1 in 4 companies delay or cancel an AI project over cost

Dive Brief:

• Artificial intelligence spending is becoming increasingly difficult for companies to forecast and manage, with unexpected costs forcing some organizations to delay, freeze or abandon initiatives, AI cost management firm Mavvrik found in a recent report.
• Sixty-two percent of organizations surveyed said an unexpected AI cost materially altered a business decision over the past year. Among those organizations, 40% required board-level escalation, 33% implemented emergency spending freezes and 25% delayed or canceled an AI initiative, according to Mavvrik’s 2026 State of AI Cost Governance Report released last month in partnership with Benchmarkit, a software-as-a-service performance metrics firm.
• “Companies are seeing significant AI cost overruns,” Sundeep Goel, CEO of Mavvrik, said in an interview. 

Dive Insight:

The findings come as AI costs are increasingly tied to the consumption of “tokens,” units of data processed by AI models that can quickly add up as workloads grow.

Mavvrik found 43% of companies cited token costs as a top source of unexpected Al spending.

In a July report, Gartner said token consumption is becoming a “meaningful component” of enterprise operating costs, while its connection to business outcomes often remains unclear.

The research firm said CFOs need greater visibility into how tokens are used, what value they generate and whether consumption is economically justified. Gartner recommends shifting from simply tracking AI costs toward “AI unit economics” that link token consumption to measurable outputs and outcomes.

The goal is not necessarily to minimize token spending but to “rightsize” it based on the value of the outcomes delivered.

A separate Gartner report projects that AI coding costs in particular will exceed the average developer salary by 2028, driven by rising large language model token consumption and the shift to consumption-based licensing.

Meanwhile, volatility in AI pricing is complicating cost estimates, according to Mavvrik’s study. Only 11% of organizations said they can forecast AI spending within plus or minus 10%, down from 15% in 2025.

Goel said the deterioration reflects a lack of discipline around AI spending that is coming into sharper focus as costs continue to rise. “It’s sort of perfect storm of wasted money,” he said.

Further complicating matters, “shadow IT” — employees adopting technology without their company’s knowledge or approval — can allow AI spending to proliferate outside traditional IT controls.

The Mavvrik report found that 98% of engineering organizations use AI coding assistants, yet only 42% include spending on developer AI tools in their AI cost reporting, meaning that many organizations are not fully accounting for these costs.

Goel compared the phenomenon to the early days of cloud computing, when development teams could sign up for services using corporate credit cards without centralized oversight. “We’re seeing the equivalent here,” he said.

That can leave companies paying for overlapping tools or unused licenses, making AI spending harder to attribute and control.

“There’s just no way companies can run a budget if there’s absolutely no controls on who can spend money on what,” Goel said.