Why Does Finland Burn Hundreds of Millions on IT Projects When AI Could Do It Better?

Finnish public sector IT projects have a long history of overruns. Hundreds of millions of euros, multi-year delays, scope blowouts that become punchlines in the press. The pattern is so familiar it almost feels normal. It is not.

The old model

The traditional model assumes that we know exactly what we want before we start, that requirements can be frozen, and that vendors can deliver against detailed specifications. None of these assumptions held in 1995. They certainly do not hold in 2026.

What AI changes

AI-native development is iterative by design. You ship a small slice, measure it, learn, and adjust. Procurement should follow the same logic: smaller contracts, faster cycles, real exit clauses, open-source defaults where reasonable. The biggest risk is not that AI projects fail — it is that we keep procuring them the way we procured the previous generation.

The fix is not technical. It is organizational. And it starts with the buyer side.