No weasel, but maybe a bubble.
Nvidia posts earnings tomorrow.
https://investor.nvidia.com/events-and-presentations/events-and-presentations/default.aspx
Peter Thiel’s hedge fund, Thiel Marco, revealed in its regulatory filing late last week that they had sold off all of their Nvidia holdings during the third quarter. As did SoftBank. The “Big Short” guy shorted Nvidia to the tune of $186M in the third quarter.
https://www.investopedia.com/peter-thiel-s-hedge-fund-dumped-nvidia-shares-just-before-its-big-earnings-report-11851890
Nvidia, of course, manufactures the chips preferred by AI data centers. And, invests in their own clients, in a sort of circular financial vortex.
https://www.forbes.com/sites/roomykhan/2025/11/03/all-roads-lead-to-nvidia-bankrolling-its-own-ai-gold-rush/
Brighter minds than mine will surely chime in to explain why this is not a matter of concern. Nonetheless, a truly huge chunk of the growth in the stock market in recent months is down to AI.
Anybody nervous?
"This is the Voice of Moderation. I wouldn't go so far as to say we've actually SEIZED the radio station . . . "
Carole Cadwalladr has been on this for a while, and posted about Thiel and Nvidia yesterday. Many of her informant techbros are seriously sounding the alarm about the imminent pop….
Brighter minds than mine will surely chime in to explain why this is not a matter of concern.
I think what you mean is, more credulous minds than yours.
At current valuations AI would have to bring in $400 per year per US resident for the AI companies to produce a decent return on investment. Which isn’t happening in the foreseeable future.
“Bubble” is exactly what we’re looking at. The question is when, not whether, it will pop. And how big an impact that will have on the economy overall. Personally, I’m going nowhere near stock in any company which is big into AI. But then, I’ve been avoiding bitcoins like the plague, too.
I’ve been reading about this for a little while. I already moved a good chunk of my retirement holdings from stock funds to bond and money-market funds. I was already up well enough and am getting closer to retirement, so it wasn’t too radical a step.
Even without the bubble speculation, the indices were staring to make my spidey senses tingle. The AI-bubble stuff I’ve been reading just pushed me out of complacency.
From the BBC: Google boss says trillion-dollar AI investment boom has ‘elements of irrationality’
In the middle part of the interview he muses about the huge energy costs of AI, only to conclude that new energy sources are going to be necessary to avoid constraining the economy. The environmental cost seems already to have been written off as a concern there. No doubt that will be taken care of automagically by the power of The Singularity.
As for the jobs thing:
So if it works it’s going to suck up tons of energy and put people out of jobs, and the irrationality surrounding its growing pains will crash economies and ruin small investors and a lot of the less secure AI firms.
And once the survivors finally get AI off the ground we can look forward to them enshitifying it as thoroughly as they have the internet, which was probably at its best in the brief moment just before every idiot with an MBA and an in with a venture capitalist kicked off the boom with a fuzzy business plan and a dream of early retirement.
Lovely.
I’ve been listening to a lot of podcasts about this and they emphasize it is not just an Nvidia bubble, but a problem with the whole industry. A couple of them point to this exchange with Altman
https://futurism.com/artificial-intelligence/sam-altman-loses-cool-revenue
The problem is that there needs to be a starting event to pop, and that might happen tomorrow, next month, next year or in 2 or 3 years. So until then, we just watch and wait.
the software industry as a whole is all-in on AI (LLMs in particular).
makers are trying to stuff it into every nook and cranny they can find. from Adobe using it so Photoshop can produce image content for you on demand (instead of you needing to use stock photos), to phones using it to automatically manipulate your photos in real-time, to audio apps using it to handle routine things like sound mastering all the way to vocal generation, to programming environments using it to analyze and correct your code for you, to apps like the one i work on using it to take over ‘help’ and data search/analysis features (“how many tables in this library use address data?”). if there’s a place a company thinks people will want to type their desires in natural language, they’re trying to use AI to make it happen.
YouTube recently rolled out a feature where, if you are already a content producer, it will look at your channel and try to figure out what kind of content you make. then it will suggest a list of related topics for you to produce new videos on. if you pick one, it will give you a list of ‘hooks’ to use to make the video sizzle. it will give you detailed outlines for a script. and it has AI-generated thumbnails ready to choose from. and at every stage, you can use AI prompting to fine-tune the suggestions. it’s literally doing everything but speaking. and i don’t see any reason why they haven’t automated that part, too.
so, the big players are definitely over-inflated. but the whole industry is using their products now.
the software industry as a whole is all-in on AI (LLMs in particular).
makers are trying to stuff it into every nook and cranny they can find.
As noted, the costs are enormous. So using AI everywhere is the only prayer of actually making any money from it.
And for non-AI software companies, it’s a matter of not looking like they are less than cutting edge. It’s folly, except in a few narrow cases. But here instinct is powerful in the industry.
Nvidia, of course, manufactures the chips preferred by AI data centers.
Nit-picking… Nvidia designs chips. All of their chips are fabricated by TSMC in Taiwan. Earlier this month Nvidia’s CEO said publicly that he had asked TSMC to increase their 3nm fab capacity because of the volume of new orders Nvidia was receiving.
Should have added, the various companies that have decided to design their own chips for AI — Google, Amazon, Microsoft, Apple, probably some others — also depend on TSMC for fabrication.
Another potential issue here. If China decides to forcibly reunite Taiwan, all that chip production is no longer secure. (Even making the heroic assumption that the various fabs are not severely damaged or destroyed.)
That would, I suspect, bring AI expansion to an abrupt halt. With obvious massive disruption to any business (not just AI providers) which has reorganized and restructured their operations to depend on it.
we need a mega-sized-CHIPS Act, not a Trump-branded one.
https://www.msn.com/en-us/money/other/what-is-the-chips-act-why-does-trump-want-to-change-it/ar-AA1C7yH1
It seems insane to me, but if China were to decide that if they can’t have 3nm chips — and Biden certainly leaned on Taiwan and countries that might act as intermediaries to block access — then no one can have 3nm chips… well, a dozen ballistic missiles with big enough HE payloads (and sufficient accuracy) puts TSMC out of that business. TSMC is building a very large campus and production lines in Arizona, but has consistently said they don’t plan on ever locating their leading edge processes there.
Since no one else has posted it, Nvidia reported record revenue and profits up 65%. The stock is up 5% in after-hours trading.
At current valuations AI would have to bring in $400 per year per US resident for the AI companies to produce a decent return on investment. Which isn’t happening in the foreseeable future.
Individuals may not spend $400 on AI-based services. Employers may purchase much more than that on behalf of each of their (remaining) employees.
Side note from the AI front lines…:
I have in the past dealt with AI hallucinated sources, and with AI suggested secondary sources that were either inappropriate to the paper at hand, or that were misrepresented by the AI synopsis…
This week, however, marks the first time I have had a student submit a paper where the AI has hallucinated quotations for the primary source and made up a new plot for the story. And this is also the first time the student turning in an AI generated paper has not recognized that AI has made up shit about a story that they had supposedly annotated, and that we had discussed at length in class.
Meanwhile, several of the students are writing projects that worry over the effects of AI on the fields that they are currently studying to become a part of, fearing that their future jobs may be transformed into something unsustainable by the time they get out of college despite it seeming like a good choice when they started.
College is a much bigger investment and a much bigger risk for them than it was for us. The cost has exploded, and the state governments are happy to allow that to happen so long as they can pass that expense on to students in the form of loans.
It’s impossible for them to make an informed decision. People like Mr. Pichai are telling them that in order to prepare for the future they will need to learn how to use AI, but they are also worried that using AI will prevent them from learning the skills they will need to be able to adapt in a changing world. Not an easy bind to resolve for a brain in its early twenties. They lack the experience needed to make good judgments about these things.
This is a podcast I’ve been watching, it’s a branch of Scott Galloway’s podcast
https://youtu.be/K8b1An5Rpcs?si=z7cix2YBS4UDHNSf
China’s lowcost open-source AI is suddenly everywhere in Silicon Valley. While the US is spending billions of dollars building ever bigger models and data centers and inflating what some analysts say looks like an AI bubble, China is doing the exact opposite, it seems. cutting prices, opting for open sourcing weights, and shipping models that are cheaper, lighter, and surprisingly competitive, and American companies are flocking to them. Airbnb CEO says his team ditched Chat GPT for Alibaba’s Quen model because it’s quote unquote fast and cheap. Venture firms are switching to Moonshot’s Kimmy LLM, and developer data show nearly half of the most used models in the US last week were Chinese. Alibaba just slashed prices on its flagship model again as part of a full-blown AI price war and a sign that involution has come to Chinese AI.
What kills me is that this is basically capitalism 101 and the Chinese are proving to be a whole lot better at it that the US.
What kills me is that this is basically capitalism 101 and the Chinese are proving to be a whole lot better at it that the US.
Say rather that the Chinese are a lot better at it than enormous Silicon Valley firms. Which is to say, firms that so dominate their niche that they no longer have to pay much attention to competition. If you are Google or Microsoft or Amazon, you’ve got money to burn. So why not spend some of it on whatever shiny object has caught your biggest shareholder’s eye?
Expect them to eventually, possibly sooner rather than later, discover why not. Although, if they can (apparently) afford to pay their CEOs hundreds of millions per year, maybe not.
Say rather that the Chinese are a lot better at it than enormous Silicon Valley firms.
Given that the so-called ‘Magnificent 7’ account for half of the NASDAQ, I think I can get away with calling it American capitalism. It’s not like there are some tiny companies, hidden in the shadows, that are somehow ‘doing capitalism right’.
It’s not like there are some tiny companies, hidden in the shadows, that are somehow ‘doing capitalism right’.
Actually, I think there are. Not just tiny ones either; medium and moderately large ones, too.
The behemoths get all the attention precisely because they are so big. As you say, a handful of them make up half the NASDAQ. But by numbers, as opposed to simple market capitalization, the vast majority not only do more business, they employ, in aggregate, far more people.
And they survive by doing capitalism right. Not “right” as a lot of economists seem define it. They take care of their customers and they take care of their people. Without screwing over either in pursuit of a tiny fraction of a percent better numbers. And their executives don’t make tens (or hundreds) of thousands of times as much as their average employees.
And they survive by doing capitalism right. Not “right” as a lot of economists seem define it. They take care of their customers and they take care of their people. Without screwing over either in pursuit of a tiny fraction of a percent better numbers. And their executives don’t make tens (or hundreds) of thousands of times as much as their average employees.
I dunno, this sounds like that AI slop that I get all the time on facebook.
The CEO looked out the window and thought about his customers: all the athletes who actually cared about the planet they trained on. The company’s mission statement, which prominently featured the word “sustainability,” sounded like a cheap lie given the audit results. The boardroom fell silent waiting for his decision…
cf. Marx General Law of Capitalist Accumulation
Two observations.
These guys have enough money invested that they can remain solvent longer than the market can remain rational.
And the fact that this is an AI bubble makes me wonder if the valuations are being generated by an LLM hallucination as well…
As usual, the stock market is driven by competing motives: greed and fear.
Banknotes stuffed in a mattress is looking better and better.
Or perhaps Giant Stone Wheels.
lj, I suppose it depends a lot on the behavior of the companies one is familiar with. I’ve certainly seen (and worked for) some that practice “devil take the hindmost” capitalism.
But I’ve also worked for companies which thought being a good person and a good citizen was important. And seen a lot more of them. Plus, of course, their owners and managers were bright enough to realize that their people were what kept the whole thing running.
Banknotes stuffed in a mattress is looking better and better.
Works right up to the moment that serious debasement of the currency (aka inflation) sets in. Which, the idiots at the top of this administration being how they are, seems a distinct possibility.
Might be better to put your money into reconfiguring your back yard (if you have one) into a big vegetable garden.
But check the soil first. I would not eat anything grown in our little backyard garden* – at least not without changing the top 1 m of earth first. Everytime anything is done on this building, the materials are stored there and I take it as a given that some nasty stuff seeped into the soil on more than one occasion in the last half century we have lived there.
*the rental flats on the raised ground floor have one (with stairs from the balcony leading down to it).
Here’s another piece that I ran across on NPR:
https://www.npr.org/2025/11/23/nx-s1-5615410/ai-bubble-nvidia-openai-revenue-bust-data-centers
The parts of this article that really had me shaking my head at these hubristic tech muppets were the reports of how much they were spending to build tech centers and how gormless the private equity pinheads are being in their rush to invest money in them.
And then we have this:
The return of the “special purpose vehicle” for financing. So very bubble.
I’d say we are better off investing in tulip bulbs, but at the rate those data centers will swallow up water and warm the planet, you’d never get those tulips to grow, and the Netherlands will be entirely underwater – just like all those mortgages were the last time we let the promise of easy money gull us.
but at the rate those data centers will swallow up water and warm the planet
doubling compute capacity every 6 months is a hell of a rate.
https://arstechnica.com/ai/2025/11/google-tells-employees-it-must-double-capacity-every-6-months-to-meet-ai-demand/