Is the AI Boom a Bubble Right Now?

That’s a question that’s been on my mind, and I don’t think it has an easy “yes” or “no” answer. To even begin, we need to be on the same page about what an economic bubble is.

It’s an economics term describing a rapid, unsustainable rise in asset prices, driven by speculation rather than a product’s actual, fundamental value. During a bubble, inflated prices are fuelled by hype and the all-consuming Fear of Missing Out (FOMO). The follow-up is usually a sharp, sudden crash as prices collapse, causing significant losses for investors.

And yes, sometimes during a bubble, you see questionable actions: grifters, fraud, and people trying to cash in and cash out before anyone notices the house of cards collapsing. Are we in a bubble right now? For me looks like we could be lets look at why. Not every bubble is created equal, some spill over to wider economy and cause huge damage. 

The Magnificent Seven: Speculation, Debt, and Toxic Interlinkages

To begin the story need to talk about Magnificent Seven. Not a superhero group but seven tech companies. Apple, Microsoft, Amazon, Alphabet (Google), Meta (Facebook), Tesla, and Nvidia. Every ever-increasing stock value and dividends give you good returns. They are the source of most stock market growth and economic investment. Economic engine that keeps USA running. Without them USA would be in recession or low growth zone. 

They’ve all seen a rapid surge in asset prices, heavily driven by speculation. Keep up the double didet growth over past decade or more and keep growing. Why there has been such a focus on cutting costs within each company. Recently claims Artificial General Intelligence (AGI) will yield growth and economic beneifts so big these companies will become even bigger giants. Becoming the sole competition and kings at the top. Within that you have people saying Artificial Superintelligence (ASI) is coming. Some even claim they already have the latter, but I’m still highly skeptical. Also quite skeptical about the claims on offer with the benefits being cost free. When companies without a viable product are easily securing massive funding, it’s safe to say high asset prices and speculation are absolutely true. The hype and FOMO are real. Why everybody is jumping on AI bagon wagon and nobody knows what quite to do with it. 

Seven giants of sillacy valley. 

A closer The business models vary greatly, which makes the uniform hype questionable. Apple sells products and cloud services, Microsoft sells cloud services and software, Amazon sells cloud services and a bit of everything, Alphabet sells advertising and cloud services, Meta is an advertising company, and Nvidia sells the hardware all the cloud services use.

Then there’s Tesla, which sells cars and claims to be working on super-awesome robots, self-driving taxis, and god knows what else, but hasn’t fully built or deployed any of it yet. This divergence makes Tesla the most questionable company riding purely on speculation.

Worse, some of these companies have a circular relationship. Take Nvidia, which holds a stake in OpenAI. This relationship effectively allows OpenAI to buy the GPUs from Nvidia that it couldn’t otherwise afford. Microsoft is also an investor/partner in OpenAI. Oracle buys GPUs from Nvidia, then sells that computing space to companies like Microsoft. Dozens of examples like this. One company is holding up everybody else up on a promise. That’s a bit of a red flag.

The Debt and The Slop

If you look at the fundamentals, the current valuations simply don’t match the reality of existing revenues or profits at all. These companies are spending hundreds of billions without guaranteeing a solid return yet. Based on current bets, they need an extra $2 trillion in extra revenue. That would mean doubling current $2 trillion they currently make together today withini a couple of years. That sort of exponential growth is crazy and would mean double didget growth carrying on.

While some of that is necessary infrastructural investment, that would be a stronger argument if GPUs didn’t get outdated within, say, 2-3 years. It’s like building more energy generation capacity but stream powered. Another example I’ve seen used is the railways, which most of them never got used. Another example is the fibre optic cables from the Dot-com boom—again, took years before they were used, and the companies who built them went under. The same happened for railway companies.

A lot of that massive power is being used to generate a bunch of text and images that offer questionable value, dubbed “AI slop.” It is deeply inefficient, really, with no proven business case. It takes huge amounts of water and electricity to run, too. They are claiming to investors that any increase in revenue is due to AI. Energy companies, data center operators, and more are making huge bets based on this “house of cards.” Now imagine how many third-party suppliers are impacted by this.

What is truly worrying is the financing. Worse still, most of this borrowing is private credit—costly loans with limited public reporting. My point is that this could be a toxic set of interlinked loans that could rival the subprime crisis, and no one knows the scale of the problem. We could see plenty of unconnected companies connected to it, all built on a dream and a sugar rush.

You could argue that this looks similar to Dutch disease, which is the economic phenomenon where a boom in a specific sector, like natural resource extraction, causes the decline of other sectors, such as manufacturing and agriculture. Here, the AI investment is crowding out other segments of economic activity on the hope it will pay off. There is some evidence that this is becoming the main driving force behind U.S.A. growth and investment. Things get complicated here due to the U.S. government trying to devalue the dollar with its trade policy, but that’s a whole another kettle of fish.

The flood of AI slop is making it hard to guess what human content is and what is just garbage, leading to the mass spreading of misinformation and disinformation. The irony is that when you look at surveys, most people have doubts about the productivity claims and can’t pinpoint the benefits of the technology, which are very limited compared to the hype. In many ways, this reminds me of the early internet. Nobody knew how to use it yet, but there was so much hype. Which means two things could be true at once: a Dutch disease bubble is growing right before us. The problem is not until you’re on the top of the bubble can you spot the bubble.

The Core Paradox: Investment vs. Profit

This leads us to the heart of the “Is it a Bubble?” debate.

On one side, the “No Bubble” crowd points to that same record investment in future infrastructure as proof that no collapse is imminent. They argue the payoff is guaranteed: revenues will grow, and profits will follow. They say it can’t be like the Tulip Mania or the Dot-Com bubble because this time, the scale of physical investment is unprecedented. Some are saying not matching the same valuations as the Dot com means no bubble.

On the other side—the “Yes, it’s a bubble” side—is the plain financial reality: valuations don’t match revenues or profits at the moment. Most investment is based on a shaky proof of concept, and worse still, some of this technology may not even have a durable, long-term commercial purpose. The truly useful machine learning applications don’t always get the most hype. On the other hand, valuations are higher than they should be, so it is a bubble. It may not be the biggest bubble, but still one.

The Labor Paradox: Undermining the Knowledge Economy

I remain unconvinced. I see AI mostly as a productivity tool, but even that is questionable. In a knowledge-based and service-oriented economy, it seems companies are trying to cut costs and privatize the massive profits, while everybody else—workers, creators, and even the broader economy—picks up the losses. What they are trying to do is destroy the very thing that allows it to exist. That makes me think it will not work out.

Part of me genuinely believes the main driver behind most of this value is pure fear of missing out. People are rushing toward it without critically questioning the “why.” Everybody else who isn’t investing looks like a fool… until they don’t. Hopefully, things turn out okay, but I’m not so sure.

Conclusion: My Lukewarm “Yes,” and the Road Ahead

For me, I would say, with a lukewarm yes, it is a bubble. I take this view due to how I understand LLMs. Large Language Models (LLMs) have a scalability problem, and far from trying to solve it, they keep adding a bigger engine. It’s sort of like adding a bigger engine to a racing car without changing the rest of the parts; yes, you get results, but diminishing returns start to set in.

Also on my mind are the weak underlying reasons for its existence. What I mean is that in its current form, the technology seems like a dead end. Smaller, less general models seem to have more promise. Take, for example, Gemini; it never stopped the logic flaws in sentences or the mistakes when I gave it this text. It sounds human but lacks the creativity to make things shorter or add certain quirks. Also, there are the security flaws. Getting bigger, and constantly going for more data, is merely hiding the inefficiency of the models.

As the bubble gets bigger, it could start pricing out investors and limit access to money. I’m not sure how or when it will pop. If it does pop, it could be a massive bang and shake the global economy. Or it could hit a bunch of private investors and loans only. We just don’t know. The chat is: if we build it, they will come, and what we’re building is worthwhile. Some people have been saying we’re building something useful behind it. There are going to be winners and losers here. The problem is you can’t pick the winners here. It’s hard to guess who’s going to be worth investing in.

More I listen to people talking about AI, more I understand people don’t understand it and are following the herd. Now it could be a transformative bit of technology, or it could be nothing. I suspect it won’t do what people claim it will do. But they have made one of the finest spell checkers and grammar checkers, it just takes a ton of water and electricity to make it work. If it sounds too good to be true, it normally is. Tech companies are expecting double didget growth and looks like more exponential growth. Yet so many people forget that.

What truly worries me is the fervent, often uncritical way some people talk about it. Perhaps I would be a smart investor only going for proven companies—the Warren Buffett method (his fund is currently holding billions in cash, arguing the market is overvalued). Good news: AI chatbots-generated text is pretty good at grammar and spell checking, but sometimes mistakes are what make stuff sound human. Whatever it is, we’re heading towards uncertain times. I just wish that if it does go pop, the fools that led us here and caused the mess face jail time or prison if they did anything wrong. Yes, the scars of the subprime crash still hurt.