en.Wedoany.com Reported - CNBC reported on June 26 that, under the pressure of continuously rising AI bills, an increasing number of U.S. companies are seeking to complete tasks of equivalent complexity with fewer token consumption to control AI spending.

Lindy, an AI startup headquartered in San Francisco with about 25 employees, is a typical example of this trend. The company previously primarily used Anthropic's Claude model. Its CEO, Flo Crivello, revealed that the monthly AI bill had exceeded the total salary expenses of all employees.

Crivello stated that the company switched 100% of its traffic to the DeepSeek model earlier this month, a decision expected to save the company millions of dollars in the coming months. Crivello believes that AI bills are a matter of corporate survival and showcased the cost curve after the model switch, noting a sharp decline in costs. His former employer, Uber, where he worked for five years, is also strictly limiting AI calls, having set tiered spending caps on some AI tools this month, with a basic tier of $1,500 per month.

Several consultants and enterprise administrators pointed out that the area where AI spending first spiraled out of control is assisted programming, as developers consume large amounts of tokens for developing new tools and services.

Faced with high costs, companies are beginning to strictly limit usage scenarios and adopt a "model routing" strategy that matches models to tasks, rather than applying the most expensive frontier models to all scenarios.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com