Dive Brief:
- Global end-user spending on AI models and platforms will jump 63% from last year, reaching $64 billion, according to a Gartner report released Monday.
- CIOs are more closely scrutinizing AI budgets as costs rise, the report said, with a focus on usage efficiency, cost control and measurable outcomes. Enterprise customers will likely gravitate toward providers that can demonstrate value across cost, latency, performance and reliability.
- Because of agentic AI, a majority of the model spend is being embedded into platforms, Arunasree Cheparthi, senior principal research analyst at Gartner told CIO Dive. “Model vendors are turning into intelligence providers to the platforms. Now the battle is between who captures the entire enterprise workflow end to end,” she said.
Dive Insight:
As AI use matures, and enterprises decide what pilots to keep backing, vendors that help companies manage where and how AI is used across the business will see more success.
AI spending will only continue to rise year over year, Cheparthi said, but CIOs and other tech decision-makers have access to some strategies to keep spending in check. A CIO can review vendor contracts, their organization’s AI architecture and governance to keep an eye on costs, she said.
Contract-level protections that enterprise tech leaders can pursue include locking in token pricing that ties input to output ratios or building in hard consumption limits within their AI workflows. In practice, that means leadership can build in consumption limits that would trigger automatic throttling, or require approval to override, once a limit has been reached.
“Wherever possible, enterprises should shift to outcome-based or value-based pricing,” Cheparthi said. “Traditional consumption models create unpredictability and often misalign with actual business value.”
Contract negotiations can also stipulate that unused tokens within a certain period can roll over, a strategy that mitigates waste, Cheparthi said.
As the CIO role continues to evolve, they have become the orchestrators and architects of their organization’s AI strategy and use. That architecture, or how an organization executes its AI strategy, can play an important role in cost savings, Gartner’s report said.
Enterprises may implement multivendor routing which deploys the most cost-effective model for each task, using lower-cost or open-source LLMs for routine requests and saving premium subscription models for more complex work.
Enterprises that track their cost per token and set reduction targets around redundant prompts or overconsumption will see some of the higher returns on their investments, Cheparthi said.
Token usage visibility should become a standard operational metric tracked across workflows, she added.
“Enterprises should enforce strict consumption limits with segmented budgets, establishing hard financial boundaries per department or application,” Cheparthi said. “Automating access throttling or escalation protocols should be in place once limits are reached, like ensuring disciplined usage from the start.”
An enterprise’s governance strategy is an often overlooked opportunity for cost savings. Many vendors embed AI premiums in bundles, which inflates renewable costs. Companies should build contract review teams that have AI expertise into their governance processes.
“This prevents capital leakage and forces rigorous outcome measurement, and also aligns departmental performance goals to AI initiatives delivering measurable value,” Cheparthi said.
