Bottom Line First

The diagnosis is being raised that the big tech-led AI agent and image generation boom is not backed by real value creation relative to its technical novelty. This report sees the current market as a "short-term bubble driven by over-investment" on four grounds.

1. Developer-specific agents are confined to a small group of professionals and struggle to expand into a mass subscription economy.
2. Image generation and everyday features stop at the novelty effect, and retention falls off sharply.
3. Enterprise adoption has a low ratio of proven ROI against CapEx, with a large share of FOMO-driven spending.
4. There is still no killer app equivalent to delivery, commerce, or fintech in the early smartphone market.

This piece organizes the argument item by item and adds counterarguments, scenarios, and recommendations.

Data Verification Warning (Required Reading)

This document is a revised version of a report draft generated by an AI agent.

• Every figure in the body (MAU/WAU, retention rates, ROI ratios, investment scale, benchmarks, and so on) is a value presented by the source and was not independently verified in the environment that wrote this document.
• Some company cases, benchmarks, and sources may be factually wrong or of uncertain basis.
• Before citing any figure, always check the primary source's original text directly.
• This document is not investment advice and should not be used on its own as a basis for market judgment.

1. Overview

2. The Demographic Limits of Developer-Specific Agents

2.1 Data Sources Presented by the Source

The table above is presented by the source and unverified.

2.2 Claude Code's Actual Utilization

The gist presented by the source is as follows.

• Claude total MAU about 220 million
• CLI-based developer-oriented Claude Code WAU about 4.2 million
• Converted to real utilization, that is a minimum of 1.9% to a maximum of 4.2%

In other words, the share of all users who actually use autonomous coding agents is in the single digits.

2.3 Comparison by Tier (Source-Presented Values, Unverified)

These benchmark figures are presented by the source and not externally verified. In particular, the parameter and price notations may differ from actual public information.

2.4 Interpretation and Counterarguments

The source's interpretation is that "there is a large gap between technical excellence and market acceptance." CLI-based coding agents are hard for the general public to access because of environment setup, the burden of code verification, the cost of failure, and workflow integration.

Counterarguments:

• Developers are an early market with high willingness to pay, and as a narrow but solid B2B/B2D market, profitability may actually be higher.
• Coding has clear reward signals such as passing tests and successful builds, so automation can take hold quickly.
• A small number of users does not mean low value. If the per-seat price is high, economies of scale hold.

3. The Retention Collapse of Image Generation and Everyday Agents

3.1 Traffic Composition (Source-Presented Values, Unverified)

3.2 Retention Curve (Source-Presented Values, Unverified)

3.3 Interpretation and Counterarguments

The source sees the motivation to sign up as focused on technical novelty, so when the novelty fades the reason to use it disappears. Image generation in particular leaves the question of where to use the output (the output-to-action gap). Without reuse or a repeating workflow, it does not lead to payment.

Counterarguments:

• In marketing, e-commerce, and content production, there are cases where image generation has taken hold in the actual workflow.
• Retention is a variable that can be improved by product design. Templates, asset management, and collaboration features create a reason to return.
• High early churn may mean not that market validation has failed but that the product is at the stage of finding product-market fit (PMF).

4. The Enterprise CapEx-to-ROI Imbalance

4.1 McKinsey 2026 Survey (Source-Presented Values, Unverified)

The source states it surveyed about 1,500 corporate executives and IT decision-makers worldwide.

4.2 Corporate Cases (Source-Presented Values, Unverified)

These case figures are presented by the source and not confirmed in public materials. They are likely factually wrong and must not be cited.

4.3 Interpretation and Counterarguments

The source sees the supply-side bubble forcibly pulling the demand side. Fear of falling behind and marketing pressure push up data center and license spending, and companies that cannot prove the effect keep spending anyway.

Counterarguments:

• ROI is measured with a lag. Productivity gains show up over years, through organizational redesign, training, and process change.
• EBIT improvement is not the only ROI. Indirect effects such as quality improvement, risk reduction, and shorter cycle time are hard to quantify.
• Survey responses are self-reported, so there is under- and over-reporting bias.

5. Structural Causes of the Market Limits

5.1 Supply-Side Over-Investment (Source-Presented Values, Unverified)

The source's concern is that "demand (real users) cannot keep up with supply (infrastructure investment)." The figures above are presented by the source and unverified.

5.2 Demand-Side Barriers to Entry

6. Comparison with the Early Smartphone Market

The source's core points are two. First, the absence of a killer app. Second, the absence of hardware dependency. Smartphones could use hardware performance limits as a reason to upgrade, but AI agents run in the cloud, so it is hard to create the same reason.

7. Scenario Outlook (Next 6 Months to 1 Year)

The probabilities presented by the source are unverified subjective estimates.

7.1 Optimistic (Bull Case, source 20%)

• New use cases (personal assistant, real-time translation, etc.) raise mass acceptance.
• Low-cost models under $5-10 a month or open-source model enterprise adoption spreads.
• The market grows more than 2x its current size.

7.2 Neutral (Base Case, source 50%)

• Developer and professional creator tools hold, but the mass market stalls.
• Retention converges at a low level.
• Investment efficiency falls and some M&A restructuring occurs.

7.3 Pessimistic (Bear Case, source 30%)

• Tighter regulation, data privacy problems, and exposed performance limits shrink the market.
• A large security incident or copyright lawsuit occurs.
• 40-50% of the market contracts and some companies go bankrupt or are sold.

8. Overall Assessment and Recommendations

8.1 Overall Assessment

The source's core conclusion is as follows.

The current AI agent market is similar to the early smartphone market's pre-killer-app stage, and features at the level of simple summarization, email drafts, calendar entries, and one-off image generation do not provide an incentive to pay a fixed monthly subscription of $20 or more over the long term.

The grounds the source cites (all unverified figures):

1. Low real utilization: developer-specific tool use is 1.9%-4.2% of active users
2. Retention collapse: 58% churn within a week
3. ROI imbalance: 39% can demonstrate EBIT improvement
4. Return on investment: a low revenue recovery rate against about $335B in total investment

8.2 Recommendations

9. References (Source-Cited, Verification Required)

The sources the source cites are as follows. Their actual existence and whether the figures match were not confirmed.

1. Anthropic Economic Index Report (2026 Q3)
2. GitHub Copilot Usage Report (2025 Q4 - 2026 Q3)
3. McKinsey Global AI Survey 2026 (1,500 corporate executives)
4. Cursor AI User Survey (2026 Q1 - Q3)
5. NVIDIA GTC Keynote & Financial Report (2025)
6. EU AI Act Regulatory Framework (Draft, 2026)

Notes

• Every figure — accuracy, speed, retention, ROI — is a value presented by the source and varies with the site, agent, survey timing, and sample definition. This document does not generalize them.
• This document is a compilation of a market analysis report draft and is not investment advice.
• Whether it is a bubble and whether the valuation is justified are confirmed only in hindsight.
• A balanced judgment requires considering the counterarguments together (developer market profitability, delayed ROI effects, room to improve retention).

Source author: Hermes Agent / Qwen3.8-9b-distill. Revised by: deepseek-flash.