Forrester Research published a model in 2024 that matters directly to public buyers, even though it reads at first like an industry analysis for IT vendors. Generative AI deflates existing service revenue by up to thirty percent at first. Tasks that used to require manual work get automated. Only after that do new revenue sources appear: demand unlocked by lower prices, AI-readiness services, new managed services. Industry-wide revenue ends up higher overall. But getting there runs through a phase where somebody pays the difference, or pockets it.
Applied to the German public IT market, this is not an abstract number. Federal, state, and local government together spend roughly 20 billion euro a year on IT services. Thirty percent of that is 6 billion euro. That is the amount vendors save every year through AI productivity gains. Under classic time-and-materials contracts, that entire sum stays with the vendor, because no contract gives the buyer a claim to any of it. The public authority keeps paying the old price for the same service while the cost of delivering it falls behind the scenes.
The other side of the ledger is larger still. Unlocked demand, AI-readiness services, and new managed services add up to a further 8 billion euro of potential per year. That potential is only accessible to the public sector under outcome-based procurement. The "new managed services" lever alone is worth an estimated 4 billion euro a year in the German IT market. Together, 14 of the 20 billion euro spent annually are in play, split between quiet losses and gains nobody is claiming.
"An agency that keeps buying AI-enabled services on classic contracts, without actively managing any of the Forrester levers, quietly loses the productivity dividend to vendor margin, barely noticeable for five years, structural after that."
The fix needs no new agency and no new budget, only a different contract structure: tenders that specify outcomes instead of technologies, procedures that bind bidders to output baselines, contract clauses that pass a share of AI-driven productivity gains back to the buyer, and oversight built on automated measurement instead of manual review cycles. The market is already moving this way. Outcome-based pricing is growing as a category, because vendors have noticed that buyers pay more for guaranteed results than for uncertain inputs. The public sector can catch up to that shift contractually, instead of just watching it happen.