Model routing on AI is a problem for OpenAI and Anthropic

Model routing on AI is a problem for OpenAI and Anthropic

Model routing on AI is a problem for OpenAI and Anthropic

https://www.cnbc.com/2026/06/05/model-routing-on-ai-is-a-problem-for-openai-and-anthropic.html

Publish Date: 2026-06-05 13:06:00

Source Domain: www.cnbc.com

• Corporate America is shifting towards a new spending discipline to rein in artificial intelligence (AI) expenditures that have far exceeded budgets for the past two years.
• A commonly used strategy has been to default to the most powerful AI model but this is changing as companies realize that many tasks do not necessarily require the most expensive models.
• A new solution being adopted is model routing, which allocates tasks to appropriate models based on complexity, thus lowering costs significantly.
• Many companies still do not use model routing, with approximately 95% still using the most expensive AI models for all tasks, even when cheaper alternatives are available.
• AI costs are a growing concern, exemplified by Cisco’s AI spending running much higher than its budget, requiring reallocation of resources across the company.
• AI companies are responding to the cost concerns with initiatives like Autonomy AI’s productivity guarantee, aimed at ensuring a good return on AI investment.
• If companies start utilizing cheaper models for routine tasks, this might impact the premium pricing models offered by top AI venders like OpenAi and Anthropic.
• The success of these new spending strategies could impact the market approach and valuations of leading AI companies.