Disney's internal AI dashboard shows how quickly enterprise AI has moved from pilot projects into daily technical work, with Claude and Cursor usage already running at massive scale.
Disney's technology teams are not just experimenting with generative AI. They are using it heavily enough for the company to track requests, tokens, power users, and costs through an internal AI Adoption Dashboard.
According to Business Insider, screenshots from the dashboard captured usage across roughly 4,800 product and technology employees in Disney Entertainment and ESPN over nine workdays in mid-April. Claude was invoked 1.4 million times during that period and processed 3.1 billion tokens. The most active Claude user alone made 460,600 calls, an average of more than 51,000 per workday, and consumed 234.2 million tokens.
Cursor, the AI coding tool, accounted for even more activity with 13.3 billion tokens. Based on dashboard data, one employee estimated costs at about $1 for every 16,700 Claude tokens and $1 for every 21,200 Cursor tokens. Applied across the reported usage window, that would put Disney's estimated spend at roughly $185,000 for Claude and $627,000 for Cursor. The snapshot also comes as Josh D'Amaro takes over after Bob Iger, making AI usage one of the operational questions now sitting near the top of Disney's agenda.
The size of the outliers matters because it suggests some employees are not manually typing prompts one at a time. At that level of activity, autonomous agents are the more plausible explanation: software workflows that can assign tasks, call models, test outputs, and repeat the cycle with limited human input.
Disney is hardly alone in trying to make AI adoption visible inside the company. Meta, JPMorgan, Visa, and other large employers have experimented with dashboards, milestones, or usage categories to see who is actually putting AI tools into the flow of work. The culture around those metrics has already picked up a name, "tokenmaxxing," a half-joking label for employees competing to burn through more AI tokens.
That framing can make the trend sound silly, but the business question is serious. Once usage becomes measurable, managers can compare teams, estimate cost, spot unusual patterns, and decide whether AI is improving output or simply creating another internal leaderboard.
Claude vs Cursor Split
The divide between Claude and Cursor also tells a useful story. Claude appears to be handling a broad set of assistant-style and agentic workflows, while Cursor's larger token count points to a heavier role in software development, where code generation, debugging, refactoring, and test iteration can consume context quickly.
For a company like Disney, that matters because the AI opportunity is not limited to creative production or consumer-facing products. A large media and entertainment business also runs streaming platforms, advertising technology, internal tools, data systems, commerce, and sports products. Those are exactly the areas where coding assistants can become everyday infrastructure before the audience ever sees an AI-generated feature on screen.
The spending estimates also look different when placed in that context. Hundreds of thousands of dollars over nine workdays is not trivial, but it is not extraordinary for a Fortune 500 technology operation if the tools reduce engineering bottlenecks or speed up maintenance work. The unresolved issue is whether companies can prove that the token consumption is tied to better shipping velocity, fewer defects, or lower support costs.
Media AI Tensions
Disney's internal embrace of AI sits beside a much more complicated public debate. Media companies are negotiating with AI labs, defending intellectual property, and facing pressure from writers, actors, and artists who worry that automation could erode creative labor or reuse protected work without fair compensation.
That tension is why the dashboard is more than a curiosity. It shows that large studios can oppose unauthorized use of their content while still deploying AI aggressively inside their own operations. Coding, analytics, and internal productivity tools are less visible than synthetic actors or AI-written scripts, but they may be where the earliest enterprise gains are easiest to justify.
Forward Implications
The next phase will be less about whether Disney employees use AI and more about how the company governs it. Usage dashboards can help control budgets and security, but they can also create bad incentives if employees learn to optimize for volume rather than useful work.
For Anthropic, Cursor, and other enterprise AI vendors, Disney is a signal that media giants are becoming customers as much as critics. The market will now watch whether heavy adoption turns into measurable productivity gains, and whether companies can build controls strong enough to manage AI at this scale without turning token counts into the wrong kind of scoreboard.
Also read: OpenAI misses revenue and user targets as IPO looms amid rising competition • O'Leary's 9GW Utah AI campus approval signals hyperscalers must own their power plants • Musk calls Altman 'Scam Altman' as OpenAI trial jury convenes
