Summary

• OP Pohjola’s Business Pulse survey finds only 22% of respondents prioritise business growth over internal efficiency in their use of AI.
• One in three companies says caution guides AI investment more strongly than the pursuit of new opportunities.
• The findings suggest adoption is advancing faster than companies’ willingness to rebuild products, revenue models, or operating structures around the technology.

Finnish companies are investing in artificial intelligence, but most are still using it primarily to make existing operations more efficient rather than treating it as a route to new growth.

Research from OP Pohjola finds that only 22% of surveyed companies prioritise business growth over internal efficiency when considering AI, while another 28% do not take a position either way.

Among Finland’s largest companies, defined in the reporting around the survey as businesses with turnover of at least €200 million, half are not primarily using AI to pursue growth. One in three respondents also says caution guides investment more strongly than the pursuit of opportunity, while 45% disagree and 21% remain neutral.

The findings complicate the assumption that wider adoption automatically means companies are restructuring themselves around AI. Employees can use general-purpose tools extensively for writing, research, document analysis, search, or administration while the organisation continues selling the same products through much the same operating model.

Efficiency is the easier entry point

Productivity applications are comparatively straightforward to test because the organisation can measure whether individual tasks become faster without changing the systems through which its products and services are delivered. A pilot can begin with limited integration and modest governance if the consequences of a poor output remain easy to correct.

Growth-oriented applications generally demand more, because a company may need to connect models to proprietary information, embed them in customer-facing services, change decision processes, or introduce automation into operational systems where errors carry financial or regulatory consequences.

Those projects expose weaknesses that an employee productivity tool can avoid. Data quality, integration with older systems, access permissions, model monitoring, cybersecurity, regulation, and responsibility for automated decisions all become harder once AI moves inside revenue-producing or customer-facing processes.

Finland’s high rate of AI adoption therefore does not guarantee an equivalent change in business performance. Finnish AI Region notes that 38% of companies with at least ten employees used AI in 2025, placing Finland near the top of the European Union, while OP Pohjola’s current findings suggest much of that use remains concentrated on efficiency.

The same reporting says only 15% of respondents can reliably measure AI’s impact, even though nearly half believe they know what type of business value they want from the technology. Earlier OP Pohjola and Accenture research similarly found operational efficiency dominating AI programmes while relatively few organisations had a clear strategy or measurable business impact.

Growth requires harder integration

Efficiency improvements can still have substantial value because lower administrative effort can improve margins, speed up internal work, and free scarce specialist staff for tasks that are harder to automate. Those gains do not need to create an entirely new revenue stream to justify investment.

The competitive limit appears when the same general-purpose tools become available to every rival. If most companies can buy similar assistants from the same model and software providers, individual productivity gains can become part of the normal cost base rather than a durable source of differentiation.

More defensible value is likely to depend on how well businesses connect AI to proprietary information, specialist processes, customer relationships, products, or operating knowledge that competitors cannot reproduce simply by purchasing another subscription.

That transition also changes the risk calculation. An inaccurate internal summary can be checked by an employee, whereas a poor decision embedded in pricing, healthcare, manufacturing, procurement, lending, or customer service may affect revenue, compliance, or safety before a person notices.

Caution is therefore not necessarily evidence of organisational resistance. Some restraint reflects the engineering and governance work required to move from general tools towards systems that can operate reliably inside core business processes.

Excessive caution can become a constraint of its own, however, if companies remain permanently at the assistant stage while competitors build AI into products and operations that produce materially different services or cost structures.

Finland’s current position sits between those outcomes. Businesses are adopting AI at a high rate, but the spending pattern suggests that many remain focused on making the organisation they already have more efficient, leaving the harder question of AI-led growth largely unresolved.