Top AI spenders cut per-employee costs by nearly 10 percent in August (the-decoder.com)

In August, the top 1 percent of AI-spending companies in the U.S. cut their per-employee AI costs by nearly 10 percent, signaling a shift in how these firms allocate their resources for AI services. According to Ramp's AI Index for September 2026, spending among these companies dropped to a median of $7,205 per employee, influenced by seasonal factors such as vacations and a strategic migration towards cheaper AI models. While usage continues to grow, companies are opting for less expensive options, with significant declines in the consumption of costly frontier models, which fell from 53 percent to 45 percent of total token usage over just one month. This trend is significant for the AI/ML community as it reflects a broader reevaluation of AI investments amidst falling token costs—down 41 percent since March. The effective price per million tokens is now $0.68, prompting companies to adopt more affordable models like GPT-5.6 Terra and Claude's Sonnet series instead of frontier models. The data indicates that while adoption rates of expensive models might be flattening, this transition to standard models underscores evolving preferences in AI utilization and signals potential challenges for high-cost model providers. As key players navigate this landscape, future developments—especially with planned IPOs like Anthropic—could provide further insights into market dynamics.