Mit: 95% of GenAi business projects do not produce value, only 5% bring returns
Unofficial AI use among employees exceeds licences, influencing job change more than official projects. This is the finding of the report 'State of AI in Business 2025', compiled by the research team Mit Nanda (Networked Agents and Decentralised AI) on more than 300 AI initiatives, 52 company interviews and 153 questionnaires collected at industry conferences
2' min read
2' min read
Despite investments of between USD 30 and 40 billion by companies, 95 per cent ofgenerative artificial intelligence (GenAI) projects do not produce any concrete impact on company accounts. This is the finding of the report 'State of AI in Business 2025', compiled by the Mit Nanda (Networked Agents and Decentralised AI) research team on more than 300 AI initiatives, 52 company interviews and 153 questionnaires collected at industry conferences.
The research highlights a clear divide, dubbed the 'GenAI Divide': only five per cent of integrated pilot projects generate value, with returns in the millions, while the vast majority remain stuck at the starting line, with no effect on company accounts. It is not 'model quality' or 'regulation' that are the determining factors, according to the research, but the approach taken by companies.
Reduced testing
.Popular tools such as ChatGPT and Copilot have exceeded 80% adoption and nearly 40% deployment, but they only affect individual productivity, not operating margin. Customised enterprise systems, on the other hand, are often rejected: 60% of companies evaluate them, but only 20% get to test them and barely 5% take them into production. The main reasons are the difficulty of integrating them into workflows, the lack of contextual learning and the lack of adaptability.
The phenomenon is reflected at sector level: out of nine sectors analysed, only technology and media show signs of structural transformation. In healthcare, finance, retail and manufacturing, adoption remains superficial, with experiments without any particular effect. Concrete savings emerge mainly from the reduction of external service costs: between USD 2 and 10 million per year in customer service and document processing and up to 30 per cent less costs for creative agencies.
An executive of a medium-sized manufacturing company summed it up like this: 'The enthusiasm on LinkedIn says that everything has changed, but in our operations nothing is really different. We process some contracts faster, but that's about it'.
