Purple sea creature

The variable cost of AI features: the next pricing revolution

As AI becomes increasingly embedded in software products, it is also reshaping the economics behind them. Unlike traditional software infrastructure, AI costs are variable, usage-driven and often outside a company’s direct control. Yet many commercial agreements and pricing models have not evolved to reflect this new reality.

Our Technology team has been working closely with several clients on these challenges. Time and again, we see the same questions arise:

  • How should AI-powered features be priced?
  • Are existing indexation mechanisms still fit for purpose?
  • And how can companies protect their margins when underlying AI costs fluctuate?

Cost risks

Token pricing varies

  • Pricing is determined by LLM providers
  • Token consumption is rising as AI is used more frequently and for more complex tasks

Commercial solutions

  1. Credit-based pricing model Customers purchase credits, you define the cost per credit
  2. Tiered volume model Higher usage results in a higher price
  3. Outcome-based pricing Customer pays per successful outcome
  4. Seat-based model with AI quota Standard seat-based model with a monthly usage cap for each user

Contractual protection

Existing indexation clauses are not designed to handle the volatility of LLM costs. In this article, we explore routes how to properly address this in a contract.

Download full article

Questions about this topic?

Discuss your AI costs & contracting challenges with one of our experts.

contact
Frédérique Joos Partner
Isabelle Dierckx Counsel
Maximiliaan Malbrain Senior Associate

How we add value to your business

Futureproofing your contracts. Drafting robust pricing clauses, threshold triggers, and fair-use provisions that actually keep pace with LLM pricing volatility.

Protecting your margins. Identifying which contractual levers you have when token costs spike and your current terms offer no room to adjust.

Surviving long-term contracts. Exploring your options under Belgian law when margins erode and the client refuses to renegotiate.

Onboarding enterprise clients. Structuring price revision clauses that work for procurement-heavy counterparties.

AI cost pass-through in client contracts. Building transparent provisions that let you recover tool and inference costs without it feeling like a hidden surcharge.

Scaling without scaling risk. Reviewing your terms before a product launch or a new enterprise deal so usage growth does not turn into a liability.

  • This field is for validation purposes and should be left unchanged.