AI Token Usage Surge Tests Bosses’ Bet on AI

At the enterprise software firm 8×8, employees have embraced Anthropic’s Claude for tasks ranging from drafting emails and analyzing customer feedback to writing code. Yet despite this growing dependence on generative AI, the finance department isn’t sweating the costs. While tech giants like Meta, Uber, and Salesforce have publicly worried about rising AI expenses and started imposing usage limits, 8×8 reports that its investment is actually generating savings.

Over the past 18 months, 8×8 estimates it has trimmed roughly $5 million in annual costs by canceling subscriptions to dozens of software and educational tools that became redundant once Claude could handle similar functions. Joel Neeb, the company’s chief transformation and business operations officer, notes that the annualized bill for Claude remains “well below” that figure. He expects the savings and costs to eventually converge as 8×8 encourages broader AI adoption and integrates the technology into more complex workflows. For now, however, the gap remains wide, which “makes my chief financial officer happy,” Neeb tells WIRED. He declined to disclose exact spending on generative AI.

As companies collectively funnel hundreds of millions into AI tools for coding, marketing, and customer service, a new fixation has taken hold in tech: tokenomics, or managing the skyrocketing cost of AI usage. (Tokens measure the volume of content an AI model processes and generates.) Last month, Royal Bank of Canada’s CEO revealed that its token usage had surged 500 percent over six months. At Cisco, a third of employees use an internal AI chatbot daily, making “the token usage … getting pretty, pretty crazy,” CEO Chuck Robbins said on an earnings call. Amplitude CEO Spenser Skates reported that some top engineers are “spending thousands of dollars a month or more on tokens.” Aaron Levine, CEO of Box, called the token budgeting conversation “one of the most important” and “heated” topics.

A WIRED review of transcripts from AlphaStreet shows that roughly 300 companies addressed AI token concerns during earnings calls or public discussions with analysts in April or May, up from just 93 a year ago. While this remains a small fraction of the thousands of calls held, the jump signals growing unease. Executives at several firms said they are developing or seeking systems to monitor token usage and select the cheapest model for each prompt. Others admitted they are still weighing whether to hire more people or increase token budgets to meet their goals.

Software has never been cheap, but the latest AI tools are causing unusual stress in C-suites for several reasons. Prices keep fluctuating, new and more powerful models debut monthly, and getting entire organizations to adopt new workflows remains difficult, meaning productivity gains on one team can create bottlenecks elsewhere.

Despite these headwinds, some companies are actively encouraging employees to use AI without worrying about costs. In April, Baseball Lifestyle 101, a Long Island-based clothing brand expecting $250 million in sales this year, told roughly 50 top managers to spend the equivalent of 20 percent of their salary on AI tokens each month. Co-founder and Chief Strategy Officer Bill Rom tells WIRED the monthly cost could exceed $100,000 by year-end, but it’s already paying off. Claude recently helped secure a $1 million order by spotting that a retailer was low on sizes of the company’s popular ice-cream-patterned shorts. “That’s a day and a half of work that can now happen in an hour or two that might make me eight figures of additional revenue over 12 months,” Rom says.