AI Agent Costs Vary Widely, Challenging Enterprise Budget Planning
AI agents from leading providers consume widely varying amounts of computational resources to complete similar tasks, with no transparency into costs and no guarantees of successful outcomes, according to an investigation by ZDNet.
The findings highlight a growing challenge for U.S. enterprise buyers seeking to deploy agentic AI workflows: the inability to predict what they will actually pay.
Testing of leading AI agents revealed that token consumption — the primary cost driver for most AI services — varied widely across agents performing comparable tasks. The lack of standardized pricing or consumption benchmarks means businesses cannot reliably forecast budgets for AI agent deployments, according to the report.
No Transparency, No Guarantees
Unlike traditional software subscriptions or even conventional API-based AI services, where costs scale relatively predictably with usage, agentic AI introduces a layer of unpredictability. Agents operate autonomously, making chains of decisions that can consume tokens at rates that differ not only between providers but between individual runs of the same task.
The ZDNet investigation found that providers offer little visibility into how tokens are consumed during agent operations, making it difficult for procurement teams and IT departments to evaluate competing offerings on a cost basis.
Adding to the uncertainty, none of the agents tested provided guarantees that tasks would be completed successfully, meaning enterprises could incur costs with no assurance of useful output.
Implications for Enterprise Adoption
The findings come during a pivotal period for the AI agent market. Major U.S. technology companies — including OpenAI, Anthropic, Google, Microsoft, and a growing number of startups — are moving to bring agentic AI products to enterprise customers. Gartner and other analyst firms have identified AI agents as a leading enterprise technology trend for 2026.
Unpredictable pricing could slow adoption among cost-conscious enterprise buyers, particularly in regulated industries where budget certainty is a procurement requirement. The lack of pricing transparency also complicates return-on-investment calculations that executives need to justify AI spending.
