-market equivalent of Big Foot sightings and Nessiteras rhombopteryx analysis. If you publish enough it, you will cease to be taken seriously, even if your revenue and web traffic grows, and even if serious people continue to indulge you as they would any other guilty pleasure.
You all know this, of course, and you know I’m pathologically averse to that sort of model. My model in late 2025, nearly a decade on from this portal’s inception, is best described as “f-ck it.” Not “f-ck you,” the reader, but rather “it,” where “it” is everything except readers. I write primarily for you, not for me. Where I do write for me, it’s a kind of therapy, not an attempt to turn my readership into an ATM by addicting them to market-themed sensationalism.
In simple terms: If people find their way to me and enjoy what I produce every day, that’s great. If not, that’s fine too. I’ll be ok either way. What’s important is that the folks who do discover my writing find something in it — and I don’t so much care what that something is; it’s different for everybody — that compels them to stick around for the long-term. The Monthly Letters, for example, are almost totally impenetrable for new readers, but completely immersive for those of you who know the “lore.” And that’s on purpose.
So, yeah. Bubbles. Back to bubbles. Back to the bubble. The equity market bubble. It’s real I think. Which is why I’m spending so much time on it. When I say the bubble’s “real,” I simply mean this: There are a lot of implicit and explicit promises built into the AI frenzy and not all of them are going to be kept. In fact, I think most of them will be broken, even if the broader promise around the technology itself is ultimately fulfilled.
A lot of people, then, are investing today in tomorrow’s broken promises, and paying up tremendously for the privilege. That latter point is really the crux of the issue. Getting in early always means risking an investment in a promise that’ll eventually be broken. But if the price is low, such gambles can make a lot of sense. Currently, the prices are high.
“We are strong believers that purchases made in bubble environments rarely prove to be winners,” JonesTrading’s Mike O’Rourke remarked, in a note aptly entitled “Buy First, Ask Questions Later.”
That short piece — it’s the October 6 installment of O’Rourke’s evening mailer — is chock-full of great quotables, including this one:
It appears only OpenAI is striking deals for hundreds of billions of dollars on a monthly basis. Despite the popularity of ChatGPT, OpenAI is forecast to be cash flow negative for several years into the foreseeable future. There is not much risk in spending money you don’t have.
Mike possesses something I don’t: The editorial equipoise to pen transparently acerbic color in a cadence that’s nonetheless measured enough to constitute plausible deniability in the event someone were to accuse him of being overtly abrasive. In other words: It’s abrasive, but then again, it’s just fact-stating.
In the same note, O’Rourke wrote, of Spotify’s participation in OpenAI’s pilot app program (unveiled on Monday), “We have hit a new low as a society when individuals need to outsource their Spotify playlist creation to ChatGPT.” I don’t say this often, so when I do it’s high praise: I wish I’d written that.
Coming full circle, readers are advised to steel themselves for a lot of “bubble talk” going forward. Just know it’s all done in good faith. Or at least when you read it here it’s done in good faith. I truly do believe we’re witnessing the inflation of a bubble that’ll one day be enshrined in those “A Short History Of Booms And Busts” books you can always pen if you’re determined to top a bestseller list but can’t actually write.
The rub is that I can’t tell you when to sell. Nor when to short the bubble. Nor even what to short to maximize hypothetical gains in a scenario where the bubble does burst. Even if I could do any of those things, I wouldn’t. Because I’m not a fiduciary and this isn’t financial advice.
But it’s not click-bait either. What I write here is designed to inform. Sometimes, particularly during periods of market turmoil like that which would accompany a burst bubble, that mission does indeed result in “the maximiz[ation] of revenue and page views,” as it did during the earliest days of the pandemic. In such conjunctures, I’ll take it. And no, it won’t go to charity. What do you take me for, a bleeding-heart liberal?
Can you truly be in a bubble if you’re aware of it? Just like, can you really be insane if you recognize your own madness?
Well, I can’t answer your first question, but based on extensive first-hand experience with my own mind, and my enthusiastic admissions to being some sort of crazy, the answer to your second question is “yes.”
Got it! So, through the process of deductive reasoning, the answer to the first question would also be “yes”.
I remember most participants in 1999-2000 thought that was happening was insane. So yes, you can be aware of it and just not care.
This sort of meta commentary is useful for distinguishing against a purely devil’s advocate style argument. It would have been useful even sooner, when it was less suggested by volume of writing that the argument was being made in earnest.
An upside of getting your info from someone with intellectual integrity is that they won’t often misrepresent their beliefs or level of certainty. A problem can arise, however, because the same people are often capable of arguing for things they don’t believe in the search for truth, so it’s easy to misunderstand them when they aren’t explicit about it.
“… the same people are often capable of arguing for things they don’t believe.”
And that, right there, is why even (and perhaps especially) a hopelessly inebriated version of me was so valuable — I’d argue indispensable — to a propaganda outfit.
Simply put: It’s impossible to know what I really believe. I could tell anybody, anything at any time, and convince them not only that it’s true, but that I believe it wholeheartedly.
There are only four people in the world capable of discerning what I really believe about anything. Those four people are all recurring characters in the Monthlies.
I will disagree with what you think of the monthlies and tell you that I think they are standalone in any magazine publication. Bio is unnecessary, most people not familiar with you would assume it’s interesting fiction. You write with an alive voice, thank you.
God, I love this place. Even when I was an active academic I couldn’t get real discussion from my colleagues.
Not just Spotify playlists, people are using AI features and chatbots to do their Fantasy Football picks
I think what makes this worse from 1999-2000 is the unclear benefit of what we are actually getting from this.
Krugman, this morning, suggests that without the AI bubble, we’d be in a recession.
https://paulkrugman.substack.com/p/why-arent-we-partying-like-its-1999?publication_id=277517&post_id=175514056
If it is fair to proclaim the birth of this bubble as the NDVA report that started it all (the earnings report heard around the world), I venture that we might need a similar event to kill the now adult bubble. With many market participants in bubble watch we could get an early exit on an otherwise normal correction that then gets bought furiously, but if we get an unexpectedly bad earnings report from an AI darling or a superscaler that is likely the catalyst that triggers a true exit and subsequent cascade down, another earnings report heard around the world. How close are we to this event? I still think there is plenty of room for further exuberance, my guess is Q1-2 2026, but what do I know? I have been adding stop loss orders on most of my positions just in case ChatGPT generated music playlists are not all they are promised to be.
After the bubble bursts, those left standing will probably fall into one of three categories: cutting edge niche AI that has become essential for certain industries and can thus become profitable at some scale; good “generic” AI that can greatly assist most people and do so profitably at a low cost; and those (perhaps many) AI companies that simply failed.
Amazon, Google, and Microsoft all survived the DotCom Bubble, and it took Steve Jobs (and a deal with Microsoft) to save Apple. Technology and product acquisitions will likely play an important roll. However, the catalyst need not be internal. If the overall economy hits a sandbar (or an iceberg), the “winnowing” of AI companies could begin in earnest.
Advanced chips are closely linked to national security and defense. I don’t see this bubble popping yet- but I’m starting to look more closely at the end users for AWS, Google cloud, etc., as well as the results for defense contractors. This is one industry where the US should absolutely should want to replace human jobs with AI. Bring on the drones, UAV’s, robotics.
The other big user of advanced chips will be nuclear power (small modular), which Bill Gates is quietly working on; to roll out by 2030’s.
https://www.rand.org/pubs/commentary/2025/02/dont-be-fooled-advanced-chips-are-important-for-national.html