Roundup #28: Let's talk about the AI Bubble, and some random things that caught my eyes
The datacenter boom is in fact starting to look scary. A piece on ChatGPT induced psychosis, some musings on the good and bad results from every technological innovation, and a few more things.
What caught my eyes
An interesting first hand account of someone who went down a ChatGPT Induced Psychosis.
I’m a tech founder and recent graduate with a master’s thesis on how humans become attached to AI companions. That is to say, I should have been more informed about the psychological dangers of AI chatbots than the average person, and more prepared for its charms.
I was not.
An absolutely massive ‘State of AI’ report from an investment firm. It’s good an in-depth, and for the most part aligns with my State of the Revolution post but goes a lot deeper into every topic. It’s long. It’s good.
Kurt Vonnegut tells his wife he’s going out to buy an envelope:
“Oh, she says, well, you’re not a poor man. You know, why don’t you go online and buy a hundred envelopes and put them in the closet? And so I pretend not to hear her. And go out to get an envelope because I’m going to have a hell of a good time in the process of buying one envelope.
I meet a lot of people. And see some great looking babies. And a fire engine goes by. And I give them the thumbs up. And I’ll ask a woman what kind of dog that is. And, and I don’t know. The moral of the story is - we’re here on Earth to fart around.
And, of course, the computers will do us out of that. And what the computer people don’t realize, or they don’t care, is we’re dancing animals. You know, we love to move around. And it’s like we’re not supposed to dance at all anymore.”
Fact checked as legit. It’s funny. I do this kind of thing all the time personally. Taking a little detour just for fun, or going to a different grocery store just to see what it’s like.
Initially I skipped reading the post Everything is Television by Derek Thompson, because I didn’t think there’d be anything new in it. Having read the earlier post on the same topic from Infinite Scroll titled the Carcinization of Content, I figured I knew enough. But I happened to come across it again and had nothing better to do, so… It turns out there are some good quotes, and interesting little facts I didn’t know:
Television’s role in the rise of solitude cannot be overlooked. In Bowling Alone, the Harvard scholar Robert Putnam wrote that between 1965 and 1995, the typical adult gained six hours a week in leisure time. As I wrote, they could have used those additional 300 hours a year to learn a new skill, or participate in their community, or have more children. Instead, the typical American funneled almost all of this extra time into watching more TV. Television instantly changed America’s interior decorating, relationships, and communities:
In 1970, just 6 percent of sixth graders had a TV set in their bedroom; in 1999, that proportion had grown to 77 percent. Time diaries in the 1990s showed that husbands and wives spent almost four times as many hours watching TV together as they spent talking to each other in a given week. People who said TV was their “primary form of entertainment” were less likely to engage in practically every social activity that Putnam counted: volunteering, churchgoing, attending dinner parties, picnicking, giving blood, even sending greeting cards.
When Putnam was writing Bowling Alone, many of his critics insisted that he was being histrionic about the decline of social capital in America because the Internet was going to solve all our problems. In his 1995 essay on the decline of reading and the rise of digital technology, Jonathan Franzen wrote that the decade’s biggest tech boosters believed that the Internet would heal the wound that television had sliced into culture. “Digital technology, the argument goes, is good medicine for an ailing society,” Franzen wrote. Summarizing the views of tech boosters, he continued:
TV has given us government by image; interactivity will return power to the people. TV has produced millions of uneducable children; computers will teach them. Top-down programming has isolated us; bottom-up networks will reunite us.
But digital media hasn’t become the antidote to television. Digital media, empowered by the serum of algorithmic feeds, has become super-television: more images, more videos, more isolation. Home-alone time has surged as our devices have become more bottomless feeds of video content. Rather than escape the solitude crisis that Putnam described in the 1990s, we now seem to be more on our own. (Not to mention: meaner and stupider, too.)
Good and bad results from technological innovations. This is probably a very malformed thought. But let’s see what I have, here is what I want to say about this. Many of the Sora2 videos are very funny. Many creative people’s ridiculous and funny video ideas are now being created. But the technology will also lead to a flood of malicious and damaging deep fakes. Thanks to smartphone cameras I have (and I checked this) close to 9,000 photos of my 4 year old son on my phone. When I feel nostalgic and want to see what my son was up to when he was my daughter’s age I’ll be immersed in a rich tapestry of HQ video and image content of him at that age. But the constant presence of cameras in every setting has also ruined concerts, member’s clubs, and teenager’s freedom to do dumb things with little consequence. The internet gave us Wikipedia, but also pro-ana subreddits. Cars gave us road trips but also sprawling suburbia where you can’t even walk to a cafe. The list goes on. Politics should spend a large part of their attention on boosting the good results and minimizing the bad ones. But that is not easy. The hamfisted Online Safety regulation in the UK has now led to 70,000 government IDs being leaked alongside those people’s Discord messages. And to make matters worse, politicians right now are themselves spending too much time online, resulting in them spending their time on things that barely exist at all, like Antifa, or trying to WIN with one side by cancelling the other.
Let’s talk about the AI Bubble
When a bubble pops, who’s on the hook? That’s always the key question. In 2007 there were stories of strippers and taxi drivers who owned 5 condos in Florida. In 2000 taxi drivers spoke about nothing but tech stocks. Plumbers and mechanics were becoming millionaire day traders. Every asset bubble is propelled by the dangerous words: “this time it’s different”. And by FOMO of course. Can’t be the one to sell first and miss out on the peak!
One difficult thing about bubbles is that….. every time is actually different. With AI, more and more people are making sounds of bubble. Why are so many people starting to sound the alarm now? Let’s take a look:
Datacenter construction has driven almost all of the US Economy growth in H1 2025. The mystery of why the US Economy seems to just motor on despite Trump’s high and erratic tariff policies, is explained by data centers. What goes into building a data center? Nvidia GPU chips, yes. But also land, concrete, steel, wiring, plumbing, and services to put the whole thing together. Sources I found with AI and then checked put the % spent on GPUs at around 40% of the total spend. That means the other 60% flows into the wider economy through the pockets of the super long list of SMEs and enterprises that provide other things and services for the project. They will hire people, those people will get paid, and they will spend their money on stuff. Who is on the hook if the datacenter construction boom were to suddenly stop? Well… everyone downstream from that money flow. That’s not great.
Funky (and bubbly) ‘vendor financing’ schemes are common now. You might have seen the viral diagram below in the last few weeks.
This is complicated, or as we say in finance ‘highly structured’. When finance gets more structured, it gets harder to figure out who is on the hook, but that is still the key question. So what is going on here?
First of all, Nvidia really IS selling tons of GPUs and as the finance bros say nowadays they are PRINTING. Nvidia made $87 Billion in NET PROFIT on Revenue of 165B in the last 12 months (LTM). That is a more than 52% NET MARGIN, and an EBITDA margin of in the 60s percent! That is INSANELY high. It’s valuation also high, with share price at ~50x earnings. But with margins like that and business on track to grow ~70% year over year, that is actually not that crazy. This is one of the most cash generative businesses in history.
But what is Nvidia doing with that money? That’s where it gets interesting. It is investing that money in companies like OpenAI on the condition that they have to buy lots of future Nvidia chips. So basically, they are giving vouchers for free chips to OpenAI, and in return they are getting from OpenAI shares in the company and a commitment that they will spend the vouchers. This is very bubbly, without a shadow of doubt. It makes it seem as if there is more demand for AI chips than there really is. Isn’t it strange that if there is supposedly infinite demand for AI chips, the sellers are subsidizing the buyers?
But as long as things keep going well, this is still fine. What can go wrong? Well, OpenAI could lose the money. They are very good at losing money so far. If that happens, Nvidia would lose that money, but it can take the hit. Other companies that are investing tons and can easily take the hit if it goes wrong are Google, Microsoft and probably Amazon. But their share prices will go down massively.
Towards the end of the great dot.com boom in the late 1990’s, hardware vendors like Cisco were making gobs of money selling server capacity to internet service providers (ISPs). In order to help the ISPs build out even faster (and purchase even more Cisco hardware), Cisco loaned money to the ISPs. But when that boom busted, and the huge overbuild in internet capacity became (to everyone’s horror) apparent, the ISPs could not pay back those loans. QQQ lost 70% of its value.
And who is on the hook for that?
Mostly rich people, and this would be a dotcom type stockmarket crash. Nvidia, Google, Microsoft, Amazon and Meta together make up 25% of the S&P500’s value. If this whole thing goes down, only Nvidia will crash as hard as Cisco did, because the others are diversified. We’ll get a bad stock market crash, but that is in itself not that big of a deal. The economy would slow and probably go into a recession because all the knock-on effects of the datacenter spending spree would stop. Shareholders in Big Tech would lose a lot of money. That’s about it.
There is a trend towards debt financing, aka we’re levering up. But unfortunately, it is not only balance sheet cash that’s being dumped into the AI Datacenter buildout. Debt is entering the picture this year in force. The Economist writes:
During the first half of the year investment-grade borrowing by tech firms was 70% higher than in the first six months of 2024. In April Alphabet issued bonds for the first time since 2020. Microsoft has reduced its cash pile but its finance leases—a type of debt mostly related to data centres—nearly tripled since 2023, to $46bn (a further $93bn of such liabilities are not yet on its balance-sheet). Meta is in talks to borrow around $30bn from private-credit lenders including Apollo, Brookfield and Carlyle. The market for debt securities backed by borrowing related to data centres, where liabilities are pooled and sliced up in a way similar to mortgage bonds, has grown from almost nothing in 2018 to around $50bn today.
The rush to borrow is more furious among big tech’s challengers. CoreWeave, an ai cloud firm, has borrowed liberally from private-credit funds and bond investors to buy chips from Nvidia. Fluidstack, another cloud-computing startup, is also borrowing heavily, using its chips as collateral.
I bolded the scariest parts. It actually gets worse. Something has been trending in finance called Private Credit. Funds lend out money to companies in private markets which are unregulated and opaque. They have in recent years started borrowing a lot from banks, and issued bonds which were taken up by institutions like life insurers. If it is true that a large share of this Private Credit is flowing into Data Center construction and GPUs, then if that sector goes bust, a lot of Private Credit will be wiped out at the same time. That smells a lot like 2008, and could have a much more severe fallout on the actual economy, with deposits and life insurances getting wiped out. Luckily, it seems what is also different this time is that many more people are shouting bubble a bit earlier.