From tokenmaxxing to sovereign alpha: Who controls your AI economics?
The business move
Canva trimmed its 2026 revenue-growth forecast from 30 percent to 20 percent, citing the high cost of its AI-powered features. The company’s experience exposes a hard truth behind the popular metrics used to measure AI progress: token consumption, model calls, and usage volumes mainly benefit AI vendors’ revenue models rather than enterprise economics.
Why it matters
Measuring AI success by tokens processed or API calls made pushes companies into what can be called “tokenmaxxing,” a model focused on maximizing vendor-driven usage metrics. This inflates AI providers’ income but often obscures the true cost-benefit balance enterprises face. The Canva example shows those AI costs can bite deeply, forcing companies to dial back growth expectations when AI workloads become too expensive to sustain. This signals a growing tension between adopting advanced AI capabilities and maintaining financially sound growth.
Who gains and who gets squeezed
AI vendors gain when enterprises measure progress by tokens or calls since they directly monetize those metrics. Enterprises that rely heavily on third-party AI in their products or workflows absorb the rising costs, shrinking their financial leverage. Customers who benefit from AI features may face slower innovation or price increases as companies try to cover these expenses. The Canva case illustrates how enterprises risk being squeezed between internal demand for AI-driven innovation and vendor costs that strain margins.
What to watch next
The unfolding challenge is whether enterprises can shift control of AI economics away from vendors’ usage-based billing and toward models that align AI investments with clear enterprise value creation. Watch for new pricing strategies or AI infrastructure options that aim to decouple cost from raw usage. Enterprises will also be testing how to deploy AI more cost-effectively, designing product features and internal workflows that optimize ROI on token usage.
AI Quick Briefs Editorial Desk