CIOs and CTOs spent years lauding AI. Now, with costs rising, they’re putting limits on how it’s used
Chief Information Officer Stephen Franchetti has enthusiastically embraced a wide array of artificial intelligence tools to bolster workflow automation for the tech firm’s 4,100 employees.
At tech firm Samsara, Franchetti has authorized Anthropic’s Claude, Google’s Gemini, OpenAI’s ChatGPT, and the AI coding agent Cursor, while also developing an internal system that monitors all AI expenses that can be tracked on a daily basis. More recently, Samsara has capped usage for some non-technical employees, while some groups like research and development—which need AI for more intensive coding and data analysis—have more room to experiment.
“It took us a while to settle on the right caps, to make sure everyone was well served,” says Franchetti. “But it puts people in the position where they’re kind of in control and they can make choices as to which models they use.”
With global AI spending projected to total $2.5 trillion this year, a 44% increase from prior-year levels, CIOs and other technology leaders within the C-suite are finding themselves in a delicate moment on their AI journey. After years of promoting AI adoption with training courses and hackathons, and making AI coding tools, agents, and chat assistants widely accessible for their employees, some are tightening up how frequently these tools can be used and retraining staff to better understand smaller and cheaper AI models that can handle many workplace tasks.
Some companies have reported that their 2026 AI budgets have blown past what they expected at the beginning of the year, without producing any corresponding value to the business. AI hyperscalers are hearing the complaints too, and have responded by rolling out cheaper models or have cut prices.
“2026 is the year of everyone finding out that AI is actually really hard,” says Will Sommer, a quantitative modeling and economic forecasting expert at research firm Gartner. “It’s not a free lunch. It requires a lot of thought and effort to get right.”
Sommer warns that companies can easily spend thousands of dollars per head on AI tools whose output is essentially junk and doesn’t bolster productivity. In June, Gartner also issued a bearish report that warned AI coding costs would overtake the average developer’s salary by 2028, due to rising token consumption and a shift to consumption-based fees.
“We’ve taken this very seriously, both in the internal use case and we are propagating those learnings externally,” says Sagnik Nandy, the chief technology officer at electronic-signature software company Docusign.
Internally, Nandy lauded that every engineer has embraced AI tools and that 75% of the code that they develop is initiated by AI. But, Nandy says he quickly realized that AI code agents were built to tap Docusign’s entire code base for context before executing a task. “That’s a lot of tokens, because you’re trying to read everything,” he says.
Nandy has adjusted these AI coding agents so that the default setting is that they only pull the context that’s relevant to a narrow task a developer is working on, which has reduced token usage by almost 50%.
Jim Dausch, the chief digital and technology officer at Yum Brands, says AI token usage isn’t yet “a material number, but the trajectory was one that we’re watching.” Earlier this year, he did notice that both AI token usage and expenses were rising at the KFC and Taco Bell restaurant operator.
But, Dausch contends that a vast majority of tasks that AI tools are asked to perform—perhaps even as high as 95%—can be handled by more basic, less expensive models. That has led Yum to promote more training on AI model usage and advise business leaders to closely manage their digital spending the same way they budget for a department’s headcount.
“We’re trying to kind of democratize where the costs live and how they’re managed, so it isn’t just an IT line item,” says Dausch.
At Cigna Group, the healthcare giant casts a wide net and has embraced multiple AI hyperscaler vendors to allow the company’s workforce to utilize small language models or earlier, cheaper versions for some tasks that don’t require a lot of reasoning capabilities. “The way you really run up costs is you use the most expensive models with no guardrails around them,” says Katya Andresen, Cigna’s chief data, digital, and AI officer.
Cigna has authorized more than 70 different AI models for internal purposes, and while Andresen says compute and AI token usage has increased, total spending isn’t rising at the same pace, due to this multimodal approach.
Shay Artzi, CTO at real-estate brokerage Compass, says his AI investments have focused on three worker groups: engineers, the company’s AI Assistant tool that automates tasks for real estate professionals, and the general corporate population. Engineers were the first to use AI expansively, but Artzi says he was cautious about AI token spending early on and didn’t mandate that code always needed to be written with AI.
Compass piloted several AI coding tools and settled on partnerships with two giants in the space, Anthropic and Google, with some financial-focused limitations. “We also put budgets for every engineer, so they are aware of how they’re spending,” says Artzi.
John Kell