Die Bezahlquote sinkt bis auf 3%. Es ist nicht mehr weit her, bis die AI-Hersteller die Nutzer mit Werbung bombardieren werden.

Recently, OpenAI, the leader in the AI industry, informed its shareholders of the latest financial forecasts. It is estimated that the cash burn will rise to $115 billion over the next four years. Just as OpenAI's management proposed this unprecedented risky course, OpenAI employees who are allowed to sell shares are expected to offload shares worth about $10.3 billion.

All this brings the theory of an AI bubble back to life. Interestingly, during the dot - com bubble around the turn of the millennium, the popular internet startups at that time also proved themselves more through "money - burning" than a stable business model.

The renowned investment firm Menlo Ventures has reignited the theory of an AI bubble. In their latest macro - report on the AI market, these figures were released: The total number of users has reached 1.8 billion, the daily active users are 600 million, but the payment rate is only 3%.

In other words, among the 600 million people who use AI most actively, only 18 million are willing to pay. What does a 3% payment rate mean for internet products? For comparison: The payment rate for video - streaming and music - streaming services in the Chinese market is about 20%, the average payment rate for online novels is 12.3%, and the payment rate for the emerging short - films is even as high as 40%.

If comparing with established internet products is considered unfair, let's take Meitu, a company that offers beauty and photo - editing software, as an example. The payment rate in the Chinese market is also 5.2%. Obviously, the Menlo Ventures report reveals the reality that the hype of AI in the capital market and on the internet does not match its actual impact. Users are willing to try something new but hesitate to pay.

What worries investors even more is that 81% of the total $12.1 billion in consumer spending in this report was spent on AI chatbots like Character.AI or so - called AI partners. So the question is: Why are consumers not willing to invest money and time in the hyped area of AI coding? One must consider that AI coding is hyped because it depicts a "everyone can code" future scenario. Technological equality through AI is the basis for the astonishing valuations of AI programming companies.

In fact, the process of AI penetration into all industries is progressing rather slowly. The reason is simple: If the current AI could really enrich everything, this should have been possible more than a decade ago with RPA (Robotic Process Automation). In other words, it's not that AI doesn't work well, but that the application barriers are too high, so the general public can't use it at all. Therefore, it is necessary to adjust the implementation and introduction methods.

Aside from other aspects, the AI enrichment of all things will inevitably lead to a corporate process division and reorganization, which will directly shock the existing management structure. One must consider that the digital transformation of many companies was only completed after many difficulties over a period of ten years. Another AI transformation could now break the back. Therefore, many business owners have already realized that the current AI is almost equivalent to an intelligent version of the office suite. Although it is sufficient as an aid, it is still unrealistic that it can completely replace human work.

The NANDA Lab at the Massachusetts Institute of Technology (MIT) showed in "The Gap of Generative AI: The State of Business AI in 2025" that many companies have high expectations for generative AI, but most projects have had no actual financial impact. Only about 5% of the pilot projects have achieved rapid sales growth, while most have stagnated and had no measurable impact on the company's profit and loss statement (P&L).

Therefore, the question is raised whether AI can actually be quickly converted into corporate profits. The foundation for the current AI boom is increasingly being questioned. The term "valuation pressure" now hangs over AI startups. Is there a way to refute the theory of an AI bubble? Of course: One can integrate advertising into AI products.

Among the leading AI products we can currently use, whether it's ChatGPT, Grok, Gemini abroad or DeepSeek, Doubao, Yuanbao in China, these are all the result of manufacturers' subsidies. The manufacturers are willing to subsidize users with real money and bear the high costs of computing power, data, electricity, and water to expand the market. The current user number of 1.8 billion shows that the market economy has already achieved initial success.

But since the payment rate is only 3%, integrating advertising for free users who pay nothing is a sustainable development option. By the way, researchers have found that those who know less about the principles and implementation process of AI tend to have blind trust. Researchers call these differences in acceptance tendency the "low cognition - high acceptance" relationship. This means that the advertising effect in AI products may be far better than traditional digital advertising.

Even AI manufacturers have ways to prevent users from being annoyed by advertising. It is known that users need tokens to use large - scale AI models. In many consumer products, these tokens are paid for by the manufacturers. The manufacturers only need to launch an activity like "watch a 30 - second ad and get 1000 tokens", and many users will be happy to do it. After Tencent Music and Qishui Music have already carried out the "watch an ad to listen to music for free" activity, this is not a new concept.

So we should cherish the good time now. The ad - free pure AI experience may soon be a thing of the past. The time when advertising is integrated into AI responses may not be far off.

This article is from the WeChat account "3e Life" (ID: IT - 3eLife). Author: San Yi Jun. Published by 36Kr with permission.