AI use has doubled: who gets paid for the time firms save?

The number that doubled

55 per cent of Norwegian businesses had adopted artificial intelligence in 2025, up from 24 per cent in 2023 (Abelia). The figures come from report no. 1-2026, prepared by Samfunnsøkonomisk Analyse for NHO, Abelia, NHO Elektro, Norsk Industri and Fornybar Norge, based on two surveys among roughly 4,000 businesses (the report).

NHO sums up the finding as more than a doubling in two years, and the projections towards 2045 show that the same output can be delivered with considerably less time spent — labour is freed up for other value-creating activity (NHO).

That is good news for the Norwegian economy. It is not automatically good news for an accounting firm.

What the report measures — and what it leaves out

The survey answers two questions precisely: how many use AI, and what the benefit consists of. The answer to the second is time.

It does not answer a third question: who ends up with the value of that time. From a macro perspective, it does not need to. Freed-up labour is a gain for society regardless of who invoices it.

In an accounting firm billing by the hour, however, that is the entire question. There, time is not a resource being freed up — it is the product being sold.

The accounting sector's version of the problem

In an hourly model, revenue is the product of hours and hourly rate. AI reduces one of those factors. The rate stays put until someone actively changes it. All else equal, that means lower revenue on exactly the same delivery.

A simple calculation — ours, not the report's — illustrates the mechanics: a year-end close taking 20 hours at NOK 1,300 per hour invoices NOK 26,000. With 20 per cent less time spent, it becomes 16 hours and NOK 20,800. The client has received a NOK 5,200 discount without asking for it, without negotiation, and without knowing it happened.

The firm, meanwhile, has paid for licences, training, new routines and quality assurance of AI-assisted work. The return on that investment was paid out to the client.

The gap between the individual and the organisation

In September 2025, Statistics Norway measured that 54 per cent of the population aged 16 to 79 had used generative AI, while only three in ten Norwegian enterprises reported using AI technology (SSB). Individual adoption is running ahead of organisational adoption.

For an accounting firm, that gap is more than a statistic. It describes a state in which the efficiency gain is already happening — in the individual adviser's workflow, in drafts, reconciliation notes and client correspondence — without being described in any delivery or reflected in any price list.

A gain that exists only in individual heads and tools never becomes a line in the service model. It becomes a feeling that things are moving a little faster these days.

Depth of integration points the same way

The report emphasises that the benefits grow the broader and more integrated AI use is within the organisation (Samfunnsøkonomisk Analyse). It is presented as a productivity point, but it is just as much a commercial one.

A tool used sporadically by a single adviser produces a gain that is impossible to price. It varies by person, by day and by client, and can neither be described in a proposal nor guaranteed in an agreement.

A way of working that is standardised into the delivery — the same method for every client in the same segment — can be described, quality-assured, promised and priced. Standardisation is not merely an operational exercise. It is the precondition for efficiency being sellable at all.

Where the gain is actually booked

Freed-up capacity in an accounting firm has three practical destinations.

It can disappear into fewer billed hours on existing clients. In that case AI is a price cut the firm funds itself.

It can be filled with more clients on the same model. That produces top-line growth, and is a genuine answer to a sector struggling to recruit enough people — but the margin problem travels along into every new client agreement.

Or it can be given a price of its own: fixed prices on defined packages, ongoing reporting, liquidity and forecasting work, advisory tied to the client's own numbers. Only then is the link between time spent and revenue broken, and the price can follow the value the client experiences rather than the hours the firm consumed.

The window is the annual cycle

Accounting firms typically revise service descriptions and prices once a year. Adoption doubled in two years according to the figures from Samfunnsøkonomisk Analyse. Two years is two pricing rounds.

A firm that becomes more efficient during the year but leaves its service model untouched going into the next contract period has, in reality, made a decision: the gain goes to the client. The decision has simply not been written down anywhere.

The point

AI is not a profitability project in itself. It is a capacity project. Profitability is decided somewhere else entirely — in the service and pricing model that defines what the client buys and what it costs.

The report counts how many have adopted the technology. The income statement counts something else: who got paid for the time it saved.