The cost landscape: Tokens, cloud and the software pricing opportunity

Across AI deployments, cloud consumption, software licensing, and vendor pricing, AI has made enterprise technology costs harder to predict, govern, and negotiate.

Technology leaders could once forecast most costs three years out, but now even quarterly reviews can be insufficient to manage the situation. The real constraint is rarely the technology. It is the absence of mechanisms to ensure spending is governed as a financial variable rather than absorbed as a budget line.

Organizations staying ahead of technology costs are treating them as a governance problem, not a procurement one, and tie technology spend to measurable business outcomes. Private equity has been the most visible enforcer, treating shadow IT, bloated application stacks, and redundant tooling as failures of governance rather than inevitable features of a large organization.

From cloud-smart to cloud-sovereign 

The cloud-first mandate of the last decade has been quietly retired. Large enterprises have settled on a cloud-smart posture, using cloud selectively for elastic and bursty workloads while keeping predictable, stable workloads on premises or in a colocation. But a newer pressure is reshaping that calculus. Concerns about data jurisdiction, supply chain dependency, and geopolitical exposure have elevated sovereignty to a board level concern. Certified government cloud variants carry a cost premium of 20 to 30% over standard commercial cloud. That premium is the price of sovereignty, and AlixPartners is helping clients model those trade-offs explicitly before they become embedded in architecture decisions they cannot easily reverse. 

The opportunity in software pricing 

The per-user model that has dominated enterprise software licensing for two decades is becoming economically unviable for vendors. When a single user with AI tools can do the work of multiple people, per-seat revenue no longer grows proportionally with the value delivered. Vendors are responding by moving toward outcome and usage based pricing, and this shift is permanent, not tactical. For CFOs who have always had a simple and forecastable budget line for software licenses, it introduces a variable cost that functions more like a revenue sharing arrangement than a purchase. Understanding exactly how platforms are used, what value can be credibly attributed to them, and what the vendor's own economics look like is the foundation of negotiating that shift from a position of strength. 

Managing AI spend before it becomes the next cloud crisis

Token costs are the variable that most AI budgets are not built around. Unlike cloud compute or software licenses, token consumption lacks visibility, varies by model and usage pattern, and the consequences of getting it wrong can hit hard and fast. In one organization AlixPartners worked with, the first four months of AI deployment consumed the full year's AI budget. AI spend is following the same curve that cloud spend did a decade ago, but steeper. FinOps emerged as the discipline that brought financial governance and accountability to cloud expenditure. A similar level of discipline is required for AI spend except the window to apply it is much smaller.