Strict budget controls on AI usage, such as per-employee spending caps, have a hidden cost. They create a "known ROI bias," pushing employees toward safe, incremental productivity tasks instead of the large-scale, uncertain experiments required to unlock AI's true economic value. This focus on efficiency inadvertently kills breakthrough innovation.

Incentivizing high AI token usage is not waste, but a form of R&D. In the new agentic paradigm, there are no best practices. Mass experimentation, even with failures, is the only way to discover future workflows and avoid being left behind.

To get teams experimenting with AI, leaders should provide an open budget for tokens initially. Being 'profligate' at the start is crucial, as imposing constraints too early leads to unimpressive results, stifles creativity, and hinders true adoption. Efficiency can be optimized later.

To foster breakthrough ideas, companies should initially provide engineers with unrestricted access to the most powerful AI models, ignoring costs. Optimization should only happen after an idea proves its value at scale, as early cost-cutting stifles creativity.

In the AI era, token consumption is the new R&D burn rate. Like Uber spending on subsidies, startups should aggressively spend on powerful models to accelerate development, viewing it as a competitive advantage rather than a cost to be minimized.

Baidu forgoes rigid policies allocating AI compute 'tokens' to employees based on seniority or title. The CFO argues the unit cost of compute drops so fast that such policies become obsolete in weeks. They prefer empowering talent with ample resources, trusting them to prioritize tasks efficiently in a nimble environment.

While large enterprises must constrain AI model usage to control costs, startups should embrace 'token-maxxing.' By giving developers unfettered access to the most powerful models, startups gain a crucial productivity and talent-attraction advantage over larger, more bureaucratic competitors.

The move from pre-agentic to agentic AI workloads consumes massive resources. This has ended the 'AI subsidy era,' forcing companies like Walmart and Uber to implement usage-based models and strict caps on AI spending to control runaway costs and enforce discipline.

Despite fears of runaway costs from "token maxing," enterprises are overwhelmingly encouraging more AI model consumption. A developer survey found 7x more companies were told to increase spending. The value gained from experimenting on AI's rapidly expanding capability frontier currently outweighs the push for cost optimization.