Survey: 82% of Enterprise AI Spend Lost to Code Problems

Entelligence AI surveyed 2,444 companies in mid-2026 and found as much as 82% of enterprise AI spending is absorbed by bug fixes, code rewrites and review and merge delays before code reaches production.

The report breaks down losses per $1 spent on AI tokens: $0.44 goes to bug fixes, $0.27 to rewrites of AI-generated code and $0.11 to review and merge delays.

Lightrun’s 2026 State of AI-Powered Engineering Report found 43% of AI-generated code still requires manual debugging in production after passing initial quality checks. That report also found no engineering leader surveyed expressed full confidence in deployed AI output. Companies including Coinbase and projects in the Cardano ecosystem have reported similar patterns of code splitting and additional human review.

Oracle reported roughly $108 billion in total debt and raised about $50 billion in 2026 through debt and equity to fund data center buildouts for AI workloads. The company reported free cash flow near negative $13 billion. More than $300 billion of Oracle’s $553 billion backlog is linked to a single client, OpenAI, which reported a roughly $14 billion loss last year. Oracle’s next earnings report is scheduled for June 16, 2026.

OKX has changed how it evaluates staff, tying employee assessments to AI proficiency. OKX CEO Stax Xu argued that AI agents speed execution while revealing employees who rely on impression management rather than measurable outcomes. He added, “It’s not AI that fundamentally changes layoffs. It’s that the AI era fundamentally changes talent requirements.” Several trading platforms and exchanges have set similar AI competency requirements for staff.

The survey findings, industry reports and company disclosures were published in mid-2026.