The jump to capable AI agents has shifted enterprise cost structures. AI is no longer a predictable per-seat software license but a variable consumption cost, akin to labor. This explains why companies are suddenly "torching" their budgets—they were budgeting for tools, not autonomous workers.

Traditional software budgeting fails for generative AI, where costs are variable and tied to tokens and usage. A CFO noted a team's daily per-person cost jumped 50% in one week. Companies must accept this volatility, run pilots to establish baseline costs, and then determine ROI, rather than trying to set a fixed budget upfront.

The end of subsidized AI pricing is forcing companies to confront its true operational expense. As AI bills begin to rival payroll, a fundamental transition is occurring where capital expenditure on silicon (CapEx) is displacing operational expenditure on human neurons (OpEx), reshaping corporate budgets.

An anecdote about an engineer spending $100M in a month on AI tokens reveals a core enterprise issue. For Lenovo's CFO, the problem isn't the amount but its lack of planning and clear ROI. This signals a shift from predictable software subscriptions to volatile, usage-based AI compute costs.

The shift to agentic AI means costs are no longer predictable per-seat subscriptions but variable expenses based on usage (tokens, compute). This requires managing AI like a capital allocation or a new form of labor, not just another software tool, a reality that early adopters are now grappling with.

The shift from predictable seat-based software to consumption-based AI creates massive financial uncertainty. One CFO reported needing to budget for AI compute costs within a 400% range of certainty, making traditional financial planning nearly impossible and highlighting the extreme volatility of "token maxing."

Companies should reframe AI spending not as a traditional IT cost but as a direct investment in amplifying human capital. This model views AI agents as 'digital workers' that provide leverage to every employee, justifying spend based on the ROI of the augmented workforce.

Just as uncontrolled cloud spending in the 2010s spawned the FinOps field, the shift to consumption-based AI pricing will necessitate a similar discipline. This involves attributing costs to specific workloads, setting granular budgets, and providing real-time visibility to prevent budget overruns and measure ROI accurately.

To solve the challenge of budgeting for AI, Andrew MacDonald proposes a novel approach: merge the headcount and compute budgets into a single pool. This forces leaders to make direct trade-offs between hiring more engineers and spending on AI models, ensuring they allocate capital to the highest ROI activities.