AI Is Not a Bubble. But Parts of It Definitely Are.

Over the last year, one question keeps coming up: “Is AI just another bubble?”

The honest answer is uncomfortable for people who like clean narratives. AI is not just hype, and it is not a guaranteed overnight takeover either. It is real technology delivering real value, while parts of the market around it are clearly drifting into overpromise and overinvestment.

From my perspective as someone who has spent years building products, working with automation, and solving technical problems inside businesses, the conversation around AI has become far too black and white. Some people still act like AI is just hype. Others speak about it like it is guaranteed to replace entire industries overnight.

Reality usually sits somewhere in the middle.

The Technology Is Clearly Real

A couple of years ago, it was fair to question whether AI was genuinely useful. For most people, it meant:

Interesting, sure. But not mission-critical.

Today, that has changed. AI is being used to complete actual work. Not in theory. In production. Every day.

The jump from “interesting demo” to “useful tool” happened surprisingly quickly. If you use these systems regularly, you can feel it: the tools are not perfect, but they are useful. And usefulness is what matters.

Real Technology Can Still Become Overhyped

The internet changed the world. That did not stop the dot-com bubble.

Railways transformed economies. That still led to overinvestment.

This is the point many people miss: a technology can be genuinely valuable while the surrounding expectations become unrealistic. That is where things get messy.

Because when expectations inflate faster than real-world delivery, you get:

So yes, some of this is a bubble. But that does not mean the underlying shift is not real.

Many People Are Still Underestimating AI

Here is the other side: while there is clearly hype in parts of the market, I also think many people are underestimating how significant this shift could become.

A lot of businesses treat AI like another software trend. I do not think it is.

Most software historically helped people work faster. AI is different because it can increasingly perform parts of the work itself. That changes the economics of entire industries.

The important question is not “Can AI replace a role tomorrow?” It is: what happens when one person can do work that previously required two people, five people, or entire departments in some cases?

You do not need full automation to reshape a market. Even partial automation changes:

And I already see this happening inside technical workflows. Tasks that used to take hours can sometimes drop to minutes:

The compounding effect across an entire business is enormous. And the technology is still improving at an aggressive pace.

The Benchmark Is Not Human Perfection

A lot of people compare AI to the best possible human outcome. Businesses do not.

Businesses compare it to cost, speed, and efficiency. Those are very different benchmarks.

AI does not need to be perfect to disrupt industries. It just needs to be economically useful.

In many areas, it already is.

What Is Actually Driving AI Forward?

People focus on the models, but underneath all of this is an infrastructure race.

AI systems are becoming faster and cheaper partly because of improvements in:

Most people focus on NVIDIA because it is the visible name. But companies involved in memory and manufacturing infrastructure matter just as much to where AI goes next.

At the same time, businesses are finally finding use cases that make financial sense. That matters more than social media hype.

Companies pay for tools that:

That is why AI revenue is growing so aggressively. Not because it is trendy. Because it is useful.

Where the Real Risk Is

The risk is not whether AI works. The risk is how aggressively the industry is building around future expectations.

Every week brings announcements about:

But announcing something and delivering it are completely different things.

A data centre is not valuable because it appears in a press release. It still needs:

And many of these are becoming bottlenecks.

AI Is Becoming an Energy Story

One of the biggest shifts happening quietly in the background is energy demand.

Modern AI systems consume huge amounts of electricity. The challenge is no longer just building better models. It is whether the infrastructure around them can keep up.

Power availability, cooling systems, transformers, and grid capacity are becoming critical constraints. Software moves quickly. Physical infrastructure does not. That gap matters more than most headlines suggest.

What Businesses Should Actually Focus On

A lot of companies are approaching AI badly. Some ignore it completely. Others force AI into every workflow because they feel pressured to.

The better approach is usually simpler: look for repetitive mental work. That is where AI performs best today.

The biggest gains are often not flashy. They are operational. Saving a team a few hours a week does not sound exciting until it compounds across months, across departments, across an entire business.

Where This Goes Next

I think AI will reshape how businesses operate over the next decade.

I also think there will be plenty of failures along the way. Not because the technology is fake, but because expectations are running ahead of what companies and infrastructure can realistically support.

The businesses that win will not be the ones chasing headlines. They will be the ones using AI carefully and practically, focused on ROI and workflow leverage.

Right now, the biggest advantage is not replacing humans. It is helping good people move faster. If you want help finding those practical wins inside your business, you can find out more on the coaching page.