As tech giants like Microsoft and Google slash jobs while investing billions in artificial intelligence, a new study from the Massachusetts Institute of Technology (MIT) reveals a stark reality: 95% of businesses deploying generative AI are seeing little to no returns.
The report, titled “The GenAI Divide: State of AI in Business 2025”, found that despite U.S. firms pouring $35–40 billion into generative AI projects, most initiatives fail to deliver measurable impact.
Researchers analyzed 300 AI applications, interviewed 150 AI leaders, and surveyed 350 employees. Their conclusion: the majority of corporate AI experiments collapse under the weight of brittle workflows, shallow contextual learning, and poor alignment with day-to-day operations.
“While some companies have seen revenues jump from zero to $20 million in a year, the majority are struggling with a massive learning gap,” said Aditya Challlapally, the study’s lead author.
The research found AI tools were effective in back-office functions such as administration but failed in high-stakes areas like sales and marketing, where human creativity and nuance remain essential.
The report also revealed a sharp divide between winners and losers. Companies that purchased specialized AI tools from vendors saw a 67% success rate, while those developing in-house solutions achieved only 33%. Leaders who invested in adaptive tools and strong managerial adoption fared better, while firms relying on off-the-shelf chatbots like ChatGPT saw little change in productivity.
MIT researchers suggest the next wave of enterprise AI will involve “agentic AI systems”—programs that can learn, remember, and act autonomously—potentially bridging the current gap between hype and impact.
Until then, however, the study warns that most corporate AI spending risks becoming a costly misadventure rather than a driver of growth.
