Source: French to English Tester Published on: 2026-04-07
Source: The Conversation – in French– By Frédéric Fréry, Professor of Strategy, CentraleSupélec, ESCP Business School
The company OpenAI, creator of ChatGPT, could soon go public. Behind this financial operation lies much more than just a fundraising: the shift of a project designed for the public interest towards a market logic. At a time when artificial intelligence is becoming a critical infrastructure, a question arises: can its development be entrusted solely to financial markets?
OpenAI was born in 2015 amidst growing concerns around artificial intelligence (AI). Founded notably by Sam Altman and Elon Musk, it adopted a nonprofit structure. Its explicit goal was to develop AI “beneficial to humanity” and to prevent it from being captured by a few dominant players. This ambition set it apart from major technology companies like Google, Microsoft, Meta, or Amazon, built on proprietary models andrent effects.
By contrast, OpenAI intended to defend the public interest by emphasizing open research and the dissemination of knowledge. However, this orientation, symbolized by its name, OpenAI (“Open AI”), quickly encountered a structural constraint: the astronomical cost of generative artificial intelligence.
Massive costs
Unlike traditional software, whose marginal cost tends towards zero (the millionth copy of Windows costs Microsoft nothing, for example), generative AI requires massive infrastructures. Each interaction mobilizes computational resources, energy, and specialized equipment. A standard request on ChatGPT, including a question and an answer,thus costs on the order of 0.1 to 1 cent of a dollar. Similarly, generating a high-definition image can cost between 10 and 20 cents. Taken individually, these amounts seem negligible, but at the scale of several billion daily requests in 2026, they become significant.
This is explained by the underlying infrastructure, in particular the graphics processors (GPUs), provided by players such as Nvidia. These chips can cost several tens of thousands of dollars to purchase and several dollars per hour in access fees.cloud. OpenAI, like its competitors, depends on tens of thousands of these GPUs running continuously in gigantic data centers.According to some estimates, the necessary investments amount to hundreds of billions by the end of this decade.
Also to read:
Why the rise in the price of computer RAM is a threat to the European economy
By the end of the 2010s, it had become clear that a purely non-profit model could not meet such capital intensity. That is why OpenAI adopted, in 2019, a hybrid status allowing it to raise funds while maintaining control through a foundation. This was a first foray into the market economy, but tempered by the ambition to resist investors’ demands.
Sudden acceleration with ChatGPT
However, at the end of 2022, the conversational agent ChatGPT radically changed the game by attracting 100 million users in just two months, beforesurpass 900 million weekly users by early 2026. OpenAI’s revenue thus went from about 200 million dollars (173.15 million euros) in 2022 to over 10 billion (8.65 billion euros) in 2025, a sixty-fold increase in three years!
This exponential growth was accompanied by the implementation of aeconomic modelincluding multiple sources of income. For individuals, OpenAI offers paid subscriptions (ranging from $20 to $200, approximately €17 to €173, per month). However, the majority of the revenue comes from businesses,viasubscriptions offered between 25 and 60 dollars (from 21.6 to 52 euros) per user per month. A company with 10,000 employees thus represents an annual revenue of several million dollars.
Corporate money
Additionally, OpenAI charges businesses that integrate its models directly into their own solutions for their usage. Each use is then billed, often on a very large scale. An application processing one million requests per day can thus generate several tens of thousands of dollars in monthly billing.
Finally, an increasing share of revenues comes from strategic agreements, notably with Microsoft, which integrates OpenAI’s technologies into its products under the name Copilot.
It is the total of these flows (subscriptions, licenses, usage by third-party companies, and partnerships) that allowed OpenAI to reach approximately one billion dollars (more or less 865.85 million euros) in monthly revenue in 2025. Yet, this commercial rise conceals an intrinsic economic fragility.
A gigantic machine to burn cash
However, despite strong revenue growth, OpenAI remainsstructurally deficient. In the first half of 2025, the company would have achieved a turnover of about 4.3 billion dollars while recording losses between 7 and 13 billion, or more than 2 billion in losses every month. In total, its cumulative losses couldexceed 140 billion dollars(121.19 billion euros) between 2024 and 2029.
This drift is explained by the very nature of OpenAI’s economic model, where each interaction generates a cost, with gargantuan necessary investments. Besides infrastructure, research and development (R&D) constitutes a major expense. To stay in the technological race in an increasingly competitive environment, OpenAI is said to have invested nearly 16 billion dollars (more than 13.8 billion euros) in research and development in 2025 alone.
Added to this is the cost of human resources,sometimes out of the ordinary. While the base salaries of the most sought-after artificial intelligence experts already range between $250,000 and $700,000 (between €216,000 and €605,000) per year, their compensation – including stock options and bonuses – often exceeds one million dollars. In some cases, the annual remuneration even exceeds 10 million dollars (€8.65 million). Once again, the bidding war among competitors, such as Meta, forces OpenAI to match these offers, fearing the loss of key talents.
A bankruptcy filing imminent?
Overall, OpenAI’s activity is not sufficient to cover its costs, to the point thatsome analystsmention that, at this pace, it could be forced to file for bankruptcy as early as 2027. External financing is therefore essential to cover these losses.
To support its growth, OpenAI has already raised about 58 billion dollars (more than 50 billion euros) since its creation, including more than 13 billion dollars from Microsoft. In 2025, an exceptional funding round would have allowed raising up to 40 billion dollars more, thus bringing its valuation toseveral hundred billion dollars. At the end of March 2026,a new fundraising of 122 billion dollars (more than 105.6 billion euros), notably from Amazon (50 billion dollars), Nvidia and SoftBank (30 billion dollars each), made it possible to reach a valuation of 852 billion dollars (or 737.6 billion euros). However, these amounts remain insufficient relative to the needs.
Industrial dependence
Dependence on industrial partners appears to be particularly problematic. Microsoft provides OpenAI with its infrastructurecloud viaAzure, while Nvidia plays a key upstream role by providing the GPUs. Just like during the gold rush era, it is the shovel sellers who got rich at the expense of the prospectors; in the artificial intelligence sector, it is the infrastructure suppliers who make fortunes, not the model designers.
In practice, each artificial intelligence request generates revenue for infrastructure providers, which amounts to a form of “invisible tax” captured upstream. In 2025, Nvidia thus generated nearly73 billion dollars in net profits for a turnover of approximately 130 billion(or 112.5 billion euros), and its stock market valuation is 1.5 times higher than that of the entire CAC 40!
Wanderings in the government
OpenAI’s economic tensions have impacted its corporate governance. The hybridization between a public interest mission and private financing mechanisms has resulted in acomplex structure. A nonprofit foundation thus controls a commercial-type companypublic benefit corporation(an American and less restrictive version of the French mission-driven company), funded by investors and responsible for raising capital and developing activities, while theoretically remaining subordinate to the foundation’s general interest mission. This structure, designed to avoid a purely financial logic, quickly sparked tensions between different stakeholders.
TheElon Musk’s departure in 2018constituted a first signal of strategic disagreement. In 2020, several researchersleft OpenAI to found Anthropic, highlighting differences on security and governance. However, it is mainly theNovember 2023 crisiswhich fully revealed the weaknesses of the system, when the board of directors suddenly announced the dismissal of Sam Altman, citing a lack of transparency in his communications.
In a few hours, the situation turned into an open crisis. Almost all employees threatened to leave the company if Altman was not reinstated. Microsoft, the main partner and investor, publicly expressed its support for Altman and even mentioned the possibility of hiring him along with his teams. Faced with this pressure, the board of directors was forced to reverse its decision within a few days. Sam Altman was reinstated, and the composition of the board was profoundly reshuffled.
This episode highlighted internal tensions, notably the difficulty of reconciling divergent logics within the same organization, between ethical stance, industrial imperatives, and investor demands.
Competition that is intensifying
To these internal constraints is added a particularly intense competitive pressure. Google, the inventor of generative AI, is making rapid progress with Gemini. Anthropic, with Claude, has established itself in certain segments, notably programming, while emphasizing security. The Chinese company DeepSeek has claimed to use less expensive processors. The French company Mistral AI advocates a frugal approach and Europe’s digital sovereignty. A sign of this turnaround: Apple, initially a partner of OpenAI to include ChatGPT for certain functionalities related to Siri, has chosen to replace it with Gemini.
In this context of ecosystem reorganization, OpenAI’s position, although still central, is being challenged. The intensification of competition increases the need to have ever larger financial resources.
The Stock Market, Rescue or Mirage?
OpenAI’s public offering is presented as a response to these constraints: funding massive investments and consolidating a weakened competitive position. The stock market could make it possible to raisebetween 50 and 100 billion dollars, by yielding between 10 and 20% of the capital. Such an operation would constitute one of the most important in the history of financial markets.
However, this transformation involves delicate trade-offs. A publicly traded company is subject to profitability and transparency requirements that can conflict with the experimental nature of artificial intelligence. Added to this is the persistent dependence on Microsoft and Nvidia, which limits the company’s strategic autonomy.
Above all, there is nothing to indicate that going public would be enough to solve OpenAI’s structural problems. At best, without a significant change in the economic model, it would only delay its bankruptcy by a few years. The economic model of generative artificial intelligence remains fundamentally unstable today.
A question that goes beyond OpenAI
Beyond the case of OpenAI, one can legitimately question the current functioning of the economy dominated by technology giants. Artificial intelligence is establishing itself as an essential infrastructure, whose effects go far beyond the economic framework. For some analysts, control over AI now carries the same geostrategic importance as the possession of nuclear weapons.
Hence, a civilizational question arises: can the development and direction of such a technology be entrusted solely to the financial markets? Can one imagine Elon Musk or Mark Zuckerberg personally possessing the equivalent of one or several atomic bombs? The IPO of OpenAI alone will not provide the answer. However, it will constitute one of the first large-scale tests.
Frédéric Fréry does not work for, advise, own shares in, or receive funds from any organization that could benefit from this article, and has declared no affiliation other than his research institution.
–ref. OpenAI on the Stock Market: can the markets govern ChatGPT and AI?https://theconversation.com/openai-in-the-stock-market-can-markets-govern-chatgpt-and-ai-279431