Is the AI bubble about to burst? Maybe we're asking the wrong question

Every company suddenly became "AI-powered"

A few years ago, every startup wanted to be the next Uber. Today the trend has shiftet to be AI-first company. From writing emails and generating images to analyzing data and helping developers write code, artificial intelligence has become the biggest technology trend in decades. Investors have poured hundreds of billions of dollars into AI, and companies are racing to integrate it into nearly every product imaginable.

But beneath the hype lies a question that's becoming increasingly difficult to ignore: If AI is changing the world, why are so many AI companies still losing money?

Why do people think AI is a bubble?

Calling something a "bubble" doesn't mean the technology is fake. It usually means that expectations, and investments, have grown faster than reality.

There are a few reasons why some people believe AI could be following that path:

• Billions of dollars are being invested before many companies have proven sustainable business models.

• Startups are adding "AI" to almost every product, hoping to attract customers and investors.

• Expectations are incredibly high, with some predicting AI will replace huge numbers of jobs within just a few years.

• The cost of developing and running advanced AI models remains enormous.

Howerver, none of these points prove AI is a bubble, but they certainly do look surprisingly familiar.

History doesn't repeat itself exactly.

But it often rhymes.

Is AI really just "autofill on steroids"?

One criticism you might come across is that AI is nothing more than autocomplete on steroids. At first glance, that sounds overly simplistic. But interestingly, it's not entirely wrong.

Large language models predict what word, or more accurately, what token, is most likely to come next based on everything that came before it. That's the same basic idea behind autocomplete. The difference is scale.

Today's AI models have been trained on enormous amounts of text and can write code, summarize research papers, analyze documents and hold conversations that feel remarkably human. So yes, AI is based on prediction.

But calling it "just autocomplete" is a bit like calling a Formula 1 car "just a faster vehicle compared to a bicycle." There's a shared principle, but the capability is in a completely different league.

The real challenge is the business model

Building cutting-edge AI systems costs an incredible amount of money. Training models requires enormous computing power. Running them for millions of users requires even more. Data centers, GPUs, electricity and ongoing research all add up to staggering costs. Meanwhile, users have become accustomed to AI being either free or available through relatively inexpensive subscriptions.

That's a difficult equation. The technology itself can be revolutionary while the economics remain uncertain. And history shows those are two very different things.

What if the biggest AI companies aren't the biggest winners?

Here's a thought that doesn't get discussed enough. Maybe the companies building AI models won't be the ones that benefit the most.

Think about how AI is being used today. Microsoft is integrating AI into Microsoft 365. Adobe is embedding AI into Photoshop. Google is adding AI across Search, Gmail and Workspace. Salesforce is enhancing customer relationship management with AI.

For most people, AI isn't becoming a product. It's becoming a feature. If that's the future, the companies that already have billions of users and successful products may end up capturing more value than the companies building the underlying models.

In other words, AI could become infrastructure.

Essential.

Powerful.

But not necessarily the most profitable place in the value chain.

What history can teach us

Technological revolutions rarely create one winner. They create an entirely new landscape. The companies making the biggest headlines today aren't always the ones leading the market a decade later.

So... is AI a bubble?

It might be,but maybe that's the wrong question.

The more interesting question is whether today's business models can justify today's AI investments. Artificial intelligence has already proven that it can make people more productive. It can help developers write code, marketers generate ideas, designers explore concepts and businesses automate repetitive work. That isn't going away.

But transformative technology doesn't automatically create profitable companies. History has taught us that lesson more than once. AI may very well become as important as the internet. That doesn't mean every AI company will become the next trillion-dollar business.

The future of AI isn't just about better models. It's about discovering who can actually turn extraordinary technology into a sustainable business.