AI Spend is Outrunning Procurement. Here’s What Sourcing Leaders Are Actually Doing About It.

· Source: NPI · Field: Business & Management — Corporate Strategy & Leadership, Operations & Process Management, Consulting & Professional Services · Depth: Intermediate, short

Summary

IT sourcing leaders in large enterprises face significant challenges managing AI spend, with 56% citing unpredictable usage and scaling costs due to the shift from per-seat to consumption- and token-based pricing models. Traditional procurement playbooks are ill-equipped for this new "tokenomics," where vendors define credits inconsistently and bundle AI features. While 50% of large enterprises have an AI Center of Excellence, its mere existence doesn't guarantee a coherent strategy. Key cost pressures for the next 12 months include token/API usage growth (27%), new vendor pricing (23%), increased employee AI tool adoption (22%), and expanding AI into production (19%). Emerging best practices involve pilot-first, shorter-term contracts, negotiating "super caps" and price-change clauses, accounting for expensive forward-deployed engineers, addressing data training legal exposure, and framing AI ROI as a portfolio-level conversation.

Key takeaway

For IT sourcing leaders managing escalating AI costs, your traditional per-seat procurement models are obsolete for AI's consumption-based pricing. You must proactively implement internal budget allocation for tokens and negotiate specific contractual protections. Prioritize shorter contract terms (e.g., one-year renewals) and demand clauses for pricing changes and consumption caps to maintain flexibility and control. Additionally, ensure forward-deployed engineers' costs are accurately modeled into your total cost of ownership, as they represent a significant, often hidden, expense.

Key insights

AI's consumption-based pricing creates unpredictable spend, necessitating new procurement strategies and internal cost governance.

Principles

Method

Treat tokens and credits as internal budgets, allocating them to business units and requiring requests for more to enforce cost discipline.

In practice

Topics

Best for: CTO, Executive, Consultant, Director of AI/ML, VP of Engineering/Data

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NPI.