Ask ten people in technology whether AI is a bubble and you get two camps shouting past each other. One points at valuations, at data centres being built faster than anyone can fill them, at start-ups raising fortunes on a demo. The other points at what the tools can now actually do, and at how quickly that changed.
Both camps are right. The mistake is thinking you have to pick one.
What a bubble actually is
A bubble is not a technology that fails. A bubble is money arriving faster than a technology can absorb it. Prices detach from what anything currently earns and attach instead to what everyone hopes it will earn. That is a statement about investors, not about the thing being invested in.
By that definition, yes, we are in one. Companies with no revenue are valued like utilities. The same handful of firms are buying chips, building data centres, and investing in each other in a loop that flatters everyone's numbers. Most of the products launched this year will not exist in three years. Some of the money spent on compute will never be earned back.
None of that tells you whether the technology matters.
The railway test
In the 1840s Britain went through railway mania. Hundreds of companies were floated. Lines were planned that made no sense. Ordinary people put their savings into schemes drawn on maps by men who had never surveyed the land. When it collapsed, fortunes vanished and Parliament held inquiries.
The railways were also the most important thing to happen to the British economy in a century. The tracks laid during the mania carried goods and people for the next hundred and fifty years. The bubble and the breakthrough were the same event, seen from different distances.
The same pattern repeats. The dot-com crash of 2000 wiped out companies whose names nobody remembers, and it also left behind the fibre, the habits, and the survivors that became the internet everyone uses now. The technology was real. The timing and the prices were wrong.
Why this time follows the pattern
The thing that separates a real breakthrough from a pure fad is whether people keep using it after the hype stops paying for it. On that test, AI is already through.
I use these tools every working day. Not for novelty, but because a piece of work that took an afternoon now takes an hour, and because I can now attempt things I would previously have turned down. Small businesses I work with are using them to write the emails they never got round to, to make sense of their own sales figures, and to learn skills they used to pay for. None of that depends on a share price.
That is what the breakthrough looks like from ground level. It is quiet, it is practical, and it does not care about valuations.
Holding both ideas at once
So how should a business owner think about it? A few rules I try to follow.
Do not confuse the products with the capability. The company selling you an AI tool today may be gone next year. The ability to have a computer read, write, and reason about ordinary language is not going anywhere. Buy the capability where it is cheap and avoid depending on any one vendor more than you have to.
Ignore the promises, look at the tasks. Anyone who tells you AI will replace your staff or double your revenue is selling something. Anyone who shows you a specific task it does well is telling you something useful. Judge it one task at a time.
Expect a correction and plan to be here after it. When the money tightens, prices for these tools will change, some services will close, and the headlines will say AI was overhyped. The businesses that quietly built the tools into their daily work will not notice much. The ones that bet everything on a single supplier will.
Learn it yourself. The lasting advantage of every previous breakthrough went to the people who understood it, not the people who invested in it. The skill of using these tools well is cheap to acquire right now and it will still be valuable when the bubble has burst.
The honest answer
Is AI a bubble? Yes. Is it the most important technical change since the internet? Also yes. The people who lose money will mostly be the ones who bought at the top, and the people who benefit will mostly be the ones who learned to use it and kept going.
Those can be different people. Aim to be in the second group.
If you would like to talk through where these tools could realistically help in your own business, without the hype, get in touch. I would rather show you three things that work than promise you thirty that might.
