The 'Invisible Costs' of AI: Server Expenses Soar Behind AI Usage — Over 40% of Companies Unaware. 80% of Self-Diagnoses Fail to Capture These Costs.

LEI, inc. (Headquarters: Shibuya-ku, Tokyo), provider of the cloud cost reduction service 'CloudCut,' has conducted and published results from its 'Cloud Cost Reduction Audit,' which examines corporate AWS usage. Among 32 companies that received free audits in June 2026, approximately 80% could not separately track cloud spending related to AI. Public data also reports that over 40% of companies fail to accurately monitor AI-related cloud expenses (CloudZero 2025). The more companies use AI, the more backend infrastructure costs—such as servers, data storage, and communication—grow, often exceeding the direct AI usage fees. These expenses are accumulating invisibly, as revealed by on-site audits.

Survey Summary: The 'Invisible' Reality Uncovered by the Audit

LEI, inc. continuously conducts 'Cloud Cost Reduction Audits' for companies using AWS, individually reviewing billing details and usage patterns. Aggregating data from 32 companies audited in June 2026, the following trends were identified:

- Approximately 80% could not separately track cloud spending on AI-related activities (e.g., generative AI, machine learning) from other expenses

(= The backend costs—servers, data storage, communication—that increase alongside AI usage are not itemized on invoices)

As AI adoption increases, so do its supporting infrastructure costs. Yet, companies cannot clearly see 'what they are paying for' on their bills—this is a reality we repeatedly observe in the field.

Why Does AI Usage Increase 'Invisible Costs'?

AI is not a 'set-and-forget' technology. Each time it is used in operations (inference), it continuously consumes computing resources and data storage. As user inquiries increase, so does the usage of backend servers, storage, and bandwidth. In other words, higher AI utilization directly correlates with increased infrastructure (server) usage.

[Background & Market Trends] Market Data Confirms the Same Trend

This on-site observation aligns with publicly available market data. The following trends are based on external surveys and reports, not LEI, inc.'s audit results.

According to CloudZero's '2025 Cloud Cost Trends,' 63% of companies are increasing cloud budgets for generative AI. The report further notes that running large language models (LLMs) in production environments can cause cloud bills to balloon by 2–3 times due to inference costs, and that inference is 3–5 times more expensive per token than training. Additionally, over 40% of companies fail to accurately track AI-related cloud spending, with some reports indicating that so-called 'shadow AI' can inflate bills by 20–50%.

This growth is not temporary. McKinsey (2024) predicts global AI infrastructure capital expenditure (capex) will reach $5–7 trillion by 2030. Gartner (2025) also notes that IT spending on data centers and edge computing will grow at an annual rate of about 12% through 2027, primarily driven by AI's computational demands. AI growth directly translates into increased demand for supporting infrastructure.

[Background & Market Trends] Japanese Companies Are Still at the 'Entry Stage' of AI Adoption

Meanwhile, Japanese companies' AI adoption lags behind global peers. According to Japan's Ministry of Internal Affairs and Communications' 'Information and Communications White Paper' (Reiwa 6/7 edition), only 15.7% of Japanese companies have a policy to 'actively utilize' generative AI—significantly lower than the U.S. (46.3%). Among SMEs, only 20.4% have implemented AI. Furthermore, MIT's 2025 'State of AI in Business' report states that only about 5% of companies have achieved measurable ROI from generative AI initiatives. Given the vast potential for growth and the impending mainstream adoption, now is the time to make rising infrastructure costs visible.

The Way Forward: Dual Focus on 'AI Adoption' and 'Supporting Infrastructure & Cost Design'

Combining audit findings and market data, four key insights emerge: (1) AI usage increases backend infrastructure costs; (2) these costs are 'invisible'; (3) AI and infrastructure demand will continue to grow; and (4) Japanese companies are still in the early stages, with significant growth potential.

The effective approach is to advance AI adoption and infrastructure/cost design in parallel—not sequentially. There is no need to halt AI implementation. Instead, while promoting AI, simultaneously optimize the rising infrastructure costs without altering existing environments or configurations. Pursuing both in parallel will determine cost-effectiveness in the full-scale impact phase.

Value Proposition of the Cloud Cost Reduction Service 'CloudCut'

CloudCut reduces cloud costs by optimizing the 'commercial flow' of contracts and billing—without modifying environments or configurations.

- Zero configuration changes, zero downtime

— Your current cloud environment and setup remain unchanged. Implementation is achieved simply by switching contract and billing interfaces (※ May vary depending on migration method).

- Up to 20% cost reduction (via combination of multiple mechanisms)

— Combines volume discounts and other mechanisms to reduce costs sustainably.

- Supports Japanese yen and invoice billing

— Use in JPY with invoice payments, eliminating concerns about foreign exchange and overseas transfers.

- No minimum contract period, zero initial cost

— Start without long-term commitments or upfront fees.

▼ CloudCut Service Website https://cloudcut.lei-inc.jp/

※ 'Up to 20% reduction' refers to the maximum reduction rate achieved in past CloudCut implementations. Actual reduction rates vary based on usage scale, contract duration, and current service configuration.

Survey Overview

- Survey Name: CloudCut Cloud Cost Reduction Audit - Survey Period: June 2026 - Survey Target: 32 domestic companies that received LEI, inc.'s free cloud cost reduction audit - Survey Method: Detailed review of billing breakdowns and usage via free audit - Survey Conducted by: LEI, inc. CloudCut Division - ※ 'Survey Summary' reflects aggregated values from LEI, inc.'s audits. Figures cited in 'Background & Market Trends' are based on external public reports (sources cited).

Future Outlook

LEI, inc. will support companies in building an environment where 'AI advancement' and 'infrastructure cost optimization' can progress simultaneously as AI adoption enters its mainstream phase. Leveraging CloudCut's low barrier to entry—no configuration changes, no downtime, no initial cost—we aim to become a foundational platform enabling more companies, including SMEs and startups, to achieve cost optimization by visualizing and compressing the 'invisible infrastructure costs' that grow in the AI era.

About LEI, inc.

Company Name: LEI, inc.