Everyone wants AI deals. The money is in boring adoption.

Share
Everyone wants AI deals. The money is in boring adoption.


My 2027 PE prediction is fairly simple.

A lot of people will overpay for AI companies.

A smaller group will make far more money by forcing boring companies they already own to use AI properly.

The first path is exciting.

AI-native targets. Huge multiples. Breathless growth. Technical diligence nobody fully understands. Foundation model dependency. Product risk. Enterprise adoption uncertainty. A management team using words like "agents" and "proprietary layer" while the deal team quietly hopes the CTO can explain it later.

Some of those deals will be brilliant.

Many will be expensive lessons with better branding.

The second path is less glamorous.

Healthcare services using AI to improve scheduling, intake, prior authorisation workflows, call handling, clinical documentation, and revenue cycle leakage.

Residential services using AI for lead scoring, dispatch, route optimisation, quote follow-up, demand forecasting, review response, technician utilisation, and churn prediction.

B2B services using AI to reduce manual account management, improve pricing discipline, automate reporting, prioritise sales activity, and remove the cheerful spreadsheet chaos sitting underneath most "professionalised" businesses.

That is where the money sits.

Inside existing EBITDA.

Inside broken workflows.

Inside labour-heavy processes.

Inside pricing decisions made by habit.

Inside sales teams calling the wrong leads first.

Inside finance teams taking fifteen days to close because half the month-end process is ritualised suffering.

AI does not need to make a portco look like a tech company.

It needs to make the business faster, cheaper, sharper, and less dependent on the three people who know how everything works because nobody ever built the system properly.

I would rather own a dull services business that uses AI to lift margin by 300bps than chase a fashionable AI asset at a heroic multiple where the moat may disappear in the next OpenAI release.

The money is rarely in the conference demo.

It is in the boring process nobody fixed because manual work was cheaper than thinking.

That excuse is dying.