Every technology shift eventually becomes a business problem. With AI, that moment is now and it’s arriving faster than most leadership teams are prepared for.

For years, the business case for technology investment followed a familiar pattern. You identified a problem, selected a solution, negotiated a contract, and your finance team built it into the plan. Costs were predictable, returns were measurable, and the commercial model largely took care of itself.

AI is now changing this for every business, with the challenge now on whether the financial infrastructure that most businesses rely on is built for what AI costs to run.

The economics of AI don’t run themselves  

The margins tell the story clearly. Businesses built around traditional software models operate at margins of around 80%. AI-driven businesses are running at 20-30%. That gap has consequences that reach well beyond the technology sector into any business now dependent on AI tools, platforms, or services.

The conversation in Boardrooms on AI still centres on adoption and capability, and there is far less attention to what’s happening under the hood - a cost structure that is different from anything that businesses have had to manage before, because AI now scales rapidly and requires continuous model management and commercial controls that most organisations simple haven't built.

Is the real world catching up?

AI adoption is accelerating faster than companies can financially control it.  Earlier this year, Uber burned through its projected AI budget far faster than expected because the commercial controls were not in place. Finance teams couldn't model usage the way they could model a seat or fixed fee. When the cost of the AI model changed, the economics downstream also moved with it, often with little warning.

This month, GitHub also announced a move to a significantly more expensive usage-based model. A signal of a broader repricing happening across the AI industry.

The question for every CFO is the same: when your suppliers change how they charge, how quickly does your business feel it?

The billing problem most businesses haven't faced yet

Most organisations have adopted AI through the tools and platforms they already use. These are often software subscriptions that have added AI features, new productivity tools and workflow automation. The cost of these tools is rising whether businesses are actively using the AI features or not. They are often absorbed into subscription renewals and contract uplifts, often before finance teams have asked whether the investment is delivering any return.

For context, the software industry built its commercial models on seat-based subscriptions. Simply, a fixed number of users paying a predictable monthly fee. That model is being dismantled and replaced with usage-based pricing, where costs fluctuate based on how much AI is actually consumed such as tokens processed, tasks completed and API calls made. For the finance teams managing these relationships, it introduces a level of variability that traditional budgeting methods were never designed to handle.

The result is a complexity that compounds quickly. A SaaS business at 80% gross margin can absorb some slippage but an AI based business at ~20% can't. At those margins, the billing gaps that were rounding errors now become materially expensive as the business scales.

Businesses we work with are now running four or five pricing models simultaneously as their commercial model aims to keep pace with a product reality it was never designed for. Seat-based subscriptions are now being layered with usage thresholds, token consumption, credits, annual commitments, overages, and hybrid pricing structures that evolve continuously as products mature.

The CFO conversation can’t wait

If you’re going through this change, business leaders need to ask the harder questions earlier - how are we being charged, how is that likely to change, and does our financial infrastructure give us the visibility and control to manage it?

This means finance teams need to go beyond headcount and fixed licences in their forecasting. It means procurement teams understand usage-based contracts before signing them. It means CFOs treating AI spend not as a technology budget item, but as a variable cost that needs active commercial management.

The first CFOs to actually look at their AI profit and loss are already having those important conversations, building the controls and billing infrastructure to support it. The companies that don't build the commercial rigor to match will find that out the hard way- when the invoice arrives. And this is probably sooner than expected.

‍

Solvimon is a billing and payments infrastructure for companies that have outgrown their first billing system. Founded by Kim Verkooij (CEO) and Etienne Gerts (CTO), both former Adyen executives, the company handles hybrid AI pricing models from seats, usage, credits, tokens, and commits in a single platform. Solvimon serves companies navigating the transition from simple subscriptions to complex, multi-model pricing. The company is backed by Northzone, and headquartered in Utrecht, Netherlands.

‍