Software pricing has historically relied on predictable seat licenses and consumption meters, giving enterprise buyers an illusion of control over IT spend. That era is cracking. As enterprise LLM tools remain brutally expensive to run while failing to accelerate top-line revenue, corporate CFOs are refusing to foot the bill for compute cycles spent on hallucinations and dead-end agentic loops. Under intense pressure from tightening IT budgets, major AI vendors are ditching token metrics for outcome-based contracts.
Paying for Completed Work
OpenAI has quietly rolled out outcome-based pricing for select enterprise accounts, according to reporting from The Information citing sources with direct knowledge of the deals. Under these agreements, corporate clients pay only when an underlying model successfully resolves a defined task, such as an end-to-end customer support ticket. While an OpenAI spokesperson declined to comment, the pivot signals a broader reckoning across enterprise software.
Startups forced this transition long before the tech giants caught up. Customer service platforms Sierra and Fin pioneered billing enterprise clients exclusively for tickets resolved without human intervention—a model compelling enough that Salesforce moved to acquire Fin for $3.6 billion. Meanwhile, coding assistant maker Cognition has offered enterprise clients credits worth up to $10 million if its platform fails to deliver value matching the contract price. Heavyweights like Adobe, HubSpot, and Zendesk are now scrambling toward similar performance-linked structures.
The Economics of Agentic Automation
Inference costs remain high, and enterprise buyers are pushing back against subsidizing vendor R&D via raw token consumption. With models like Anthropic's Claude intensifying competition, enterprise leaders are demanding direct financial justification for every automated workflow.
"The point isn't just to say the software closed a certain number of calls and charge two dollars for them. It's to say the software raised revenue by a certain amount, so the customer should pay two dollars because the AI brought in 20 or 40."
As Marc Benioff, CEO of Salesforce, made clear when pitching Agentforce to Wall Street, enterprise contracts are pivoting directly to customer cost reductions and top-line expansion. Corporate leaders can now negotiate bespoke agreements tied to hard financial gains, permanently shifting software procurement away from compute volume toward verified commercial returns.
The Problem of Success Attribution
Tying enterprise billing to business performance introduces a fierce operational battleground: baseline attribution. When a vendor charges against incremental revenue or resolved volumes, proving whether an autonomous algorithm or your in-house team closed the gap is messy. Payment processor Stripe highlighted this friction in its attribution guidance, noting that operational improvements and revenue surges often trace back to product updates, marketing pushes, or seasonal tailwinds rather than third-party AI.
For enterprise leadership, the immediate priority is renegotiating incoming SaaS contracts. Demanding ironclad outcome definitions, baseline metric calculations, and neutral third-party attribution clauses is no longer optional—it is the only way to stop vendors from claiming unearned credit for routine business growth.