Enterprise IT and AI Vendors Clash Over Token Pricing as ROI-Based Deals Prove Elusive in 2026
Updated
Updated · Computerworld · Jun 2
Enterprise IT and AI Vendors Clash Over Token Pricing as ROI-Based Deals Prove Elusive in 2026
7 articles · Updated · Computerworld · Jun 2
Enterprise buyers still cannot settle on AI pricing because they want contracts tied to measurable business returns, while vendors prefer token, usage or task-based fees they can meter and forecast.
SAP executive Irfan Khan said the sequencing is the core problem: pricing must be locked in before agentic AI projects start, even though day-one costs and day-one value rarely line up.
Consultant Justin Greis argued AI is being forced into infrastructure-style pricing even though it behaves more like labor augmentation, leaving CFOs unmoved by token counts if process gains never appear.
Outcome-based pricing is not a clean fix either, because vendors cannot control adoption, data quality or change management, and paying AI mainly for results could push autonomous systems to game metrics.
The report says companies may need AI review committees, quantified upside and downside scenarios, and executive bonus exposure before CIOs can negotiate pricing that both sides will accept.
With AI costs spiraling, are enterprises being forced to choose between innovation and budget predictability?
Is outcome-based AI pricing a true partnership or a dangerous incentive for machines to game the system?
Enterprise AI Economics 2026: Surging Costs, Elusive ROI, and the Battle for Value
Overview
In 2026, U.S. companies face a surprising AI cost crisis: even as the price per AI token drops, total spending on AI is soaring. CFOs are forced to choose between investing in AI or human workers, as annual AI budgets are quickly depleted. This is mainly due to businesses adopting AI without clear strategies, often using a 'thousand flowers bloom' approach that spreads licenses widely but fails to deliver real value. At the same time, employees struggle to keep up with rapid AI changes, and cheaper AI access does not guarantee cost-effective results, leading to financial strain and tough decisions for enterprises.