On July 3, The Information reported that Tesla recently notified its employees of a new policy. Starting from July 6, the weekly cap on internal expenses for AI computing power and model invocation will be set at $200. Any costs exceeding this limit will require prior approval. Previously, engineers frequently racked up AI invocation costs that reached thousands of dollars per week. Over the past six months, Tesla has been actively developing its unified AI platform, dubbed "Bottle Rocket." This platform integrates multiple models from sources such as OpenAI, Anthropic, and xAI, facilitating a shift from decentralized AI usage to centralized management across the entire company. However, as AI becomes increasingly prevalent in customer service, sales, and automation processes, the associated cost pressures have become more pronounced, prompting Tesla to tighten its budgetary controls. This trend is not exclusive to Tesla; other major companies like Meta, Uber, and Walmart have also been reining in their AI expenditures. For Tesla, this adjustment is especially critical, as CEO Elon Musk has repeatedly emphasized that the company's future value will primarily hinge on AI-driven products such as the Robotaxi and Optimus robots, rather than traditional automobile sales. Notably, Tesla's revenue has plateaued for two consecutive years.