[AI Bubble] “Few Paying Subscribers and Weak Retention” Concerns Over an AI Bubble Spread as Companies Seek Breakthroughs While Maintaining Optimism

AI apps show strong early monetization but struggle with long-term retention
Absolute number of paying subscribers remains limited, profitability under pressure
AI firms begin monetization push in markets such as India while maintaining optimistic outlooks

Skepticism is growing in capital markets about the real profitability of the artificial intelligence (AI) industry. The number of paying subscribers for AI-powered application services is showing sluggish growth, while user churn is becoming increasingly pronounced. AI companies, however, continue to maintain an optimistic outlook as they search for new avenues of growth, including expanding sales of paid subscription products in large markets such as India.

AI Apps Confront Retention Barrier

According to the “2026 Subscription App Ecosystem Report” released on the 11th (local time) by subscription management platform RevenueCat, AI apps demonstrate clear advantages over non-AI apps in early-stage monetization. The paid conversion rate of AI apps was 52% higher than that of non-AI apps, while their revenue-generation capability per download was also 20% stronger. Monthly average user value (RLTV) for AI apps reached $18.92, compared with $13.59 for non-AI apps, representing a 39% difference. On an annual basis as well, AI apps maintained profitability roughly 41% higher.

The problem is that such momentum is proving difficult to sustain over time. Over the past year, the annual retention rate of AI-based apps stood at 21.1%, 9.6 percentage points lower than that of non-AI apps at 30.7%. Monthly retention also lagged, with AI apps posting a rate of 6.1%, compared with 9.5% for non-AI apps. Retention refers to the proportion of existing customers or users that a company, application, or service manages to retain without losing them over a given period.

Churn among paying subscribers is also accelerating. RevenueCat analyzed that the annual subscription cancellation rate for AI app subscribers is about 30% faster than that of non-AI apps. Refund rates were also approximately 20% higher for AI apps, with some services recording refund rates as high as 15.6%. RevenueCat assessed that “AI apps tend to have strong conversion rates from free trials to paid subscriptions, but long-term user value and user experience show considerable volatility,” adding that “revenue volatility is high and companies may struggle to sustain long-term service quality.”

Limited Scale of Paying Users

The absolute scale of paying users itself remains stuck in a stagnation phase. According to analysis by agent integration firm AI Venture Studio, roughly 1.3 billion people—about 16% of the global population—use free chatbots out of simple curiosity. By contrast, the number of paying users willing to spend $20 or more per month amounts to only 15 million to 25 million people worldwide, equivalent to about 0.3% of the global population. This estimate combines the number of ChatGPT Plus subscribers—about 20 million—announced by OpenAI late last year with estimated paid subscriber figures from competing services.

Similar conclusions have emerged from other analyses. According to the “State of Consumer AI” report published last month by Silicon Valley venture capital firm Menlo Ventures, the number of global AI service users stands at roughly 1.8 billion, with about 600 million using AI services daily. The estimate is based on a survey conducted by Menlo Ventures involving 5,000 adults. However, assuming a monthly subscription fee of $20 for AI services, only about 3% of users are paying subscribers.

As a result, concerns over the profitability of AI companies are rapidly spreading in the market. Despite the enormous scale of capital flowing into the sector, revenue foundations remain fragile. Several experts share similar concerns. For example, Joseph Stiglitz, a Nobel Prize–winning economist and professor at Columbia University, warned in an interview with U.S. business magazine Fortune on the 8th that the current surge of investment in AI could turn into a bubble.

He noted that the AI boom is delivering positive short-term effects on the macroeconomy but argued that expectations for the revenue AI can generate are excessive. Stiglitz stated, “Market expectations that AI companies can earn high profits are based on the assumption that competition will be limited, but in reality competition in the AI sector is intense, with not only large U.S. technology companies but also Chinese firms participating.” He added, “Even if AI succeeds technologically, intensified competition could drive profits down to zero, preventing companies from achieving the profits investors expect.” He also warned that if such a bubble collapses, it could inflict severe short-term damage on the macroeconomy.

AI Companies Seek New Growth Paths

Industry participants, however, maintain an optimistic outlook that the number of paying subscribers will continue to grow. OpenAI, for instance, expects ChatGPT’s weekly active users (WAU), currently around 800 million to 900 million, to rise to 2.6 billion by 2030. Of these users, the company projects that 8.5%, or about 220 million people, will subscribe to the Plus plan. The strategy involves dramatically expanding the total user base while simultaneously increasing the share of paying subscribers. The current subscription fee for the Plus plan is $20 per month. If 220 million users subscribe to the plan, annual revenue would reach $52.8 billion.

Cases are also emerging in which AI companies are beginning to generate paid subscription revenue in regions that had previously relied on free promotions. India is a representative example. According to market research firm Sensor Tower, India was the country that downloaded the most generative AI apps in the world last year. App installations surged 207% year-on-year, surpassing the United States, where many major AI companies are headquartered. In response, AI companies such as OpenAI, Google, and Perplexity have been expanding free premium services to secure demand in the Indian market.

However, this wave of aggressive early promotions is gradually drawing to a close. Perplexity ended its bundled “Pro” service offered in partnership with Indian telecom operator Airtel in January, while OpenAI halted free access to “ChatGPT Go” in India. The companies have effectively begun a full-scale monetization push in one of the world’s largest markets.

Whether these monetization strategies will succeed, however, remains uncertain. India’s AI user base tends to prioritize value for money and adopt a cautious approach to spending. Sensor Tower data shows that in-app purchase revenue from AI apps in India accounts for only about 1% of global totals. Some observers therefore expect that AI companies will eventually introduce localized subscription strategies tailored to the market’s characteristics, such as low-priced plans and telecom-bundled subscription products.