How to read this page
Adoption counts documented deployments: each one is a named organisation's AI use case described in a public source, such as the company itself, established news media, a government body, academic research or a vendor case study that names the customer. The counts show where deployments are published, which is not the same as market share. Large, English-language and US/UK organisations publish more.
Value is modelled, not company-reported. Every case goes through the same standard model: set-up and running costs come from company size and complexity, and benefits from the type of AI, company size and the strength of the evidence, ramped up over five years. Return per €1 is the modelled five-year benefit divided by the five-year total cost of ownership. Profitable in 5 years means the net present value is positive at a 10% discount rate. Use these figures to compare options against each other, not as forecasts for a specific company.
New use cases are researched daily with AI-assisted web research. Each one must cite a source page that was actually retrieved. The library currently holds — cases (updated —).