AI model routing tests business flexibility, IBM exec says

AI model routing is becoming a test of whether companies can adapt as AI models, costs and use cases change, Neil Dhar, senior vice president of IBM Consulting, argued in an IBM newsroom article published on August 31, 2026.

Dhar said companies have spent recent years trying to decide which large language model to build around, but that improving open alternatives and more commoditized models mean one model is unlikely to dominate all business uses.

He cited Axios as reporting that businesses are increasingly using routing to automatically match AI workloads with models based on factors such as cost, performance and security.

Dhar also cited IBM Institute for Business Value research saying that by 2030 most organizations will primarily use smaller, customized models as chief executives move toward hybrid AI strategies using both foundation models and specialized models.

Dhar said model routing is valuable because it lets companies redirect work as performance, costs and business needs change. He argued that the same flexibility should apply to investment decisions, business processes and workforce skills, so companies are not locked into assumptions that may quickly become outdated.