If all the AI was paid, would you be ready to pay to access it? “Not because for the moment, not using AI is not excluding”, “no, I do not use it daily and answer anything”, or “only if I need to work and it is my company that pays me the subscription.” These are the types of answers that can be heard when asking the question.
Today, the AI has not yet been essential to the eyes of people, not to the point of convincing them to pay a subscription to use it. But the development of artificial intelligence is a financial abyss that companies that put them in the market seek to fill. As a result, AI could become one day mandatory.
Ai again at the Freemium stadium
It should be said that most of the AI we know as chatgpt, Grok or Gemini has accustomed us to free access to their basic services. Only a few payments are offered for those who use it regularly and wish to benefit from augmented characteristics (unlimited applications, more efficient models, etc.).
This economic model, known as Freemium, is quite widespread in the digital ecosystem. It consists of a platform to primed customers with free, often unlimited services. Once a user is accustomed (even loyalty) to the free service, the platform will encourage him to go to the upper level and take advantage of an improved experience. But for this, the user will have to get the credit card.
This monetization strategy applies to both chatbots and chatgpt, as on electronic commerce platforms such as Amazon (with Amazon Prime) or even video games, where there are many examples (for example Candy Crush, League of Legend, Fortnite EITHER Rocket League).
The profitability of these models mainly implies what are called “whales” (translated by “whales” in French). It is a minority of users who will make many transactions and bring a lot of money to the platform, which will compensate for the free use of other people (as if a user paid for 10).
In the case of Chatgpt, only three million users in their 500 million messages pay a subscription to take advantage of Chatgpt Plus (23 euros per month) or Pro (229 euros per month). This brought almost $ 10 billion to the company in 2024 (an average income of 3,330 euros per payment subscriber).
An expensive technology
The AI will undoubtedly be one of the most promising sectors of the coming years. But technology is expensive, very expensive to develop. In 2023, Dylan Patel, semi -analysis chief analyst, estimated with The information That the daily cost of ChatgPT amounted to $ 700,000 per day.
If in 2025, the daily cost of operating Chatgpt is not known, it is most likely that it has reached new heights. Several parameters can explain a considerable increase in the price, such as the number of users that increase, models that always require more data and energy, or the development of increasingly diverse tasks.
This leads Openai in a loop where the company loses billions of dollars each year. In 2024 alone, the firm registered a deficit of $ 5 billion, despite the 10 billion won thanks to its subscriptions. And Sam Altman’s expenditure only intensifies since 2030, Operai plans to spend almost $ 200 billion.
A profitability that can be achieved
To compensate for these expenses and finally end their losses, Operai has established high financial objectives. The American company wants to succeed in generating $ 125 billion alone in 2029, more than 12 times more than it generates with its subscriptions currently. The goal would be to become profitable that same year, win more than you spend.
To reach such sums, there is no miracle. We will have to sell subscriptions or completely review the company’s economic model (including advertising, for example). That is why it is not impossible to see one day from Chatgpt only available in a paid version.
And in terms of alternatives such as Gemini, Grok, Claude, Le Cat, perpleity? In order not to leave the monopoly in Chatgpt, competitive companies could borrow and optaai to develop their own AI, and then follow the same financial scheme.
The paid has not yet convinced
But companies that innovate in the AI sector only know how to sell their services? According to a study published at the end of June by Menlo Ventures, not so much. In all chatbots, only 3% of users have opted for a payment subscription. This 3% still represents a gain of $ 12 billion per year. But for 100% paid users, we would reach 432 billion.
What are the main reasons for such disinterest? First, the price of a subscription, usually very expensive. Count between 20 and 350 euros for a monthly subscription, it is a matter of making it profitable. These packages paid generally have access to more powerful models, unlimited generation tools or even exclusive characteristics (we think of Grok Companions).
The most expensive subscriptions will be more appropriate for professionals. However, an individual who uses a chatbot from time to time will not see the usefulness.
Another argument that arises often remains the reliability of AI. The latter is still subject to recurring hallucinations. This is the reason why users still have some distrust of technology.
Free, still for a long time?
This disinterest even feels on electronic devices, whose main sale is precisely the integration of artificial intelligence. Copilot+ laptops are the perfect example. As the means of registration reports, the PC range manufactured by Microsoft is struggling to flow. The co -pilot+ are sold on average 57% more expensive than the average, a higher price that would be justified by the presence of AI.
“The manufacturers increasingly incorporate the characteristics of AI in their devices, which makes the PC equipped with AI a need instead of an choice for many buyers. But that does not mean that consumers actively seek these characteristics,” said the Marie-Christine registration registration record, an analyst in context.
Not finding a clientele ready to pay, the AI must still remain free for a few years. In any case, the companies that develop them have all interest in it. A first reason is to give time to your tool to democratize, retain or even become essential in the daily life of potential customers.
But, above all, this allows you to obtain an astronomical amount of training data for algorithms. These are essential for the development and improvement of models.
Source: BFM TV
