Model Routing Emerges as Enterprise AI Cost Control Solution, Pressuring OpenAI and Anthropic

According to CNBC, AI leaders including Cognition CEO Scott Wu and Cisco CPO Jeetu Patel revealed this week that model routing—a solution that matches tasks to appropriate AI models based on complexity—is becoming a key cost discipline in corporate America.

Currently, approximately 95% of enterprise AI usage defaults to expensive frontier models even for routine tasks, according to Glean CEO Arvind Jain. Model routing could deliver 5 to 10 times better cost efficiency on boilerplate work using cheaper alternatives. Cisco's cost analysis underscores the pressure: at $200 token usage per employee weekly, the company faces roughly $900 million in annual AI spending for its 90,000 employees. If enterprises increasingly route high-volume, easy tasks to cheaper or open-source models, OpenAI and Anthropic would receive payment only for complex work, potentially reshaping the pricing models both companies have built their valuations upon.