Tesla caps employee AI spending as firms grapple with soaring token costs

Tesla announced that, starting July 6, each employee will be limited to $200 per week for third‑party AI tools such as Anthropic’s Claude, OpenAI’s ChatGPT and Google’s Gemini; the cap does not apply to Tesla’s own system, Grok. The move follows similar limits introduced by Uber, Meta, Walmart and Coinbase as companies confront rapidly rising AI usage fees.

Palantir CEO Alex Karp warned that the industry’s “token‑maxing” model is unsustainable, highlighting the steep cost increases for AI compute. An analyst’s forecast projects that by 2029 AI token spend per engineer at top‑tier firms could exceed the engineers’ salaries, with Anthropic’s compute costs already 2.3 times payroll. Median firms now spend about $137 per engineer, while the leading 1% spend roughly $8,900 annually. Scenarios suggest AI spend may become a structural expense comparable to wages, especially as open‑source alternatives pressure pricing.

Together, the policy changes and cost projections illustrate a broader shift: large tech firms are tightening AI budgets and evaluating the long‑term financial impact of token‑driven AI services.