Companies are boosting productivity by adopting artificial intelligence, but an analysis found hidden costs in actual operations are far larger than expected. A survey result also showed as much as 82 percent of total engineering costs was consumed by fixing and rewriting bugs in AI-generated code and by delays in review.

On May 28, blockchain media outlet BeInCrypto reported that contrary to expectations that AI adoption would raise development efficiency, operational burdens and organisational restructuring costs are rising quickly at the service deployment stage.

A survey by Entelligence AI of 2,444 companies found firms spent an average of $0.44 on bug fixes for every $1 spent on AI token use. Another $0.27 went to rewriting AI-generated code, and $0.11 was incurred during delays in code review and merging. That means 82 percent of total costs went to humans reworking AI outputs.

Distrust of AI code quality also persisted. According to Lightrun's "2026 State of AI-Powered Engineering" report, 43 percent of AI-generated code that passed quality checks still required additional manual debugging in real operating environments. None of the engineering leaders surveyed said they were fully confident in deployment outputs. The outlet said Coinbase's AI adoption case and Cardano's strategy of separating AI code also showed a similar trend.

Financial burdens from expanding AI infrastructure are also growing. Oracle has increased total debt to about $108 billion to expand AI data centres. It also secured an additional $50 billion in 2026 through debt and equity financing. Free cash flow, however, was tallied at about negative $13 billion.

Oracle's backlog of orders of $553 billion, with more than $300 billion tied up in OpenAI-related contracts, was also cited as a burden factor. OpenAI is known as a customer that posted a loss of about $14 billion last year. The market is watching whether a strategy of large-scale preemptive investment aimed at growing AI demand can translate into actual profitability.

The mood is also shifting toward faster changes in organisational operating methods. Star Xu (쉬) OKX's chief executive said the spread of AI agents is changing the workforce evaluation system itself beyond simple automation. He said talent requirements are changing in the AI era and explained that AI utilisation capability has begun to be directly reflected in employee evaluation criteria.

The industry has produced analyses that AI is reshaping hiring and organisational management standards beyond being a simple work-assistance tool. Major big tech earnings announcements and engineering indicators scheduled for June are expected to be a key test of how much the gap between expanded AI investment and actual profitability has been narrowed.