More than 40% of agentic artificial intelligence projects will be cancelled by the end of 2027 due to rising costs and unclear business value, according to a new report by a US research and advisory firm, Gartner.
Agentic AI refers to systems that can act independently to complete tasks or goals. While companies like Salesforce and Oracle have invested heavily in the technology to improve margins and reduce costs, many current efforts are still in early experimental stages and may not deliver expected results.
Gartner says many vendors are rebranding existing tools like AI assistants and chatbots as agentic AI without offering true autonomous capabilities. The firm estimates only about 130 companies among thousands of vendors actually offer real agentic AI products.
The report finds that many current agentic AI projects are driven more by market excitement than practical use, with limited return on investment. Current AI models often lack the ability to follow complex or long-term instructions, making them difficult to apply effectively in real business settings.
