Is AI in a Bubble? History Says It's Complicated

Matt: [00:00:00] All righty, everybody, this is another edition of The Web News, and today we wanna talk about the AI bubble.

And we've seen tons of people talking about this. Is there an AI bubble? Is there not? When will it pop? Because there has to be one, et cetera, et cetera, et cetera. So I wanna be clear that this is not, like, some hardcore, like, lots and lots of research into economy stuff and all this. This is sort of an editorialized discussion about the A- about AI, AI bubble.

But also, I did do research into previous bubbles, and we can kinda compare and contrast those historical tidbits, if you will, into this AI bubble or not bubble or whatever. We'll get into it. So I'm just gonna read through this research really quick. Again, just, just brief research for editorial purposes, not hardcore dates and everything like this.

Okay. So I researched three other bubbles in history, and the first one is all the way back to the 1840s, and that is [00:01:00] Railway Mania, one of history's first major investment bubbles. Railroads were brand-new technology and promised to completely change transportation. Investors rushed to buy shares in hundreds of railway companies.

Many railways were poorly planned or never became profitable. This is a really key thing. I just wanna, like, a brief thing out of the script here. This is really key because if we think about railroads now, we have data on that, right? Like, we know where we should put railwa- railroads, where we should put rail yards, where the supplies have to be, right, how often to run the trains, how often to have the...

or how long to have the, the train cars and things. When the railroad was new, think about the investment in just people working to build a rail and imagine it going to some place that would never have worked. It would never have actually B- bore fruit, if you will. And so now there's just, like, a rail that goes to no town because they thought a town would go there.

Or, [00:02:00] you know, we, we, we didn't have enough, we didn't have enough supplies along this road, or this road was too, or this route was too dangerous, this rail was too dangerous. So you have to remember, like, this is in the beginner, in the beginning of, like, the railway, and so we, we as humans, even though the experts, even though railway experts are, quote-unquote, "experts," it's still a new technology.

And so you're not gonna know what's gonna happen. So by the mid-1840s, the bubble burst and investors lost huge amounts of money. The railroads themselves were not a failure. It is clear as day that railway, that railway, railways, railroads, whichever, were going to revolutionize the world. We knew that. We knew it was gonna revolutionize travel, and in Canada and in the United States, it was going to help settle or tame the West as it was, as we learned in our history classes.

And of course, they went on to transform trade, travel, and the economy for generations. You can move a whole [00:03:00] bunch of stuff with a railroad. But unfortunately, even though we knew it was revolutionary, it still had a bubble that popped. So let's go to one that's a little more modern. In fact, quite a bit more modern.

We're all familiar with this, I think, called the dot-com bubble, but many people don't know really what it was. They just know that there was some sort of... Everyone... I, I was told by some people it was everyone was buying dot-coms, and then they realized that that, that that was bad, and then they stopped. And I was like, "Well, they weren't spending billions.

Like, each person individually wasn't spending billions. Like, there's, there's more to this story," and of course, there is. The internet back in the n- late 1990s to the 2000s was, to the early 2000s, excuse me, the internet was becoming a mainstream utility, and everyone wanted a piece of it. Companies added dot-com to their names, and investors poured money into almost anything internet related.

Many startups had little or no revenue but were worth billions on paper. In 2000, the bubble burst, and tech stocks crashed, wiping out trillions in market [00:04:00] value. Thousands of companies disappeared, but companies like Amazon, Google, eBay, and later Netflix came out of that era and helped shape the modern internet.

The internet, once again, was a revolutionary piece of technology that we all knew was gonna be revolutionary, or at least we thought. There was obviously some skepticism. There always is. Some skepticism, some pessimism. But the people who were building on the web or understood the web or been, had been using it understood that this was gonna be a revolutionary tool, and even then, we didn't know what websites to build.

We didn't know the best ways to do e-commerce on there. We didn't know the best ways to conduct business on the internet. So from the ashes of that unknowing, we build what we have today, which is we know when to do e-commerce. There's SEO. There's certain ways to build websites. There's web apps, and we've sort of figured it out.

Not perfectly, of course, but again, from those [00:05:00] ashes, we build what we have today And then this one, I wanted to do one bubble that was not technology related, and that is the financial crisis in 2008, otherwise known as the housing bubble. So this was not driven by technology, again, housing. Home prices rose rapidly because people believed that they would keep climbing forever.

Banks made increasingly risky mortgages and packaged them into investments that were sold around the world. When homeowners started defaulting, the financial system began to unravel. The result was the global financial crisis, major bank failures, and a worldwide recession. So this is a lesson here.

Bubbles don't just happen in tech, they happen whenever speculation gets too ahead of reality. So it doesn't have to be the new phone, the new app, the new internet box thing at the time. It doesn't have to be that. It can be people speculating and saying, "I'm just gonna keep dumping money into this thing.

It'll [00:06:00] never go wrong. Nothing will ever go wrong here." Famous, famous last words, I would say. So now that we've kind of gone through a little bit of the research, what we want to talk about is, you know, where AI is at today. Are we in an AI bubble? Are we not in an AI bubble? All that kind of stuff. Mike, why don't you take it away?

I've been chatting for quite a while

Mikhail: Yeah, absolutely. I think one thing I wanna go back to is the dot-com bubble, 'cause I think that's gonna be the most relatable bubble to all of this. And I will say that this was not a bubble that got it wrong, right? Like, they weren't, they weren't wrong about dot-com. They were just wrong about some of the utilizations of the dot-com thing, and they, they made bad bets on a wide horizontal scope of companies, Their, their idea was that every, every one of these companies might make it, or they were already a- aware that only a certain percentage of the companies will make it, and they were trying to just s- make their bets safer by investing in a lot of different [00:07:00] companies. That's what VCs were do- were doing. And they, they were right. Like, if that was their assumption, they were right, that a lot of companies that were worth mil- millions or even billions sometimes, um, not make it. But I believe that their expectations were actually undervalued to what it became. if we look now, the, the highest value companies in the world, all of the highest value companies in the world are all tech companies in this kind of dot-com space. So they thought it was gonna be big. I don't know if they thought it was going to be that big, Now, there was a correction, and I think that's where kind of differ in calling it a bubble, because, yes, there was definitely a ti- a period in time where nothing was It was harder to get investment, and there was, uh, uh, you know, companies that didn't make any val- didn't have any value were falling left, right, and, and [00:08:00] center.

So that, that I consider a correction, and I do see that parallel in AI right now very closely. I think investors are smarter now in the sense of, like, they know that this is gonna happen. They know that there's gonna be a correction, so they're investing a lot of money. But again, they're make sure, making sure that they have their bets in, in the right places. companies are still gonna try to get g- still gonna get mass- massive investments they don't deserve necessarily. the reality is is that, yes, if AI goes the way of the dot-com bubble- We might not see a burst. We might see a correction. Well, we, we probably will see a correction of some sorts depending on the economy, but we will not see a burst because the re- the reality is that potentially what AI brings in terms of value will outpace even the dot-com bubble. So [00:09:00] they, all those companies, the Anthropics, the OpenAIs of the world, are set up to be the most valuable companies ever on t- over top of the tech companies before them. Now, obviously, the tech companies are trying to play into it, but if you think about it from a growth perspective, investing in OpenAI, know, a year ago or two years ago when they were getting a ton of investments, the investors' mindset was, "Hey, if they're gonna be bigger than Google, then my investment is gonna go up 10X, 20X, 30X, 100X over what I could get right now."

So the, the risk reward there is quite high. I, I-- The, the, well, the point I'm trying to make is I see what people are doing with the investments. I don't know if it's gonna be a catastrophic bubble burst. I don't think that it's gonna be as catastrophic as people are thinking. It might be more of a soft landing.

I know that, that approach, that, that term is used a lot in the econ- the econ, the current economy that we have, uh, [00:10:00] versus like a, a complete correction just due to where the trajectory of AI, right? Like it's if I'm looking at it right now from utility perspective, looking at the value that it's bringing, it's high.

Like it's very, very high. Like it's, changed the way I work 100% across the board from, you know, even a year ago to now, and it's only getting better.

Matt: But though I would actually, like, pose a question to you there 'cause, so you, you told me the other day when you purchase a plan from, like, OpenAI or Anthropic and you purchase a... I forget the ratios you told me, but let's just say you purchase a $100 a month plan, you're getting, like, $2,000 or whatever it is in, in AI credits, and that is being subsidized by investments and, and, and things like this, like angel investors and things for the time being.

'Cause obviously they're trying to get the compute to eventually come down. So effectively what's happening here is they're getting hit [00:11:00] in the front and the back, these companies. So they're getting hit in the back end because there's a lot of cost and they can't charge a person like me $2,000 a month 'cause I'm just not gonna use it, so they have to charge me 100.

And I'm not even paying that, I'm paying the $28 Canadian, which I think is like a $20 American plan. I'm paying for that currently. I might upgrade soon 'cause we have some work to do. But then you're also getting hit in the front because people are skeptical on AI. There's lots of people that hate AI, and they d- really don't like it.

They don't like the, the data centers and all these type of things, and there's, there's fair arguments in there too. So they're getting hit in both the front and the back. And then what your comments are is you're saying, "Well, it's a revolutionary thing and it changed my life." And I would say that's the exactly the same thing that the people who are investing in the railroads said.

This is a revolutionary thing. This is absolutely gonna be amazing. This is gonna tame the West. It's gonna do this, it's gonna do that. It's gonna be amazing. This is gonna be, uh, you know, absolutely the, the, the best thing [00:12:00] ever. I don't think there's a question about AI changing things and changing the world, changing the world...

I, I don't, I don't wanna say forever because there's always, like, a new toy, but there's no, there's no... I don't think there's any question in my mind that AI is going to change things in some way with a lasting effect. I don't think there's a question there. I think the question is how much and in what, what areas?

Because- So let me ask you a question. Let me ask you a question then, 'cause I said I was going to. Do you think that a, this AI bubble... 'Cause I think it is a bubble to some extent. I know you're saying it's, it's a correction, but a bubble is technically a correction. We can get into all the... We're not economists, but my question is, if the bu- if the bubble were to burst next year, do you think that these companies would survive because they're getting hit on the back end [00:13:00] by selling us compute at a loss?

Mikhail: So yeah, let, let me clarify that sell and compute at a loss statement. So The reason that we say that they're subsidizing is that we're comparing the that we're getting from the $200 subscription to what it would cost the same amount of credits from just an API.

Matt: Okay

Mikhail: know, what we don't have a clear indication of, is if the API credits being sold with profit on top of it,

Matt: I see

Mikhail: right?

So we don't know that part. We don't, we don't know how much it's costing them to actually serve those credits. So to them, it-- to, like the API side, yes, the two-- the $200 subscription in Codex gets you like $19,000 worth of credits. That seems like a lot, and it is a lot. to them, maybe it's only a little bit of subsidy.

Maybe it's only $500 extra of subsidy from the actual inference perspective. Now, there's many different aspects of this. So one is that they are trying to price into the API, and [00:14:00] to the subscription, honestly, uh, a return on training. So training is separate from inference. When you're training a model, you have-- you-- it takes up a bunch of compute, it takes up a bunch of GPUs, and it takes a lot of time, and it costs a lot of money, right?

So they're trying to... They're probably trying to bake in profit margin into their services to bring back some of that money,

Matt: Sure

Mikhail: right? Obviously. And they're baking in probably a little bit of profit on the actual inference costs. So we don't know what their actual, like, you know, compute inference costs are quite yet. might be making a profit at this point on the API, and they might be making a profit on the subscriptions because, as we all know, people will buy $200 subscriptions and only use like 10% of the usage. Now, that's outweighed by like-- W-when, when I say that, like, you-- to get, you know, you can get $19,000 for the $200, that means you're using the, the subscription to its maximum [00:15:00] potential at all times of the day.

So you're burning your limits at all times of the day. So you have to know how to do that. That, that's like a skill. That's called token maxing,

Matt: Yeah. Yep

Mikhail: You have to really-- Most people are not doing that. Most people are using it you know, for their work,

Matt: Mm-hmm

Mikhail: think averaging it out, I bet you they're not losing that much money on the subscriptions.

I, starting to think that that's the case, that we're gonna see a profitable AI company fairly soon. I think Anthropic is already kind of putting themselves in that position, they're going to come out with like a revenue, like a, a, a positive revenue. And so t-TLDR, my answer to your question is, I think the big companies, Anthropic, OpenAI, right?

The, the new big companies, I think they're going to survive the, the correction Um, the companies that won't survive the correction are the really small companies that don't provide the [00:16:00] value yet, Like if-- And the, it all depends on when the correction happens. So these companies are betting on like, like cheaper tokens or whatever and stuff like that for

Matt: Mm-hmm

Mikhail: cheaper inference, whatever. If the correction happens all of a sudden because of an economic downturn, like w- uh, an external factor, and they are not there yet, they're gonna die. just the reality. They're not gonna get any more investment. They're not gonna be able to pay their debtors, and that's it. They're going to die. the, that's the reality of the situation. But these large companies, that's where all of the investors are going to shift to. They're gonna stop, they're gonna lower their risk by a ton, going to shift their money back into these larger companies that have already gained the market share and

Matt: Right

Mikhail: gained all their money until they start to make a profit. And I, I think we're closer to that than we think, like closer to them breaking even on some stuff. Um, th- they're burning a lot of money still, obviously, because they're also doing a [00:17:00] data center and, uh, data center build-outs. They're also investing in hardware. There's so many different verticals that they're putting their money into, they could probably adjust that in a different market as well.

Like they could probably stop some of those projects, you know, slow down on some things, and just focus on core business for a little while to s- to, to weather the storm if it needs to happen

Matt: Well, I, I do think, like you, you said that we're probably better at handling bubbles, and I think we have more information than ever because, well, every, every bubble experience you're gonna have more information 'cause you can look back and see all the other bubbles that have come and gone. So you can adjust things and, like, kind of design your economy, your investment strategy and things to try to mitigate it.

But I, I, I do-- I, I still speculate here because I, I, I get what you're saying with the API, the API costs and things like this, but we are seeing companies that are using coding. They're getting their, all their engineers, say they have five engineers, they get all their eng- engineers their, their Claude subscription or their Codex subscription, and they're [00:18:00] losing money like crazy.

Like, they're spending money on the engineers ripping through tokens and ripping through compute at the end of the day, ripping through compute like, like crazy. And we're starting to see people start to complain about that and how, how expensive it is, and they're complaining about how expensive it is now.

And we're, we're assuming, like you're saying, it's a bit of a black box, but, like, l- I would say, like, let's assume it's being subsidized by the investors. And even if it is not, so if it, if it's being, if it's being subsidized by the investors, then the price is only gonna go up. So you're complaining about the price now?

The price is only gonna go up, so that's not good, and it's gonna go up, up as an actual price increase, not just natural inflation. And then if-- Let, let's just say they are profitable, and these, these tokens are being sold at fair value or, or minimal profit or whatever it is, and they're being sold at that, at that.

That's also still not good because you're already complaining. So I, I, I just, I... What I actually f- what I actually think is gonna happen is I do think that we are gonna see a, a, [00:19:00] a bubble pop. We are gonna see, kinda like what you're saying, a correction. How drastic, I don't know. Because we are seeing a lot of, a lot of, uh, different things come from AI that we don't know if it's gonna play f-- like, have, bear fruit.

So, for example, something that may, that is, that we're working towards right now is trying to make it so that there's, like, a robot in your house or several robots in your house that can help you with domestic tasks, the laundry, et cetera, right? Clean the floors, what have you. Th-- I-if in the event that a bubble pops, those companies have not gone into the market super hard yet.

We don't see that everywhere. And so those companies are likely to suffer or completely die out, and then we w-- and then all that compute and all that stuff is no longer being purchased from these data centers. You see, it's like a bit of a ripple effect, right? Now, I'm not saying that these domestic robots are a [00:20:00] major portion of the compute.

I'm, I'm not saying that. However, there's many little projects that are trying to grow. Domestic robots, you got the AI pins, people testing out AI phones. There's AI glasses and things like this. People testing out open source models, people testing out local, locally running models, people testing out all these things.

And I think that by us having all these options that are all in their infancy, we're doing the exact same thing that happened in the railway b- the railway bubble and in the dot-com bubble as well. We're actually seeing this happen again because you're speculating. Like, in the railway bubble, it's like, "Should we build, should we build to...

There's a meadow past this, this, this mountain. Th- going through the mountain's gonna cost us $100,000," and this is the 1800s, so it's crazy. Tons of, tons of people, tons of, tons of supplies and everything. "Should we go through this mountain? Because the meadow's beautiful. Maybe we should build a railway to that, that meadow."

You blast through the mountain, you put the tunnel in, you put the [00:21:00] bridge over to the meadow, you get everything going, and the people go, "Why the heck would we go to that meadow?" And that entire line is worthless. And, and w- that principle can be seen here, where we've seen AI pins and people go, "That's stupid.

Why would we do that?" 'Cause the market also dictates a lot of this stuff. Where's, where... If people, let's say, go back to the domestic robots for a second. If people are not ready or don't want to pay for or don't or, or do not want domestic robots in their home, then that market is not going to work out.

Maybe in 10 years people would be like, "I'm sick and tired of cleaning my floors," or whatever. But I would also like to say something else, is that the homes in Canada, like, many people are buying, or many people have bought, the condo market's in, like, a weird spot right now, but that could change, of course.

But many people have bought condos and Mike, like, you're, you're in a condo now, and these condos are not big. So would you really [00:22:00] want to s- pay $6,000 even? Let's say, let's say $1,200. Is it worth $1,200 for not that much square footage to just use a vacuum real quick? So if you have a lot of people that are buying smaller places or renting smaller places 'cause the cost of living, which is an external factor we haven't mentioned yet, but the cost of living, at least here in Canada, is skyrocketing.

So people are not buying the biggest houses that they can and the, and renting the biggest apartments and buying the biggest condos. They're, they're having to sort of downsize, you know, before they've even gotten started. And so, like, to me, if I have two rooms, or if I'm in a bachelor apartment with one room and a kitchenette, I don't need a robot to clean that.

It's gonna take me five minutes. And so the market comes in and dictates this stuff, right?

Mikhail: Absolutely. I, I, there, there's a lot of speculation on the robotics stuff. I think that there's, you know, there's going to be some failures there, a lot of them. reality is I, I, I do think that a lot of that is priced in. We know that that's gonna happen. We understand it. These large [00:23:00] companies understand it as well, that, hey, this is

The correction will happen. Uh, so we need to be prepared for it. It ... Like, we're, we're in full-on expansion mode now, but I'm sure they have contingency plans for, like, a contraction. where I, that's where I think that, like, this is a different situation they, they understand that this is a bubble.

Matt: For sure

Mikhail: fully aware of that situation, and they understand that a lot of what they're doing is experimental. Like, they have many, many different runs going for each one of those models that they create, like each one of those models. And a lot of them, they just throw out, and they'll ... Like, they'll throw out, like, million-dollar runs, like millions of dollars of compute.

They'll just be like, "Nope, it's bullshit. Throw it out." They understand that that's the case, but the end goal, like, where they're heading is that, hey, this is going to replace all of humanity's, you know, work. That's where ... That's wh- that's the price points that they're starting to get. [00:24:00] Like, they're starting to get infrastructure price points.

Like, they're not, they're not getting price points for being a ChatGPT or, or even a coding model.

Matt: Sure

Mikhail: getting, they're getting, like, evaluated as if they're going to be the next, like, internet provider or the next water provider, and, like, no one will be able to live without it. It's, it's tough for me to say that that's ever gonna happen, but it's t- also tough for me to say that the market will ever, like, back, like, turn around from that because turning around from that

Like, what they've done is they've all started to rely on each other. NVIDIA relies on OpenAI. OpenAI relies on, uh, you know, NVIDIA one to one. Google relies on Apple. Apple relies on, you know, Google and NVIDIA and everything else, and Qualcomm and all that stuff. Like, they're all It's a very circular economy that they've created, so that if one of them does go down, it's going to be cat- so catastrophic that they can't let that happen. [00:25:00] Like, the- these massive giants can't let one of their own giants go down. So they, they would probably come in similar to how Microsoft came in and bailed out Apple back in, what, the '80s, I think it was?

Matt: Ba- yeah, back in the Pirates of Silicon Valley days anyway

Mikhail: Back in those days, this, the same situation they, would happen because they can't let it happen.

They can't let a giant fall, and the small companies will fall. There's no doubt in my mind that small companies are gonna fall. They always

Matt: Assuredly

Mikhail: But that's not going to cause a massive explosion in the industry. That's just gonna cause another company to come up from the ashes of those smaller companies.

I, I've s- I've, I used to think, like, i- if you asked me six months ago, I would say, yes, I think OpenAI would fail. that that was gonna be the inevitable one out of all of them, and that would be the one that would break the camel's back and cause the bubble. But the more it's gone and the more I'm seeing them being a little bit more fiscally responsible after they've hired their [00:26:00] CFO, like their new CFO, and after they've started getting ready for an IPO and stuff like that, starting to come to -- I'm starting to turn around on that to think that they're probably gonna be okay. Like they're, you know, they're starting to think about scale just like I thought they would. They're starting to figure out other avenues of income. They stopped worrying about being like only a ChatGPT, now they're more of a coding bot. Like their fi- their fi- they found their revenue path after looking at Anthropic already. And I don't-- Being in the midst of it kind of hurts me, like in, in terms of being, um Uh, just looking at it impartial, impartially because I'm in it. Like I am using this shit day to day and I'm seeing the progress happen.

Matt: Sure

Mikhail: And I, and I also know how many people aren't using it. I, I had a recent thing with my dad where, um, I, I, I spun up a server for [00:27:00] him, like a Linux box to, to do like Plex and, you know, organize his movies and stuff like that. And I put l- Ubuntu on it, and I put Codex on it so that I could get Codex to help me with some of the configuration. And him and I were just sitting there and I was like, "Oh, this, something's not working," and I just started typing it into Codex. I mean, and like just talking to Codex like I would like an IT, you know, an IT admin, assist admin. And he's like, "What do you mean? Like what, what is it gonna do? Like isn't it just gonna tell you something from the internet?" I'm like, "No. Like it's gonna go in and look at the logs of this box and figure it out itself, and then if I need to do something it'll interject or it'll just do it itself because it's Linux." And I gave it pseudo permissions. And he's like, "What?" And then it just went, and then you just saw it happen. Like you just saw it go into logs, you saw it start to, like do Linux commands all over the place, and he's like, "It can do that." And then literally l- last week I was talking to him and he's like, "You know, I started to do that.

I started to treat it like a person, and it's doing like a bunch of troubleshooting at my job now. Like and like I'm, I, I've been [00:28:00] doing like more work and I've been able to do more complex things like, uh, DB admin stuff and stuff like that that I was never able to do I just like ask it to do things like I would an actual programmer or something. And then I just paste it in or whatever and do like regex edits and stuff like that." So I think there's a good chunk of society that's still not doing that, like a very large chunk of society. And once they start to do that, there's a lot more potential for revenue there, and then that's, again, that's one use case. There's gonna be more use cases of it, and it's going to become more, whether you like it or not, that's the other part of this, is like yes, there is some pushback from society, but it-- we're not seeing that affect and capitalism. We're just not seeing that on any scale yet. That could change

Matt: It-- Well, one of the things I, I do-- Like, I'll concede a point where, like, absolutely a bunch of people are not using this thing, and the adoption rate is not where they'd want it to be, of course. Not everyone's [00:29:00] using Codex. Many people are more or less using AI just as like a Google replacement. They're using the little AI overviews or maybe they're using ChatGPT.

At least from the people that I know in my sort of small sample size of a social circle, certainly they're not using agentic AI or anything like that. So I, you know, I concede that point. But I do wanna say that, like, there is something unique about this bubble, at least to me, where-- 'cause I still think it's a bubble.

There is something, there is something that i- is unique to this bubble, is that this is the first time ever that we're building something that doesn't involve humans working around it. It's working to actively replace a whole bunch of humans. I know we've had automation scares in the past and things, but it was always, "Oh, instead of having ten people on the factory floor, you now have four."

This is now the brain part, the intelligence part is being replaced, and then robotics has, has all but come, has all but caught up. So if [00:30:00] they're able to get good robotics with good intelligence, now there really is a question of why do we have anyone on the factory floor? Like, that is a, that is a question.

So I'll, I'll concede that because for the first time ever, we're building technology at a general intelligence level where it's not just like, I built this little machine that allows, uh, that allows a plane to have autopilot, but we still need pilots in the plane. Now it's like, oh, like this machine actually makes a whole bunch of decisions and decides a, decides a whole bunch of things and, and looks at a whole bunch of...

Like it look, it does like a, what a human does, right? It looks around and does those things. And so, like I, I concede that, that that's fair. But I do wanna say a b- a major part of the reason why I still think this is a bubble is I think that we assume all this is gonna go off very smoothly, but the, I think that, that we've built up too many things.

Not only have we dumped money into this thing and there's gonna be parts that die off and things, and I don't know if OpenAI is gonna die or [00:31:00] Anthropic or anything. I don't have any speculation on, at, to that micro level. I do not have speculation there. I mean, it's possible that two of the big titans suffer in a, in a, in a, uh, what do you call it?

A, um, correction and a bubble pop, and they buy each other, they merge or something like that. Like, that's always possible. We've seen mergers all the time with large companies, of course. Big companies buying little companies, two big companies becoming one, et cetera. There's lots of different financial ways for these companies to sort of survive straight up or just kind of live on through another company, I guess.

The thing here, though, with this AI bubble is I think we've built up so, so much And we're now talking about changing everything. Like, the people who are, like, super pro-AI or the people who are, like, thinking that AI's gonna become s- like, im- immensely more powerful, immensely more capable, immensely, immensely, m- uh, more disruptive, that disruption is gonna be disruptive.

I mean, it's gonna be a thing where, like, do you think that the government's gonna be ready to [00:32:00] go when 50% of the people are unemployed? How about 70%? What about 90? And here's the thing, our financial f- our financial system still exists. People still have to pay their mortgage, but if no one's working, who's paying their mortgage?

So now what does the bank do? So that's bad, and that's a whole bunch of stuff that's gonna have to be figured out. Are we gonna get some sort of dividend from the AI, like a citizen's dividend? I've heard that, that term be thrown around. Who knows? Like that sounds so foreign to us, citizen's dividend. So then some people say, "Well, we should have mincome."

Okay. But some of these people were making a quarter of a million dollars a year. Heck, you know, $150,000 a year, maybe both spouses $150,000 a year, bought a, a decent swath of land and then bought a really big house. Is the mincome for both spouses gonna be able to afford whatever kids they have and then the big house and all the land that they bought?

So are they gonna have to then downsize? Well, that's g- that hurts the economy right there. Just somebody downsizing like that and not pumping the money back into the economy like that immediately hurts the, hurts it, hurts the economy, and then you're gonna see that over and over again [00:33:00] potentially.

You're ... Like, we're, we're gonna see a reset and, like a potential reset is what, what is potentially being advertised here. So AI maturing itself could, very speculato- very speculatory, but could bring a b- bring upon its own bubble pop because the economy's not gonna be ready. I would ... I'm gonna, I'm gonna estimate the economy's not gonna be ready 'cause we humans are very reactive.

We do not do things proactively. And so we're gonna wait until unemployment hits an unbelievable level. We're gonna be running around with our heads cut off, not knowing what the heck to do, and we have to remember that there's a bunch of other market factors 'cause the market just keeps on ticking. You got real estate crises in certain countries.

We got cars that are super, super expensive. We got gas that's super, super expensive, and we keep being told to go on EVs and everything else, and the grid's not ready. So we have all these other things that are still happening. Now, of course, in a [00:34:00] complex modern society, naturally speaking, you're gonna have a bunch of complex issues.

And if you look at them in the, through the very worst lens, through the most pessimistic lens, yes, every single one of these issues, or at least many of these issues, are existential crises to the industry they're in or to humanity or whatever the scale is. Obviously it differs based on the problem. But what I'm thinking is, is that we are, we already have a bunch of these problems, and we've had these problems, and because of this we've had bubbles pop in the past.

And now we're talking about completely taking the system, flipping it up over its head. Meanwhile, it's all held up by a- angel investors and we go, "It should be okay." We've basically taken the scaffold on the top with the Earth on the bottom, we're flipping it over and being like, "Yeah, the Earth should w- should be held up by that scaffold.

No worries." And I speculate that that's not gonna happen. I speculate that that is not gonna go smoothly, and it's gonna be a complete bubble pop in my opinion

Mikhail: So, I, I think I'm not gonna play [00:35:00] devil's advocate to this fully because this is the biggest concern that I have as well. Like, a g- a societal collapse versus a, a company collapse or, like, an economy collapse, I think that's the biggest risk. The AI companies are aware of this, and what they've done is they've embedded themselves into the economy to such a level that governments can't go out and just start banning AIs massive repercussions on their own economy. So like, if US came out tomorrow and they're just like, "We gotta, we gotta pause AI, and we gotta stop new models, and we gotta take away models." Like, no more AI in the workplace. Let's say they do that tomorrow. That economy collapses.

Matt: Yeah

Mikhail: they're i- they're, they're reeling from an econo- economy collapsing situation right away, right off the top. They've done that on purpose. Like, these AI companies have done that on purpose. They've embedded themselves with the economies. They've lobbied the right people. They've gotten to the point where they can [00:36:00] get-- Th- they can, they can scare the government to not ban them, That's one thing that is a little bit scary, but also on the, on the other side of it is what we talked about before the podcast, was the debt ceiling. I believe what, what's going to happen is we'll see unemployment go up, we'll see these sti- stipends come out, and they're just gonna, they're just gonna crank money out, out the printer like crazy. gonna go through the roof, right? That's just inevitable, and they're just gonna try to keep everything afloat for as long as they possibly can. I don't know what that's gonna bring. There's probably more complex things that they're gonna do, but I don't see any other solution to this at

Matt: Well,

Mikhail: point

Matt: see, this, this is one of the issues here, is that one of the, one of the customers of OpenAI and other companies are consumers. If the consumers have no money, they don't pay for the subscriptions. Now, yes, maybe they have big government contracts where, [00:37:00] oh, I'm gonna have this ChatGPT-like thing control the, the power grid or something in the future or something.

Like, granted, fine, AI will... And, and you're saying they're embedding themselves into places, and large companies always lobby governments and things. But I, I, I, I'm just w- I

I'm at a point where I'm starting to, like, realize something, and that is that Like at what point i- is this invention for humanity or against humanity? And you're saying like, "Oh, you know, lobbying and all this stuff." There comes a point where it might have to be a scuttle in order to stop like mass riots or something.

Like think about this, if everyone suddenly becomes very, very poor, impoverished, and you're, you're gonna start having unrest. We've seen it in many countries, including our owns, our own countries, North America and everything else. And you'll start ... Y- there's a good chance of having some unrest, and I think they're gonna s- they're gonna realize, "Hey, we can't have our cities figuratively or [00:38:00] literally burn be- just because inst- just because we want it to go from 1 trillion to 2 trillion."

They're gonna have to realize that ' Cause at, at what point, it, it, with, with railways, with the dot-com bubble, with the housing bubble, it was all for people. People live in houses. The economy and all that, there was a bunch of greasy stuff happening, but people live in houses. People use the railway. They travel on the railroad.

Mail's transported. Goods are transported to people on the railroad. The dot-com bubble, they're trying to make life easier. They're trying to get access to information, the information age. What's the AI doing? Like, at some point we have to realize that, like, it's a human invention, and I would hope it's gonna serve us, right?

And then you start getting into the super intelligence thing, and it's gonna kill us, and yada, yada, yada, and we start getting into that point. But what I mean is, is like, I wonder, like, there, there... I, I believe that there is a point in which lobbying only goes so far Like if you're like, "No, no, no, like, like it's okay, 90% of the people can be unemployed.

Don't [00:39:00] worry, OpenAI will be worth another trillion next year. Don't worry, guys." Eventually people are gonna be like, "What in the heck is going on here?" And it's not gonna be nice. It's not gonna be a nice solution

Mikhail: the AI companies themselves understand this, and that's why they're the ones that are proposing this like min income and all that, because they understand that it, they can't survive unless there isn't the-- if there is unrest in that situation,

Matt: Right

Mikhail: right? So they're the like, you know, Anthropic themselves are like, "Hey, maybe we should tax credits." And OpenAI has put, put money into universal

Matt: Oh,

Mikhail: income

Matt: classic pass it, pass it to the, to the consumer classic. Yeah

Mikhail: Correct. Yeah. Well, no, no, no, it's, uh, the credits would be taxed from, like, from the profits for the

Matt: Oh, I see. I see

Mikhail: yeah. So, like, each credit would, like, 20% of it would go into a, a humanity fund or whatever like that. I can't remember what the actual numbers are.

Just go look it up. But Anthropic has talked about that. Dario Amodei has talked about, like, let's, like, today let's start charging, like, putting a tax on [00:40:00] every single credit, every single token. So, like, they know this fact. They understand that if there is unrest, there won't be companies. So they're gonna do everything in their power to make sure that there isn't full unrest.

Now they're gonna, they're gonna redline it. They're gonna absolutely redline it, so they're gonna make it so that it's, like, the most profits versus unrest that they possibly can. So there will be some suffering, there's no doubt about that. But they're, they can't let it ha- they can't get it to the point where it's actual riots in the streets and people can't purchase their products.

Matt: Well, also because why would you have an AI manufacture products for people that can't afford said products? Like y- y- you've broken, I think you've mentioned a loop in the beginning, and I ca- I think it was in this context you mentioned a loop. But it's like obviously you have companies that make products for people, and then those people buy those products with money that they get from a job from those, by working in those companies.

Like it's, it's a cyclical [00:41:00] thing. If you break the chain, I mean, who are you making... Like Walmart, oh fantastic, we made more Rubbermaid containers for three cents, but everyone has one penny. It's like well

Mikhail: Here, here's the utopian vision. I, by the way, I'm more of a doomer. Like, I, I would say I'm more on the doom side than the g- like, I think I'm taking a devil's advocate approach just to make this a little bit more realistic and interesting. But here's the utopian situation. Company-- Again, these AI companies realize that profits go up the more people consume. To get people to consume more, we have to work less. That's the reality. So we just lower the amount of work that people need to do because AI can do more of it. We give people min income, we give-- we, we, we figure out a way to change society in some way, shape, or form. This is why I'm saying utopian, because it's probably not gonna happen. But we figure that out, right, quickly, then people will work [00:42:00] less, have more money, have more time to spend, and therefore economy keeps booming. That's, in my mind, that's probably what they're trying to achieve in some way, shape, or form. don't know how achievable that is in a, in a, in, in our, in today's society. not achievable. Um,

Matt: Well, we have to be more proactive with stuff. Like that's, that's the thing here is in order to achieve, in my opinion, the utopian vision. Like is the utopian vision possible? I mean,

Mikhail: Maybe

Matt: But also we would have to, we would have to become less reactive and more proactive. Like doing that tax thing like you're saying, and I don't know, stashing that away or something and getting it prepared for a citizen's dividend or whatever ends up becoming the case.

I mean, that's proactive, where they're like, "Oh my God, we need to do a citizen's dividend." They're not suddenly taken aback and they have no money set aside. They have money set aside for that. Okay, so now we could potentially do a citizen's dividend

Mikhail: I mean, and it's, it's not unheard of. Let, let me be clear. Like, there's c- there's countries out there that do have that. [00:43:00] They have, like, a common weal- like a, you know, a citizens' wealth fund. There's plenty of countries that have that, like the, you know I don't wanna I know m- some, there's some Middle Eastern countries that have, like, a, that they, that they generate out of their profits from oil.

Matt: Oh, sure

Mikhail: l- and for the citizens specifically of that country, they set up, like, bank accounts, and they s- they set up education accounts and all that, and, like, you live a decent life as a citizen. Now, those are smaller micro-societies, I would, like, much smaller populations, and they, they're very limited on, like, immigration and stuff like that. That's not what we have. But there are examples of it, is what I'm trying to say, of, like, maybe something like that working, um, in,

Matt: Well, we have a welfare system. Like we have like a safety net to some extent

Mikhail: we do. So we just have to, like, think a little bit bigger on that. Like, we, we need to be more proactive and consider the implications of all this happening right now for it to actually be caught in time, which we're not doing, or maybe we, I don't, I, I don't think we're doing, so [00:44:00] it's kind of a moot talking point.

But, like, is a way for stuff to happen without devastation in society. I hope the right people are in power to, to handle that or like that at the time when it needs to be handled. You know, obviously my, hope is very, um, optimistic, I would say, and don't know. I, I, I think, I think that's, that's all I would have to say on the bubble. Like, hopefully a lot more knowledge is put in place right now than it was before. There is a way to do a soft landing. Hopefully we see it.

Matt: Uh, I am not really a doomer or, uh, hyped on it, I think. I usually am, like, pessimistic as heck and be like, "Ah, this is all gonna fall apart." And, like, I might say that in a casual conversation here and there if I'm fed up with something. But I think I'm at the point now where there's so many moving parts where I'm, um, in my education phase.

I'm [00:45:00] just keeping an eye on it. I'm listening to podcasts that are related to it. I'm listening to experts. 'Cause you'll have an expert that says, "Everything is 100% great," and then you'll have an expert the very next day come on the same show and say, "Everything is negative 100%. It's a mess. We'll, we will never survive."

And it's like, well, both of these people are experts. So I am personally just sort of educating myself, preparing myself, and just getting ready for whatever and just trying to prepare myself. I mean, will we be freed of all this labor as some people have said? Maybe. Will we be taken into, like, some terrible impoverished state?

Maybe. So I just, as things ad- Like, I, I, I advise, 'cause this is not investment advice, I am quite literally just like, "I'm gonna educate myself, and based on my education, make decisions." But I think that it's time, even if you hate AI, you should be listening to what's going on. 'Cause even if you're like, "I want it all to, all to shut down," [00:46:00] fair enough, and you probably have some good points, however, if it doesn't happen that way, you don't wanna be caught, you know, in a terrible situation financially or otherwise.

Just basically educate yourself 'cause it's, there's too many questions. There's too many what-ifs. There's too many when's. There's too many oh for sures. Then there's too many, "Oh, no, no, definitely not" for the same, the same opinion. "Oh, no, no, it definitely will happen." Actually, it can't happen. It's like, oh, well, here, here we go, two opposite opinions, and the people are 100% sure in either way, opposite directions.

So that, that's my advice and, and I'm, and I'm taking my own advice. I'm just, I'm just educating myself. Educating myself and taking action based upon my education more or less.

Mikhail: Sounds about right. Yeah, I th- I think that's good advice, honestly

Matt: And I think that's the web news. Uh, that's been A- the AI bubble. I'm gonna name it something a little more snazzy than that, but I hope you enjoyed it, [00:47:00] and that's it. We are signing off for the week

Mikhail: Goodbye