If AI Cuts Every Cost, Who’s Left to Buy?

I read a paper last night called The AI Layoff Trap.

I know. Cheery little bedtime number 😀

But it put words around something Helen said to me on a walk with Hugo last week, and I’ve been mulling it over ever since.

We were talking about automation, Collab365, and the slightly awkward place businesses are now in.

Helen said:

“Companies racing to automate so they can slash costs are just firing their own customers.”

That’s the trap.

And I think we’re all walking into it.

We use AI every day. In fact, we’re rebuilding Collab365 around it. Without AI, I’m not sure we’d still be standing.

But the paper makes a really uncomfortable point:

A company can make a perfectly rational decision for itself and still help break the market it depends on.

Put it this way.

Company A uses AI to cut costs.

Maybe it doesn’t hire the ads manager, the designer, the copywriter, the junior support person, the admin assistant, or the poor soul who used to spend two days cleaning spreadsheets and chasing invoices.

Profit margins improve. The business survives longer. Yay. The founder breathes out for the first time in months.

Totally rational.

Then Company B does the same because it has to compete.

Then Company C follows or it goes bust.

Then the whole sector joins in, because nobody wants to be the noble idiot carrying a higher cost base while everyone else is using agents to undercut them.

Each decision makes sense.

Every single one.

But those “staff costs” and “agency costs” on the spreadsheet are also customers.

The ads manager who didn’t get hired has less money to spend. The designer who lost three retainers is not buying a “how to build your personal brand” course. The copywriter watching AI eat the bottom of the market is not upgrading to another membership.

The junior marketer who never gets the first rung on the ladder is not paying for software, training, events, templates, coaching, conferences or all the other stuff businesses like ours sell.

They cut back too.

Nobody’s buying the stuff.

That’s the bit I can’t shake.

One business automating is survival.

Every business automating at once starts to look like demand destruction.

And this is where The AI Layoff Trap gets properly uncomfortable.

The authors argue that displaced workers are not just workers. They are consumers/buyers. If AI removes their income faster than they can reskill into new work, the spending power firms depend on starts to erode.

At the limit, companies automate their way to incredible productivity and shrinking demand.

Or, in simple terms:

Everything gets cheaper.

Nobody can buy the stuff.

It sounds dramatic, but the mechanism is painfully simple.

A firm captures the full benefit of cutting its own costs, but only feels a fraction of the demand it destroys when workers lose income. The rest of that damage gets spread across everyone else.

So your spreadsheet says:

“Great news, we saved £80k.”

The economy quietly says:

“Consumer spending is down again.”

It’s a prisoner’s dilemma dressed up as progress.

And the annoying bit is that knowing the trap exists doesn’t magically solve it.

Try walking into a board meeting and saying:

“We could automate this department and protect margin, but I don’t think we should because it might damage demand later.”

Good luck.

You’d be replaced by someone with a nicer deck and fewer morals.

That’s why this scares me.

Not because firms are evil.

Because they’re rational.

And when rational firms are forced into the same corner, they can collectively produce a stupid outcome.

We can see a tiny version of this inside Collab365.

Take marketing.

A few years ago, launching a proper campaign meant pulling in people. An ads person to structure it. A copywriter to write the angles. A designer to create the assets. Someone to cut video. Someone to build the landing page. Someone to look at the results and tell us which bit was on fire.

Now a frightening amount of that can be done by me, Helen and a stack of AI tools.

Claude helps with angles, copy, variants and analysis. Canva handles more of the creative production. Agents can help read performance data, suggest tests and generate new campaign ideas. Builders can spin up landing pages in a fraction of the time.

That is brilliant for us.

It means we can move faster, spend less and survive longer.

But it also means the ads person doesn’t get the brief. The designer doesn’t get the work. The copywriter doesn’t get the project. The video editor doesn’t get the clips.

Then we wonder why the market for courses, memberships, tools and services starts feeling thinner.

We are not outside the problem looking in.

We are inside it, doing the rational thing, and that is what makes it awkward.

The paper also makes another point I think people will miss.

This is not just bad for workers.

It can be bad for company owners too.

If enough companies over-automate, the lost spending power can come back around and hurt profits. Not because capitalism suddenly grew a conscience, but because revenue needs buyers.

Companies don’t sell to "efficiency gains".

They sell to people and organisations with money.

That is why the usual answers feel thin.

“Just retrain people!”

Yes, we should. Properly. But retraining takes time, and the layoffs can happen in a quarter. The awkward question is whether AI creates enough new jobs quickly enough to replace the income it removes.

I’m not sure.

“Universal Basic Income will fix it!”

Maybe UBI helps people survive, and I’m not dismissing that. But the paper’s point is that general income support does not necessarily change the company’s incentive to automate in the first place.

It raises the floor.

It doesn’t stop the race.

And .. where are governments going to get the money to pay UBI? Print it again?!?!

The paper’s unfashionable answer is a Pigouvian automation tax.

And no, sadly, nothing to do with pigs.

In plain English: if replacing workers with AI imposes a demand cost on the wider economy, tax the automation at the point it happens and recycle that money back into the system....

Wage insurance. Serious retraining. New job creation. Maybe shorter working weeks. Maybe support for entry-level roles so young people still get somewhere to start.

I know “tax the robots” sounds like something from a 2017 panel discussion where “future of work” was the latest buzz.

But in the short term (next 10 years), I’m struggling to see how we get through the transition without some mechanism that keeps people buying while companies automate.

Because if we don’t recycle the gains somehow, the logic gets ugly.

Firms cut labour to protect margin. Workers lose income. Demand falls. Firms see weaker demand and cut more. Competitors automate harder. Prices drop. Margins compress. More people are removed from the system.

Round and round it goes until everyone is very efficient and strangely broke.

At Collab365, we’re making the same bet lots of small businesses are making.

Stay lean. Use AI properly. Keep humans in the loop. Build something useful before the ground moves again.

But I can’t pretend it doesn't feel awkward to me and Helen.

AI is helping us survive.

The same pattern, multiplied across the economy, could damage the demand we all need to survive.

That is the bit I cannot shake.

The transition question is not just:

“How do we help companies adopt AI?”

It is:

“How do we keep enough people earning, spending and moving forward while they do?”

Because if every business treats people only as costs to be removed, eventually the spreadsheet runs out of customers.

So here’s the question I keep chewing on:

If your competitor halves their headcount next year and undercuts you by 40%, what do you do?

And if everyone does the same thing, who is left with enough money to buy what any of you sells?