AI cost surprises force board escalations and budget freezes

A new report from Mavvrik and Benchmarkit finds enterprises are still struggling to govern AI spending, with surprise costs forcing board escalations, spending freezes, and cancelled projects in 2026. The study also shows most companies cannot fully attribute AI costs, making ROI calculations and margin management unreliable.

Why it matters: - AI spending is now affecting core business decisions, not just IT budgets. - 62% of organizations said an unexpected AI cost materially changed a business decision in the past year. - 40% escalated the issue to the board, 33% imposed emergency spending freezes, and 25% delayed or cancelled an AI initiative. - AI costs have eroded gross margins at four in five enterprises for a second year, according to the report. - 81% of organizations still cannot fully account for AI costs.

What happened: - Mavvrik and Benchmarkit released the 2026 State of AI Cost Governance Report on July 29, 2026. - The study examined how enterprises forecast, attribute, govern and measure AI spending across cloud infrastructure, GPUs, large language models, autonomous agents, developer tools and private infrastructure. - The report is based on survey results from 396 enterprise organizations across Technology/SaaS, Financial Services, Retail, Manufacturing and other industries in April and May 2026.

The details: - Only 11% of organizations can forecast AI costs within ±10%, down from 15% in 2025. - 89% miss forecasts by more than 10%. - Self-assessed maturity scores showed little correlation with actual forecasting performance. - 68% of organizations now operate hybrid AI environments that span cloud and on-prem infrastructure, up from 61% in 2025. - Only 44% include on-prem AI infrastructure in cost reporting, up from 35% in 2025. - 15% of organizations running agentic workloads cannot attribute agent-related costs at any level. - 98% of engineering organizations use AI coding assistants. - Companies use an average of 2.4 AI coding tools at the same time. - 39% said coding tool costs exceeded expected license or usage levels. - Only 42% include developer AI tools in AI cost reporting. - Data platform overages are now the most frequently cited source of unexpected AI spending. - Costs tied to data movement, storage and processing often exceed expectations and remain largely invisible to finance teams. - 49% of organizations were forced to reprice AI-powered products because of unforeseen cost increases. - AI-native application companies posted the worst forecasting performance of any segment. - Just 3% of AI-native application companies forecast AI costs within ±10%, despite more than half describing themselves as advanced in AI governance.

Between the lines: - The report suggests AI ROI discussions are often built on incomplete cost data. - Productivity gains and faster development cycles are easy to measure, while developer tools, data platforms, agentic workloads and private infrastructure often remain outside finance reporting. - That gap helps explain why board escalations and emergency freezes are showing up alongside broader AI adoption. - Sundeep Goel, CEO of Mavvrik, said organizations succeeding with AI are not necessarily spending less; they are building governance systems that show where costs originate, how they scale and how they affect margins. - Ray Rike, CEO of Benchmarkit, said ROI cannot be calculated accurately without complete cost visibility.

What's next: - The full report says it covers forecasting accuracy, attribution maturity, hybrid infrastructure visibility, agentic AI governance, industry benchmarks and recommendations for building sustainable AI financial operations. - Mavvrik says enterprises need better visibility, attribution and control across AI, autonomous agents, GPUs, cloud, on-prem and SaaS environments, including systems with no billing API. - The report points to a continued push toward stronger AI financial governance as companies try to rein in bill shock and protect margins.

The bottom line: - Enterprises have embraced AI faster than they have built the financial controls to manage it, and the result is budget volatility, weaker margin visibility and more frequent executive intervention.