Box CEO says companies will need to figure out how to budget for workers running up AI token bills

In an era where artificial intelligence is transforming industries, Box CEO Aaron Levie emphasizes the urgent need for companies to reassess their budgeting strategies for AI usage. Levie predicts that the surge in AI-related expenses will initially impact engineering departments, where developers often run several agents simultaneously or work on projects overnight. However, he warns that this trend will eventually extend to all sectors of knowledge work, including fields like legal and sales, which could become significant consumers of AI resources. Tokens, the units that measure AI consumption, play a crucial role in this financial landscape. These tokens—representing words or parts of words—are fundamental to Large Language Models that fuel well-known chatbots like OpenAI's ChatGPT and others. Notably, the pricing of these tokens varies, with more complex queries and advanced models resulting in higher costs. Levie's comments come in response to concerns raised by Nvidia CEO Jensen Huang, who expressed alarm that an engineer earning $500,000 might not be utilizing AI tokens commensurate with that salary. Huang highlighted the expectation that such high-earning engineers should be accountable for their token expenditures, with any significant shortfall leading to serious discussions about their efficiency. As AI technology continues to evolve, Levie believes that workers who effectively leverage these tools will inevitably increase their token consumption. He notes that the compute budgets for these employees are likely to rise consistently over time. Moreover, he points out that AI agents, which can operate around the clock, are expected to be the largest consumers of tokens. Unlike basic chatbot interactions, these agents are capable of processing vast amounts of data and creating innovative solutions at scale. However, the financial implications of these advancements are not universally welcomed. Venture capitalist Chamath Palihapitiya shared his experience with his startup, 8090, revealing that they had to halt the use of Cursor, a widely-used AI coding tool, after witnessing a dramatic increase in token-related expenditures. He noted that their costs had more than tripled in just a few months, highlighting the significant financial burden associated with AI technologies. Ultimately, Levie suggests that companies will need to adapt their budgeting frameworks to accommodate these rising costs. He speculates that AI spending may not be categorized solely under IT budgets but could instead fall under broader business allocations, potentially shifting financial responsibilities to CFOs in the process.