Dude, you know how everyone keeps saying the same thing, claiming the AI bubble is going to burst because of a GPU shortage. They say factories won't be able to keep up with production, that we’re short on materials, components, power, and that we’ll run out of water to cool all these data centers. But man, I’ve been thinking to myself, and I honestly think everyone is looking in the wrong place, you know?
Because dude, the AI bubble is going to burst, and to me, it's about to happen, but it’s not for lack of chips; it’s going to burst because of a lack of debt, man. 730 billion is what Big Tech is going to spend on data centers this year. They've probably spent most of it already, since the year is almost over, man.
But if you stop to think, that's 78%more than last year, compared to 2025. Only, that money isn’t coming out of the cash reserves of these AI companies, bro. It’s coming from credit, from bonds, from private credit , from structured financing, and so on.
To give you an idea of how serious this is, the Bank of England issued a warning in July, and basically, Capital Economics projected the S&P 500 falling 21%. Meanwhile, over 170,000 people have already been laid off from tech jobs in just the first 9 months of 2026 . And that’s when I stopped to ask myself, man, just to try and make sense of this.
I asked myself, I said: "Dude, when is this thing going to blow up? " >> Because, man, it is definitely going to blow up. And another thing that creates a totally absurd doubt is: what’s going to happen, for instance, to those free AIs?
What’s going to happen, for example, to those subscriptions that aren't exactly cheap, that we're paying for, like, you know, to use Cloud Code? How are we, Vibe Coders, going to survive from here on out, man? What is going to happen to us?
Could it be that vibe coding is going to turn into something more enterprise-level? Because listen, like I said, this bubble is going to burst, and it feels like it’s very close. And when it bursts, the first things to disappear will be, man, the cheap stuff we use every day to test things out, dude.
So, grab your coffee so we can chat more about this. Go ahead and subscribe to the channel, leave a like, and turn on notifications so you don't miss our lives, our videos, nothing, man. This right here is the biggest tech dive bar .
We talk about technology here in a raw way, with no course sales, no mentorship sales. Let me show you here, my friend, what I found, because, man, opinion is just paper and on paper you can write whatever you want, you know? July 2026, the Bank of England published the Financial Stability Report, which is basically a report, it's a more technical report, you know?
that gets published. And the message this report leaves is interesting. It says that artificial intelligence companies are using credit, man, in a crazy way, you know?
Super accelerated. These guys are using public bonds, they're using private credit, they're using leveraged financing, all at the same time. And if it were in Brazil, man, they’d even be taking loan shark money to inject into this crap, bro.
Money from loan sharks, you know? And there's more, man. Just in the first half of 2026, man, big infrastructure companies had already issued more bonds than in all of 2025.
Man, do you have any idea, dude, that it's the whole of 2025? Because there's a crowd that gets the impression and says: "Man, I think we're reaching a more mature phase of AI now. " AIs aren't as accessible as they were before.
The price is already higher and it's becoming something okay . I think it's paying for itself, but it's not, man. It's not.
It was the whole of 2025 and it’s outdoing itself every 6 months, man. Then you’ll say: "Dude, but debt for a tech company is normal, man. " Uber until the other day, I don't know if even today, hell, was growing and wasn't generating any profit, man.
But, man, this thing is much more serious than we imagine, dude. Just for you to have an idea of when was the last time this happened, it was the Telecom bubble back in 1999 and the Dot-com bubble in the year 2000 and also the housing bubble in 2007. To give you an idea, man, when people start taking private credit, that's usually when the bank closes the door, you know?
It says : "Man, we're not going to release any more money to you, it's over. " And what is the consequence of this? The interest on this thing is incredibly high.
It doesn't have much regulation, and the deadline for this thing, man, any day, any hour, it could expire, bro . And when it expires, my friend, I think the bill is going to come to us, vibe coders, you rascals, pro players of cloud code. Now, before I go on about what's going to happen, man, let me tell you something else, you who are a vibe coder, just like me.
You dev, you chrome-user who is building a billing product, a SaaS product, something where you need to charge someone, man. You have to understand that even though we joke that we are vibe coders, we have a very big technical responsibility, man. And we know that payment infra is not a business decision, no.
It's not a decision for the folks who just hammer the gavel and pay the bills. No, man. We are talking about architectural decisions.
If you mess that up, man, the payment system goes down and you're the one who gets screwed. You're the dev who will be there late at night, man, unable to sleep. Max, our partner, solves this beautifully.
They have a wonderful, documented API, man. They have anti-fraud. And their anti-fraud is proprietary and uses machine learning.
It’s not like those other anti-fraud tools that go around declining legitimate purchases. No, man . Besides that, they have native split and the best part, they have zero-fee Pix for you, dev, you who frequent this channel, man.
I left a little link in the description. Go there, integrate it , and come back later to thank me. At the end of July, the results came out.
For Q2, which is the second quarter, man. Amazon had 220 billion in Capex, Alphabet 200 billion, Microsoft 175 billion, and Meta 140 billion. Now, my player, I’ll add it up for you, no need to grab your calculator.
Let's add it up, man. That was 730 billion dollars burned, my player. 78%more than in 2025, man.
Then you’ll ask me: " Hey, buddy, but where is all this money coming from? " >> A portion of the revenue, because these companies are actually bringing in revenue, they aren't just burning it, they’re making money from this thing, because we're talking about big tech, not some random startup or a group taking venture capital money to just mess around, no way. We’re talking about a more mature crowd.
But look, man, Alphabet spent 44. 9 billion in Capex in Q2, and they generated 39. 1 billion in operating cash.
The free cash flow, buddy, it was negative. And man, just so you grasp the gravity of this, it's the first time this has happened since Google went public back in 2004, when I know a lot of people here were still in diapers. We’re talking 22 years ago, man.
The first time this has happened. And we know the difference, right? It’s coming from somewhere else and not from their cash reserves, right, man?
Then in September , Capital Economics analyzed eight bubble indicators. Five of those indicators were already, man, at maximum alert at an extreme level. The S&P 500 projection dropped 21%by the end of 2027, man.
21%, buddy, that’s not a correction, man. It’s a crash. It’s a disaster.
Did that cause a problem? But who am I going to leave that topic to? My friend Raul Sena, the "Sardine" investor.
Now we’re going to cover the subject that relates to us programmers, us coder vibes, man. Because man, there’s something nobody is asking that really worries us, which I don’t see the business crowd, or these AI, mentorship, or course-selling channels talking about. The folks aren’t questioning this at all, man.
When this bubble bursts, what’s going to happen to us, man? Like I said, I’m not an economist, I’m not a futurologist, I’m a programmer just like you. YouTube is just a hobby for me, you know?
Because I’m a dev, I’m on the front lines taking the hits, losing sleep, fixing bugs. And that’s why what worries me as a programmer isn't the S&P 500, man. It’s what’s going to vanish along with it for us.
Because think with me, the coding agent we use nowadays that we’re paying $ 200 for, man, what’s going to happen to that? Do you think this is going to turn into an enterprise thing? And those AI agents we’ve been using, damn, I don't know, inside OpenRouter, we're trying out some alternatives, checking out some Chinese stuff; do you think China will really grow in this?
Do you think Chinese models will make a leap because China has more money to invest? I don't know, man. Because using a code agent is one thing, and having, for example, an LLM model deployed in your application is another.
And when we say you have a model plugged into your application, you have a cost. That cost goes directly into those Illuminati reports for the business guys to see if the company is actually turning a profit or not. If that cost skyrockets in the coming years, it could be that for many companies that have AI plugged into their tools—where it's actually useful, because there are plenty of tools where the integration is, man, completely useless.
It’s just a little chat that serves no purpose at all. Like Canva, for example. Man, Canva’s chat is horrible, seriously.
It generates some pretty messed up stuff; it helps out in a pinch, but it’s not something you use that much. But there are some things where AI was implemented, man, that turned out wonderful and actually became a core part of the company. If that drives up costs, we’re talking about a lot of companies going bust.
That’s going to splash back on us. It’ll splash back, maybe, in more layoffs; it’ll splash back, maybe, with companies needing to be flexible to restructure; it will splash back, for instance, on the dev who uses AI like crazy, mindlessly, having to start looking at what they can get out of an AI, plugging in a machine learning model, plugging in a Chinese model, you know? So, man, this is going to have an impact on our day-to-day lives.
But there’s another side to this: we know these big techs, man, they’re just burning cash to get us addicted to their ecosystem. It’s basically like gambling, it’s the new gambling. And their plan is exactly that: get you hooked first, charge you later.
Go back, look at when that course launched and you paid, like, a $ 10 subscription, see if you’re paying that today. And every company does this . Uber did exactly that, and then came the "enshittification.
" Now Uber is super expensive, the cars are terrible, and the driver acts like they're doing you a favor just by giving you a ride. iFood did the same thing; it used to be really cool. Now, man, it’s just completely absurd.
But Uber and iFood had venture capital, my friend, look. Just throwing money at it. Big Tech companies don't have venture capital throwing money at them.
They are funding their own party with the profit they have—the little profit they have —and with massive amounts of leveraged debt, man. And when that debt comes due and it all bursts, man, we're screwed. Then you'll say: "Damn, man, but that’s an exaggeration, dude.
" There will always be a free tier. There will always be a free tier. So alright then.
So for you who are new, you who are young and emotional, you're probably a React developer who became a dev during the pandemic and thinks I'm talking nonsense, that I'm just an old guy criticizing, thinking everything is going to end. Back in the year 2000, there were dozens of startups giving away free storage. There was this stuff that gave you free email, free hosting, it gave you—man, where are those companies now?
When the dot-com bubble burst, man, what was left were exactly the companies that had real revenue. The rest, my friend, those companies all vanished. "Oh, but AWS gives a free tier.
" Sure, but what does AWS give you a free tier for? Same scheme: to get you hooked on their environment and then charge you, man. And their bill arrives hefty.
It's not just AWS, no, GCP too. All the cloud companies do this as well, man. Now just think about your own workflow, man.
What in your workflow today is subsidized? I'm talking about the free tier of an assistant, I'm talking about that cheap plan for your code agent, I'm talking about that cheap API from OpenRouter. All of this, buddy, if we stop to think about it, it's a somewhat artificial price, you know?
You can't rely on it, because all of this depends on someone burning a ton of money on the other side, man. And that money will run out, and it's getting close. That's why we're talking about this so much in this video, I keep harping on it because, man, this has been consuming my mind, man.
But on the other hand, I don't think AI is going to disappear. I think a maturity phase is coming, and it’s already happening. If we stop to think, man, where did all those AI video generator apps go, man, the ones that gave us free stuff?
They vanished, man. They vanished. Well, there’s one or two out there, maybe.
Now there’s Cling, Luma, stuff like that coming out . Even Runway officially put it in their tool, but it’s a paid thing; those apps to generate puppy memes for free are over. So, could it be that these code agents, these, man, using cloud code, will become an enterprise product, a big company product, you know?
Like an IBM. You don’t see an end user having something from IBM that they personally signed up for, do you? No.
Usually, who uses IBM, IBM closes a deal with the company, and the company provides it to the employees. So, is that the future of artificial intelligence? Will it be an enterprise thing?
Will it be something you won't be able to pay for out of your own pocket, man, to afford? Then I ask: " What about the independent dev, that Joe SaaS, that indie developer, that freelancer, that guy who's just starting to program with AI, man, what’s going to happen to these guys? " I don't know, but I think these guys, man, will be the most affected, especially those entering the market, because they won't have a way to train, unless there are those student packages like there are, and so on; maybe that’s a way to hook those people.
But, man, I think there’s going to be something weird in the coming years, man, because this subsidy will end, the free plan will become paid, the paid plan will get super expensive, and that talk of "anyone can create their SaaS," the barrier of code is gone, man. Man, that’s history. And there’s another side to it, too.
The people addicted to artificial intelligence, who don't know how to program, who never had an interest in learning to program, whose life is just tossing prompts, and they have so much confidence. Actually, I even made a video called "AI Midwit," where we have a person who is very dim, but they think they’re very confident and think they’re amazing because they like to complicate things. And there's another side to this, man, because the same big tech companies that are, like, blowing 730 billion dollars on data centers, on the other hand, are laying people off, using that line that AI could do people's jobs.
But we know the real reason for these layoffs is to free up cash to invest in more AI, because there's still no ROI. And when that debt comes due, what’s going to happen? Will they rehire people or lay off even more?
Man, I don't know. The future is going to be uncertain, bro. So, it's not that AI is going to disappear, it's just that, man , in my mind, I think it’s going to get much more expensive.
And when it gets much more expensive, it's the kind of thing not everyone will be able to use; it’ll just be the heavy hitters, the devs. So, if you’re a programmer like me, man, and you’re building something out there that needs billing, talk to my friends at PMAX, dude. They’ve got payment infra, architectural decisions.
Don't put this off until later. Don't wait for the bubble to burst, man. The time to integrate is now, my friend.
And comment down below what your view is, what you think. Let’s talk, man. Am I being desperate for no reason?
Is it possible, man, that I have another theory that the models, the generalist models, they're going to die and we'll have more specialized and much cheaper models? Will that actually work out? Could Brazilian AI be our salvation?
I don't know. Leave your comment here. Let’s discuss it, let’s trade some ideas.
Not subscribed to the channel yet? Subscribe to the channel, hit the like button, and turn on notifications so you don't miss any videos. Help me reach 400,000.