As the AI gap widens, the firm becomes the client's AI department

The common reading of the new AI numbers is that small businesses are falling behind and need to catch up. That reading is both correct and useless. Correct, because the numbers show exactly that. Useless, because it assumes a fourteen-person company should solve a problem that's structural, not a matter of attitude. A more precise reading is this: the gap SSB is measuring is a price list nobody has written out yet.

The gap is widening, and it's widening faster at the top

In 2026, 8 in 10 large firms use AI technology, compared with 4 in 10 small firms, according to Statistics Norway. The smallest firms, with 10-19 employees, have moved from roughly 3 in 10 in 2025 to around 4 in 10 this year. Firms with over 100 employees went from 6 in 10 to 8 in 10 over the same period.

Both groups are growing. But one is growing twice as fast measured in percentage points. Erik Fjærli, section chief at SSB, points out that the largest firms' lead grew further this year, and ties it to large firms having more resources and stronger expertise to adopt new technology. That last sentence is worth sitting with. Resources and expertise aren't traits of an industry or an ambition level. They're traits of size. And size isn't something an SMB can simply choose next quarter.

The list of barriers reads suspiciously like a service specification

For 2026, SSB has also mapped out what's actually stopping companies. In table 14944 in Statistikkbanken, the barriers to AI adoption are recorded by industry and number of employees: high costs, lack of relevant expertise, incompatibility with existing equipment, software, and systems, and difficulty accessing or ensuring the quality of the necessary data.

Four barriers. Read them again, but this time as a description of what an accounting firm does for a living.

1. High costs are a barrier when the investment has to be carried by a single income statement. They're something else entirely when spread across three hundred clients in the same portfolio.

2. Lack of expertise is a barrier when a fourteen-person company has to hire someone who understands both finance and AI. Expertise can't be divided into fourteenths. It can, however, be shared across a client portfolio.

3. System compatibility is a barrier for whoever owns a single system. The firm owns the stack, and in practice decides which systems its clients end up on.

4. And data quality is the most obvious of all. The firm already produces the client's structured data: vouchers, chart of accounts, payroll, VAT, accruals. In many SMBs, there isn't a single other dataset with anywhere near that level of standardization.

Four barriers, four points where the accounting industry has a structural advantage the client can't replicate on their own.

That's why the gap doesn't close from the inside

This is where the common interpretation breaks down. A gap caused by fixed costs and indivisible expertise doesn't close because the weaker party tries harder. It closes because someone else takes on the cost and sells the result onward. That's also how technology has historically reached the small-business segment. Few SMBs have built their own ERP systems, their own payroll engines, or their own reporting solutions. They bought them, often through the advisor who already held the data. The fact that the SSB numbers show growth even among the smallest firms suggests AI is arriving at SMBs the exact same way: baked into systems and services someone else has set up. The question, then, isn't whether SMBs adopt AI. It's who they pay for it.

Internal efficiency is a margin, not a revenue line

The way AI is used across much of the industry today, it's an internal tool: faster voucher processing, automated reconciliation, document understanding, less manual entry. The gain is real and immediate. It's also fleeting. Internal efficiency sits on the cost side. As more firms achieve the same gain, it disappears into the price, either through lower hours billed on fixed-fee work, or through competition for the same clients. Efficiency can be sold once. It can't be sold every year.

External insight can. The difference between the two isn't technological, it's commercial: in one case, AI is something the firm does, in the other, it's something the client buys. And what the client buys isn't a model. It's answers. How long liquidity will hold. Which clients are actually profitable once time spent and follow-ups are accounted for. What happens to the margin if the purchase price rises nine percent. Which deviations in this month's numbers are worth reacting to. These are questions SMBs have always had. What's new is that the data needed to answer them now sits, fully structured, in the accounting system, and that the cost of answering them has fallen dramatically.

Trust is the part competitors don't have

When data quality and data access are listed as standalone barriers in SSB's mapping, that also says something about who the SMB lets into its numbers. The accounting industry has an authorization regime, confidentiality obligations, liability insurance, and an established position as the party that already knows the client's finances. Pure technology vendors have the models but have to build the trust. Firms have the trust, and can buy the models. That's not a symmetrical situation.

The objections are worth taking seriously

Three caveats belong here.

First, SSB measures use of AI technology as a category. That says little about how deeply the technology actually sits in day-to-day operations, and some of the growth in both groups may be functionality that arrived baked into software companies already used.

Second, the barriers in table 14944 apply to the accounting industry itself too. A great many Norwegian firms are themselves small companies, with exactly the same cost and expertise constraints as their clients. The scale advantage isn't automatic, it only appears once the same solution is delivered identically to many clients. Without standardization, a twelve-person firm is just another small business in SSB's bottom category.

Third: insight delivered for free inside an existing fixed fee doesn't become a service market. It becomes a cost the firm has taken on on someone else's behalf.

Why 2027 is the budget year that matters

The most interesting part of SSB's release isn't the level, it's the direction. The gap between large and small grew from 2025 to 2026. If the trend continues, the gap will be wider still when the 2027 numbers are published, and willingness to pay among an SMB segment watching its competitors pull ahead will be higher than it is today. A gap that widens year after year is, in practice, a market that grows year after year. What decides whether the accounting industry gets paid for it is whether AI expertise stays an internal operational advantage inside the firm, or leaves the building as a packaged, priced, billable service.

SSB has delivered the numbers. The industry writes the price list itself.