Finnish companies are buying AI, not building it
Finland offers extensive R&D support, yet too few companies use it to turn AI adoption into sustained innovation and business value.
Text: Dr Umair Ali Khan, 22.9.2026 | Photo by Adobe Stock Photos (edited with AI)
Finland has set a national target of raising its research and development expenditure to 4% of gross domestic product by 2030. Reaching it will require Finnish companies to turn new technologies into better processes, products and growth. AI is the most immediate test of whether this investment can lead to genuine business renewal.
Companies no longer need to be convinced that AI matters. Its use is growing rapidly, and the share of regular users has more than doubled in one year. But adoption is advancing faster than the research and development needed to turn AI into new products, redesigned processes and sustainable competitive advantage.
Using AI is not the same as developing a distinctive capability with it. Buying the same AI assistant as one’s competitors cannot necessarily create a lasting advantage. Development helps a company identify a valuable problem, build and test a solution, measure its effects, and refine it through several iterations. It also keeps the solution useful as business needs, data, models, and regulations change.
What looks simple rarely is
One of our industrial projects shows why this matters in practice. A partner brought us a simple-sounding request to turn customer purchase orders into sales orders that could pass straight into its own system. In practice, the solution had to extract item details from documents in many formats, price each item against a catalogue in a traceable way and generate a final sales order with reliability measures.
The real challenge was handling the diversity of real-world documents and reproducing the judgement employees used when information was incomplete or inconsistent. That is not a software purchase. It is a cycle of building, testing, measuring, and improving.
Yet many companies still expect AI development to work like a conventional software purchase. In my work recruiting companies for EU-funded R&D projects, I often encounter demand for a quick, inexpensive product, but less willingness to commit staff and management attention to a two- or three-year co-development process.
Even large companies struggle to commit
In my experience as an adviser, underused collaboration with universities and research institutes is one reason AI adoption is not producing scalable solutions. The reluctance is understandable.
Free support is not the same as costless support. Even when the expertise is paid for from public funds, someone in the company must define the problem, prepare the data and run the project, and then wait years for the benefit. Small, owner-led companies rarely have anyone to release for that.
I have found even large enterprises struggling to name anyone willing to own the work despite having financial resources and people, let alone SMEs.
The innovation system adds friction of its own. Funding applications take a long time, and academic timetables do not match industrial ones. A fixed project duration of two or three years can be too long for some companies. In certain funding instruments, the requirement for company co-financing is also a barrier, particularly for smaller firms with limited cash flow.
Start small, stay accountable
So, the problem is not the absence of support. It is the absence of a form of engagement that matches a company’s capacity. That engagement should begin with a manageable step rather than a large funding application.
A company can start with one measurable business problem and a short readiness assessment, then test the idea through a thesis, student project, prototype or proof of concept. Only when the evidence is promising should it commit to a larger publicly funded project.
Starting small does not remove the company’s responsibility. It still needs to appoint someone to lead the development, make relevant data available, and reserve employee time for experimentation. Publicly supported expertise can reduce the risk, but it cannot replace commitment from the company.
Research and support organizations must, in turn, reduce the cost of participation. Companies need faster contracting, staged projects, help with applications, and clear commercial milestones. Success should be measured through implemented solutions, new products, productivity and growth, not simply through the number of events or consultations.
Finland has already built much of the expertise and infrastructure that companies need. The next task is to shorten the distance between a company’s business problem and the knowledge available to solve it. Publicly supported services create value only when companies turn them into better processes, differentiated products, and sustainable growth.
