Today, Explained - The AI bubble

Episode Date: October 28, 2025

And what happens when it pops. This episode was produced by Avishay Artsy and Kelli Wessinger, edited by Amina Al-Sadi, fact-checked by Laura Bullard, engineered by Adriene Lilly and Patrick Boyd, and hosted by Noel King. Enormous valuations across the AI industry are raising questions of a bubble. Photo Illustration by Thomas Fuller/SOPA Images/LightRocket via Getty Images Listen to Today, Explained ad-free by becoming a Vox Member: vox.com/members. New Vox members get $20 off their membership right now. Transcript at ⁠vox.com/today-explained-podcast.⁠ Learn more about your ad choices. Visit podcastchoices.com/adchoices

Are we in an AI bubble? A simple answer is, we better hope not. You are far too young to remember Tulip Mania in 1637, Google it, but perhaps you remember the housing bubble that caused the 2008 great financial crisis. It was the worst day on Wall Street since the crash of 1987. Shock and panic evident on the faces of those on the trading floor. How many you people want to pay for your neighbor's mortgage that has an extra bathroom and can't pay their bills?

When bubbles pop, there's pain. But while consensus is growing that we are in an AI bubble, there's a fight raging over what type of AI bubble we're actually in and what will happen when she goes pop. What to keep your third eye open for coming up on Today Explained. Paramount Wolfe Check out the big stars, big series, and blockbuster movies streaming on Paramount Plus. Cue the music. Like NCIS, Tony, and Ziva.

We'd like to make up our own rules. Tulsa King. We want to take out the competition. The substance. This balance is not working. And the naked gun. That was awesome. Now that's a mountain of entertainment.

Paramount Wolf. You know what's better than the one big thing? Two big things. Exactly. The new iPhone 17 Pro on TELUS' five-year rate plan price lock. Yep, it's the most powerful iPhone ever, plus more peace of mind with your bill over five years. This is big. Get the new iPhone 17 Pro at TELUS.com slash iPhone 17 Pro on select plans.

Conditions and exclusions apply. This is an artificial intelligence version of Drake and U.L.S. name told to the explain. 20205 has been a boom year for the stock market, despite a ton of economic uncertainty. Powering the boom is AI. Big investments, big faith in AI. Does AI deserve it, or is this a bubble? Lily Jamali covers technology for the BBC. Lily, let's start this way. What is a bubble? That is a great question.

A bubble can take hold when there is an expectation of future profits. And when that expectation is so alluring that it ends up leading to speculation. So there's a long history of boom-bust cycles in our country. And it's not always clear when you're in one. Even when you think we're in a bubble, you know, the timing of that can get pretty tricky. At the moment, everybody's talking about an AI bubble. Why is that? I think it's because of, first of all, all the interest in artificial intelligence.

You know, I live in the San Francisco Bay Area, and when I drive down the 101 freeway, it is all billboards about AI. So there you go. You know, that's sign number one. What are the billboards on the 101 saying? But more significantly, you know, I think there's a real concern right now about the potential blast radius that an AI bubble bursting could have because there's just so much money pouring into artificial intelligence. So take a company like OpenAI, which brought artificial intelligence

into the mainstream in 2022 with ChatGPT. That company was actually founded a decade ago as a nonprofit by its current CEO, Sam Altman and Elon Musk actually. They're now suing each other in court over what happened next. But its mission at the time was to create artificial intelligence tools for the benefit of humanity. Fast forward a few years. And Altman has been reorienting the company to be more of a for-profit enterprise. So they've been taking investor money left and right to the point now where OpenAI is valued at $500 billion. It is the most valuable private company anywhere. And so you might think, well, if anything goes wrong, it's those people who've been able to invest that are going to be affected.

The thing is, Open AI and some other AI developers have been taking investments from publicly traded companies, companies that you and I can invest in, and that we actually might already own through our retirement funds without even knowing it. So there's this growing web of deals and sticking with Open AI, let's look at that. NVIDIA, the AI chip designer, is invested in OpenAI in a big way. Chipmaker AMD, Rodcom, they both just struck deals with this company in recent weeks. Microsoft had a massive early partnership with OpenAI, same with Oracle. It is a very long list, and it's growing all the time.

AI development, at least the way that OpenAI and the other AI companies in the U.S. are doing it, involves serious infrastructure, so data centers, and a lot of sophisticated chips need to be purchased, the kind of chips that Nvidia actually designs. So, Nvidia just announced this $100 billion investment in Open AI a couple weeks ago. The issue there is that a lot of that money will presumably make its way back to Nvidia. So that construct where, Noelle, I give you money, and then you in turn by my stuff is sometimes known as vendor financing or circular financing. Instead of one company paying another for its products, these AI giants are seemingly passing cash back and forth through stock agreements and future chip purchasing deals.

Vendor financing has a pretty spotty history, frankly. It's been a feature of past bubbles that we've seen. Invitya says Open AI can spend that money however they please, but undoubtedly a lot of it is going to go back towards Nvidia chips. So the concern is, does that muddy the waters when it comes to understanding how much demand there really is for what a company is making? And in this case, whether there is actually as much demand for AI tools as there seems to be right now. So you've got this tangled web of deals that has a lot of people in Silicon Valley noting the parallels to the dot-com era. Does OpenAI make anything that I use other than chat GPT? Like, what is all this valuation about?

They are producing things all the time. So just in the last couple of weeks, for example, they have unveiled SORA 2, which is this video generation tool that's made a lot of headlines. It's the most powerful imagination engine ever built. I say to you today, my friends, I have a dream that Tung, Tung, Tung, Tung, Sahoa will get a low tape of fade and it will cost six, seven barbershop hairquarters. It is loved by some, reviled by others who think it is going to just flood all of our feeds and our lives with AI Slop. They recently announced a new browser that will compete with Google Chrome. So Open AI is announcing live right now that it is launching ChatGBT Atlas and AI powered web browser. It's able to remember previous chats and searches.

There's even a mode that completes tasks for you. And so there are real things coming out of this company, real tools that people can use right now if they want to. The question is, what is the value of that? Is it actually boosting productivity in a way that is good for the economy? There was a study out from MIT this summer that cast doubt on that notion, finding that the vast majority of AI initiatives in the workplace aren't. helping companies turn a profit. And actually, I talk to a lot of rank and file workers in the

tech industry who tell me that a lot of times they're using these tools in a way that is somewhat performative just to be able to show their bosses that they're using AI, but they don't always feel the benefits in their actual work. So this notion of how much AI is actually impacting the economy in real ways, impacting the workforce in real ways, that doubt could go away. Let's see what happens. But I think there's a lot of eyes on the stock market where you have a bucket of American tech companies, many of them working on AI in some way or another, that make up 80% of the gains that we've seen in the stock market this year.

80% sorry, you just flew right by that. 80% of the gains that we've seen in the market have come from open. AI or all AI companies? From a handful of companies that are either working on AI in some way or are basically very much devoted to AI. And that is a metric that has so many people concerned right now. Right. Because if one industry, one unproven industry is making up that much a percentage of the gains, that means that if you are exposed to the market, if you've got your retirement in there, if you got any money in the market, you are exposed to this risk. What happens if we are in an AI bubble and it bursts?

What does that look like? If that were to happen, it would certainly be reflected in the stock market, you know, just to give you an example from the dot-com era. So one day I'm online and I find beyond.com. Cosmo.com has videos and ice cream. eBworld.com. We're inside your world. Pets.com, because pets can't drive.

When that bubble burst, Microsoft shares lost about two-thirds of their value. Amazon, which was then known as Amazon.com, almost went to zero. Oracle, which has been one of the big winners in the AI boom. Well, back then, their shares were down 80% when the music stopped. After what they are already calling Black Friday, they are counting up the damage on Wall Street. This closing bell might as well have been an alarm. Savage was the silly. Even the most venerable high-tech giants, such as Intel and Sun Microsystems,

we're now walking wounded. And countless dot-com vunderkins were now pushing up the daisies. Webvan, dead. Excited home, dead. Theglobe.com, pets.com, boo.com, dead.com. Dead. Dead. So with that in mind, somewhere on the order of two-thirds of Americans own stock, those accounts are going to be deeply effective.

if we are, in fact, in a bubble and that bubble were to burst. Retirement accounts would be touched by this. By one estimate, $20 trillion in wealth held by American households could be wiped out. And that is going to have impacts across the American and the global economy. We're talking about potential job losses and impacts on wages for jobs that are left. Is anyone hoping that we are in a bubble and that the, the bubble bursts. Is anyone rooting for this? There is a camp of skeptics, I would say, who hope the bubble bursts. Part of the artificial intelligence race is about achieving AI that can

surpass human intelligence. Some call it artificial general intelligence or AGI. Others call it superintelligence. And then you have people who look at this and say, is that a good idea? Do we really want intelligence that can surpass what humans are capable of. And you have doomers who think that this could spell the end of humanity as we know it. In addition to that, there are very real concerns about the potential ecological damage that running towards AI unfettered in this way can cause. Residents' utility bills are already going up. We have water issues out the wazoo. And we know that data centers are extremely energy and water intensive. A single data center building is using as much as a city worth of power.

So we could see everyday Americans contend with higher electricity rates, limited water availability. These are all very real concerns that we've already seen play out in various ways in different parts of the country. We started this conversation, Lily, by asking, are we in a bubble? And you said, often, if not always, you don't know until you know. what should we be looking out for? How are we going to know? Well, I'm going to be looking at what the ads on the 101 freeway say. That's my first go-to. Have we moved on from AI and are we back to touting the metaverse? Because that's generally where the trends come first. I think we, I'm imagining we would know in this case if we're seeing indications in the stock market, if people are suffering from job losses. we might end up in a situation where we have a whole bunch of data centers sitting empty across the country. That's one physical manifestation that we might see because we are currently in a relentless building spree. There is just frankly what feels to a lot of people like a lot of hype in this space.

and it's just interesting getting to have FaceTime with people like Sam Altman, because when you are in the room with him for even five minutes, the message is always, we just need more. We need more compute. And if you just make it happen, we can deliver so much more. The question becomes, is whatever they want to deliver, something that people really want and will use and will have impacts on us that will make us more productive,

that will ideally allow us to do more with our time. That was the BBC's Lily Jamali. Do know that Vox Media has partnership with Open AI. Many media outlets do. It does not affect our editorial decisions, though, in any way. Coming up, an argument that we are definitely in an AI bubble. It's just not the AI bubble that everyone thinks we're in. Support for today's show comes from Quo, formerly open phone.

It's hard to estimate how important customer communication is for a business. A quick response can be the difference between making a sale and not making a sale. A quick and personalized response can potentially win you lifelong loyalty. didn't ask me to give you any personal experience, but let me just say, I personally agree. Quo formerly open phone is a business phone system that says they can make sure you never miss an opportunity to connect with your customers. It's the same platform you may have heard us advertised before just with a new name. We call that a rebrand in the business. You can get started for free, plus get 20% off your first six months at quo.com slash explained. That's

QUO.com slash explained. And if you have existing numbers with another service, Quo will port them over it, no extra charge. Quo, no missed calls, no missed customers. Support for today explained comes from American Giant. Perhaps you are interested in finding clothes that look better the more you wear them. Clothes that develop a patina. Clothes that shape themselves to your body and lifestyle. Clothes that you can make memories in.

Clothes that wear in rather than. then wearing out. American Giant says their clothes are designed for everyday life and wearable for years to come. As those seasons shift and get cooler, you could try their hoodie, which they've named the greatest hoodie ever made, and which they say is constructed out of high-quality materials and cut and sewn here in the U.S. of A. Our box colleague, Andrew Melnizek, has tried American Giant, and here's what Andrew has to say. So American Giant just sent me their men's horizon shirt in olive green. Honestly, I mostly got that because I've been told I, look really nice and green. So it makes me feel good. When I got out of the mail, took it out of

the box, my immediate reaction is like, wow, this actually feels really nice. It has a quality feeling to it. Get 20% off your first order when you use the promo code explained at American dash giant.com. That's 20% off when you use code explained at American dash giant.com. Support for today, Explain, comes from rippling. What's rippling? You got to wait for it. Rippling wants you to know that finance teams can waste weeks chasing receipts, reconciling spreadsheets, fixing errors across a slew of disconnected spend tools. And that's what Rippling is here to fix, says Rippling, helping you keep your spending under control without the busy work of clunky finance software.

How's this for streamlined? Rippling is a unified platform for global HR, payroll IT, finance. They say they've helped millions replace their mess of cobbled together tools with one system designed to give leaders clarity, speed, control by uniting your employees, teams, and departments in one system. Rippling can help remove the bottlenecks, the busy work, the silos, your software might have created, says Rippling, automated, perfectly in sync, seriously simple to use. Rippling can give your company a one-stop shop for your people, their data, everything they touch. With Rippling, you can run your entire HR and IT and finance operations as one. And right now you can get six months free of all this when you go to Rippling.com slash explained.

Learn more at R-I-P-P-L-I-N-G.com slash explain. That's rippling.com slash explain for six months free. Terms and conditions, do apply, guys. You're listening to Today, Explain. Paul Kedroski is a partner with SK Ventures and a fellow at MIT's Institute for the Digital Economy. Paul has been causing all kinds of trouble lately, arguing that we are not paying attention to the right AI. bubble. We're assuming that AI is untested and overvalued. Paul doesn't think so. AI is obviously a hugely important technology. It's going to be hugely useful. It probably

already is useful. I know it is to me. It probably is to you. So that's not the point, though. The point is that we're in this moment where there's an AI capital expenditure bubble, meaning that we're spending this prodigious amount of money on data centers and on sort of the underlying infrastructure for AI with probably no likelihood of recovering most of that cost and a significant likelihood that most of those assets become worthless because of the speed at which they depreciate. Build baby build, as they say, that is the White House Silicon Valley slogan, urging companies to invest in artificial intelligence infrastructure inside the United States. Today we have $50 billion of data centers that we own or are building another $50 billion

pipeline behind that. What's driving that is AI. Around 2010, we were seeing about 311 data centers that was nationwide. And then by the end of last year, that number had nearly quadrupled to 1,240 data centers. But increasingly, and this is the piece that got me concerned when I first started looking at this, an increasing fraction of the money that's being spent on. All of these things that allow us to distribute AI like electricity is coming from. an increasing share of it's coming from debt. And it gets really, really complicated in a hurry. But the point is that debt comes with obligations.

You don't get to just walk away from it. So that makes this moment even more perilous. If AI is so important, if this is like a game-changing, life-changing, humanity-changing technology, why does it not make sense for trillions of dollars to be rushing in? Isn't this what we should be doing? Yeah, we should be. But the problem, of course, is that there's this idea of what's called like a rational bubble. Everybody thinks they're doing the right thing, but when you add what everybody's

right thing together, you end up with a prodigious amount of waste. So it's no different than if you go back to the 19th century railroad bubbles in both the UK and the U.S., there was simply too much track, too many enthusiastic railroad builders building almost adjacent tracks to the same locations. And this led to an incredible amount of waste, but it also led to company failures. It led to various market crises across the 19th century in the U.S. and repeatedly in the U.K. So it's not as simple as saying, well, this is important, so we should build it and not care what it costs and not care the consequences. If so many smart people think that we are in a bubble, why is money still flowing in to data centers and other AI infrastructure at the rate that it is? Why don't people pull back a little bit? I'm not convinced that many people think it is a bubble. As I talk to people in technology, the most common response I get from people is not only is this not a bubble, but it's probably the most important technology of our lifetime.

We have an opportunity to build a superintelligence, a godlike intelligence on top of all of these chips and buildings and this AI electricity thing we're creating. And so to say we should slow down at this point, according to the technology community, is just a huge error. So there are people outside of technology who say, oh, you know, this is an incredible amount of spending. The Bank of England said it last week. Other people are cautioning about it, but not inside of technology. The United States and humanity broadly certainly has had no shortage of bubbles throughout history. You mentioned the railroads. Walk us through some famous American bubbles.

So the railroad's probably among the most prominent in the U.S., and that was, again, an enthusiasm for the idea. And that created an incredible frenzy of new companies, of track laying. The same thing happened in the 20s during electrification. So in the 1920s, we went from a single-digit percentage of American rural areas, having access to electricity. And by the end of the decade, it was more or less ubiquitous. everyone had access to electricity. But at the same time, that gave rise to this proliferation of utility companies, of ventures that were doing all kinds of questionable things in terms of overspending.

And in part, you could argue that electrification and the frenzy around it gave rise to the stock market, rise of the 20s, which led to the crash of 29, and helped precipitate the Great Depression. Okay, so that was 29. We're talking a century ago. We have seen other more recent bubbles, yeah? Oh, of course. But people are pretty familiar with the telecom and dot-com bubbles. But the closest historical analogy to what's happening now

is genuinely is railroads and electrification. And that it's 100 years ago is in part why people's memories are so dimmed of it. And instead they babble about, oh, you know, fiber optic cables during the dot-com bubble. Well, that all worked out. These are all the wrong references. The right references to what's going on today, both in terms of scale and scope, was the railroads and the electrification.

There's likely only to be one winner. In the same way that we don't need to have two sets of tracks to Philadelphia, we probably don't need the same number of companies delivering what are called these large language models, these AI models that people are using. These will naturally shrink. We're only going to have a few major providers because it's very hard to differentiate.

Even if you try them today, I defy you to tell the difference between getting an answer from OpenAI, Microsoft, or Claude, Anthropic, they all more or less are already beginning to feel like the same thing. Paul, ultimately, how destructive are bubbles, and what do they tend to destroy? All of them leave behind scar tissue, do immense damage. It's just a question of the kind of how big the bubble is. where the damage goes, because it's almost the story of bubbles in 200 years of the, in modern finance, is that there's no such thing as a bubble that doesn't cause damage. And that's why I find it a bit strange when people say, oh, you know, it'll all work out because we'll have something left behind. And it's, it's as if they always ignore the idea that all financial, all financial bubbles have consequences, whether it's wealth destruction because stock markets decline or whether

it's because misallocated money causes money to go to one place rather than somewhere else. So if you're just holding an index fund and thinking you're being very conservative, you're actually soaking in AI right now. So if everything reverses those 20 or 30% in the other direction, you're much poorer than you were and that'll change your spending and that has implications for recessions. Yes, it does. But ultimately, things come back, right? What is, what is, tell me if I'm wrong, but what is strong,

survives, gets more nimble, gets more creative, markets recover over time, jobs come back, etc., etc. Like, isn't it always the case that the bubble bursts and then what it leaves behind is, I don't know, maybe not something beautiful, but something workable? No, no, that's not the case. That's kind of a line of pattern from the technology community and others in other bubbles, but the reality is almost every, Every financial, every technology revolution has caused huge damage and can take decades before we get back to where we were before. And as the famous line in economics goes, in the long run it may work out, but in the long run, we're also all dead. In the long run, we're all dead. It's a quote from John Maynard Keynes. If you didn't know,

Now You Know, Paul Kodroski, MIT, SK Ventures. Today's show was produced by Avishai Artsy and Kelly Wessinger. It was edited by Amina El Sadi, in fact checked by Laura Bullard, and engineered by Adrian Lilly and Patrick Boyd. I'm Noelle King. It's Today Explained. Support for today explained comes from American giant freedom you may know ain't free and cheaply made clothes ain't actually that cheap. Not only do they fall apart quickly and have to be replaced more often, their costs for the environment and costs for the communities that make those clothes.

The founders of American Giant said they decided to see if they could do things differently. So all of their garments, they say, are designed for durability to outlast a trend cycle. And they're all made in the USA. Our colleague, Andrew Melnizek, has tried American Giant. Here's what he has to say. In the box that I got, they sent me a set of pants and a shirt. Taking out of the packaging and tried it on for the first time was a good reminder that, hey, actually, nice quality clothing matters. It just feels nicer.

It doesn't feel like it's going to fall apart after a few washing. and I don't have to worry about the stitching or the color fading after a few uses. Get 20% off your first order when you use the promo code explained at American dash giant.com. That's 20% off when you use the code explained at American dash giant.com.