AMSTERDAM (Reuters) -Nearly every large company to have introduced AI has incurred some initial financial loss, an EY survey published on Wednesday said, often due to compliance failures, flawed outputs, bias, or disruptions to sustainability goals.
Reputational damage or legal problems were less frequently reported. EY, the British business services firm formerly known as Ernst & Young, conducted the anonymised survey among 975 executives overseeing AI at companies with annual sales exceeding $1
billion from around the world in July and August 2025.
Total combined losses were estimated at $4.4 billion, with metrics such as revenue growth, cost savings and employee satisfaction trailing expectations. Despite that, the firms polled were still optimistic that AI adoption would ultimately yield significant benefits, EY said.
"AI is absolutely improving efficiency and productivity, people are doing more, faster. But the value capture lags because those gains are being reinvested into doing more work, not necessarily into cutting costs or driving immediate revenue, said Joe Depa, EY's Global Chief Innovation Officer, in an email to Reuters.
EY's survey was focused on what it terms "Responsible AI" adoption - a series of metrics assessing whether companies have established internal governance policies for AI, communicated clear usage guidelines, and monitored compliance.
Companies with more fully developed "Responsible AI" policies reported stronger performance on sales, cost savings, and employee satisfaction metrics, EY said.
(Reporting by Toby Sterling, Editing by Louise Heavens)
