OpenAI, which wanted to reinvent the relationship between humans and machines, now seems primarily focused on generating revenue. The artificial intelligence laboratory is transforming into a kind of conglomerate that appears to be driven by urgency: consulting, hardware, advertising, adult services, browsers, and perhaps soon office suites. This does not seem to be the result of a vision or product strategy, but rather a matter of survival. Behind the talk of innovation lies an economic constraint: AI must be made profitable before it goes off the rails.
In short:
- OpenAI is multiplying its commercial initiatives (consulting, hardware, browser, advertising, adult services) not because of a clear strategy, but to respond to growing economic pressures.
- ChatGPT’s growth is slowing, with paid subscriptions stagnating despite a still massive user base, which undermines the model’s profitability.
- Some activities, such as consulting or hardware with Jony Ive, are aimed more at reassuring investors and building a narrative than at generating immediate revenue.
- OpenAI is exploring very different, sometimes contradictory business models, which cloud the clarity of the project and expose the company to increasing operational complexity.
- The company seems caught up in a race to monetize in order to finance costly technology, at the risk of diluting its initial mission of innovation in artificial intelligence.
Euphoria, and then what?
For several months now, ChatGPT adoption curves have been flattening out. According to a Deutsche Bank research note, European user spending has stopped growing since May 2025 (ChatGPT Is Already Stalling Out on New Subscribers). TechCrunch has observed the same phenomenon on mobile devices: an 8% drop in downloads and a decline in daily usage in the United States (ChatGPT’s mobile app is seeing slowing download growth and daily use, analysis shows).
However, caution dictates that we qualify this statement: NerdyNav points out that the service still claims more than 800 million weekly users (Latest ChatGPT Statistics: 800M+ Users, Revenue (Oct 2025)) and revenue estimates are around $13 billion (OpenAI Hits $13B Revenue, Fueled by Consumer Subscriptions). But the momentum is no longer that of a rush, and for a company whose infrastructure costs are skyrocketing, the stagnation of part of its paying base is a warning sign (Is generative AI doomed to never be profitable?).
This is when a model needs to reinvent itself, and OpenAI is no exception to the rule.
Artificial intelligence discovers consulting
Last July’s announcement of a $10 million consulting contract raised eyebrows among observers:the company that created the product that was supposed to “kill consulting” by automating human expertise is now inventing its own consulting business. Forbes detailed how it works: dedicated teams, integration of customized models, deployment support in large organizations (OpenAI’s $10 Million+ AI Consulting Business: Deployment Takes Center Stage).
Economically, the operation makes sense: low volume, high margins, and a positive image, but it muddies the message: artificial intelligence is now being sold as a human service. In any case, these services, while lucrative but limited, represent only a fraction of revenue (OpenAI’s $10M AI Consultancy Revolution: The New Era of Premium AI Services), a temporary stopgap rather than a structural pillar.
AI was supposed to automate cognitive jobs, but in the end, it is recreating them.
The Ive Project: Giving the Machine a Body
The acquisition of io Products, founded by Jony Ive, for approximately $6.5 billion marks another turning point interpreted as OpenAI’s desire to design a “post-screen” device (OpenAI acquires iO for $6.5 billion: strategic pivot or headlong rush? and OpenAI closes its deal to buy Jony Ive’s io and build AI hardware). But this shift from software to hardware implies a change in business: supply chains, manufacturing, logistics, and maintenance are all industrial realities that are foreign to the culture of the laboratory.
The idea seems to be to anchor artificial intelligence in a tangible, identifiable object, perhaps in a future autonomous personal assistant, but the reality is less simple, with design delays and a dispute over the product name (OpenAI and Jony Ive may be struggling to figure out their AI device). The integration of Ive’s team is proving more complex than expected, and no commercialization is expected before 2026. I also noticed that the pages of the OpenAOI website that had announced the acquisition and the project at the time have been removed.
At this stage, one might wonder whether this is initially more of an image-building exercise than a short-term revenue driver. OpenAI wants to show its investors that it is aiming for a “post-ChatGPT” era, but that era still needs to exist.
The browser, a new frontier
On October 21, 2025, OpenAI announced ChatGPT Atlas, its own web browser: “a new web browser built with ChatGPT at its core.”
It’s a logical step: if AI becomes an interface, access to the web must be controlled. But it is also a sign of dispersion: OpenAI, which until now has focused on content generation, is entering a saturated and highly capital-intensive market, as we are no longer talking about a traditional browser.
At the same time, the company seems to be “exploring the office software market” to compete with Microsoft 365 and Google Workspace (OpenAI explores office software market as ChatGPT gains business traction). There has been no official announcement, but OpenAI is reportedly considering an office suite that natively integrates its models. The idea is more than coherent, namely to control the environment in which its models are used, which is a real competitive advantage enjoyed by Microsoft and Google, but it reinforces the impression of industrial dispersion (Does OpenAI want to, should it, and can it become the new Google?).
The blurred lines of diversification
Since the end of the summer, several weak signals have indicated that OpenAI is also exploring advertising, although this has not been officially confirmed (OpenAI Looks to Build In-House Ad Infrastructure). The idea would be to insert recommendations or branded content into certain conversations. The hypothesis is not absurd, as the model of massive but free use calls for some form of monetization, but it contradicts the original discourse of sobriety and ethics in the user experience and what Sam Altman himself said recently about preferring to avoid advertising and instead make money with his product (An Interview with OpenAI CEO Sam Altman About Building a Consumer Tech Company).
Even more delicately, it appears that OpenAI now allows certain developers to create “AI companions” with erotic content, officially reserved for an adult audience (ChatGPT is about to get erotic, but can OpenAI really keep it adults-only?). The company explains that it wants to “regulate” this type of use to prevent it from developing outside its ecosystem, but the technical boundary between moderation and permissiveness remains porous. But it is no longer a rumor: OpenAI is experimenting with a form of intimate conversation, cautiously, but it is doing so.
These initiatives, as different as they may be, have one thing in common: they show a search for new sources of revenue, including in areas that OpenAI had recently ruled out.
Incompatible business models
Some may see this as a moral issue, but the problem is also structural.
These business models do not have the same costs, cycles, margins, or scalability. However, investors prefer clarity: too many business models obscure the risk and weaken valuation.
OpenAI has the advantage of not being listed and therefore escapes immediate market sanctions, but its main investors are not financing a laboratory to turn it into a conglomerate. Diversifying business lines is a way of reassuring investors about revenue, but it does not reassure them about the consistency of the project.
Under cost pressure
Each GPT model requires astronomical amounts of computing power and energy. OpenAI has placed record orders for chips from Nvidia and Broadcom, spending billions of dollars to support the next generation of models (OpenAI big chip orders dwarf its revenues — for now). This is an investment, but above all, it represents a huge structural dependency.
The plan to design its own processors with Broadcom could eventually reduce the bill, but it exposes the company to a second level of risk: becoming both a producer and consumer of hardware in a sector already saturated by giants.By seeking to control its costs, OpenAI is actually multiplying them.
The unstable balance between pragmatism and vision
All this does not resemble a long-term strategy but rather a series of short-term reactions. The advice to compensate for subscriber stagnation, the hardware to fuel the narrative, the browser to capture data, advertising to extract value from a free base, “adult” chats to occupy a lucrative niche left to others.
Each move is rational when taken in isolation, but together they look like a headlong rush: a company that is no longer exploring artificial intelligence, but rather the business models that can finance it.
The deliberate leak
These wide-ranging moves, from hardware to advertising, consulting, and browsers, are less a strategy than the reflex of a company that is experimenting to buy itself the time it needs to make its core business profitable.
OpenAI’s strategy therefore seems to be movement. Its initial ambition, to push the boundaries of artificial intelligence, has given way to a race for monetization. It is no longer a question of exploring what the technology can do, but what it can bring in.
Each diversification is a tactical response: consulting to boost revenues, hardware to reassure investors, advertising to make usage profitable, and derivative products to occupy the media landscape.
OpenAI is no longer seeking only to invent artificial intelligence, but also to invent the revenues that will keep it alive.
The question now is no longer how far technology can go, but how long the business model can hold up.
Conclusion
What OpenAI is going through is nothing unusual: all companies born out of a technological promise eventually have to face the reality of the numbers. What is unusual here is the speed at which things are happening and the gap between visionary rhetoric and financing logic.
On the one hand, the organization still seeks to present itself as a cutting-edge laboratory, but on the other, it is piling up a variety of activities to generate revenue. This diversification can be seen as a sign of pragmatism or as a symptom of a company that has become a victim of its own success. Artificial intelligence has never been free, and the bill is simply starting to show.
The question is therefore no longer whether OpenAI will invent the next cognitive revolution, but whether it will be able to survive in the real economy.
To answer your questions…
Faced with the slowdown of ChatGPT and colossal costs, OpenAI is seeking new sources of revenue. Consulting, hardware, browsers, or advertising: each initiative aims to make AI profitable and reassure investors. This diversification reflects less a long-term vision than a reflex for economic survival.
The consulting business allows OpenAI to monetize its high-margin expertise. This shift, which seems paradoxical for a company that advocated automation, is primarily aimed at compensating for stagnating subscriber numbers. It is more of a prestigious showcase than a sustainable pillar, but it is useful for attracting large companies.
By acquiring Ive’s company, OpenAI aims to create a “post-screen” device that embodies AI. This ambitious project symbolizes the search for a “post-ChatGPT” era, but it remains risky: delays, industrial costs, and difficult integration. It is more of an image gamble than an immediate source of revenue.
ChatGPT Atlas wants to make AI the new interface for the web. By launching its browser, OpenAI controls access to information and strengthens its ecosystem. But this saturated and costly market illustrates a strategic dispersion rather than a clear vision.
OpenAI empile des modèles économiques très différents, aux coûts et marges incompatibles. Le conseil, le matériel et la publicité complexifient sa structure et brouillent sa cohérence. En cherchant à réduire ses coûts, l’entreprise semble surtout multiplier les risques.
Image credit: Image generated by artificial intelligence via ChatGPT (OpenAI)
Related articles:
What do the AI giants really want?
Is generative AI doomed to never be profitable?
Does OpenAI want to, should it, and can it become the new Google?
AI heading for an economic dead end?
Generative AI: a bubble, a crash, or a turning point?
OpenAI acquires iO for $6.5 billion: strategic pivot or headlong rush?
