From the outside, OpenAI seems to have positioned itself at the center of a technological revolution. Its flagship product, ChatGPT, demonstrated nearly two years ago how generative artificial intelligence could change digital life. Its potential has attracted cascades of investment and hundreds of new startups, it has driven strategic changes at the world’s most valuable companies, and it is being touted as the embryo of history’s first conscious machine.
From the inside, the situation can be very different.
Last August, OpenAI saw two more of its founders leave. One went to Anthropic, its biggest competition. The other, Greg Brockman, its president until now, announced that he was leaving the organization for an “extended” period in order to devote more time to his family. With these departures, only two people remain of the eleven who gave the starting signal. And among them Sam Altman, its famous CEO, who was fired in November 2023 after losing the confidence of the board of directors, but was reinstated a few days later after a labor riot and complaints from investors.
High labor mobility is a hallmark of tech startups. It’s not so common in companies valued at $86 billion and with collaboration agreements with industry heavyweights like Microsoft or Apple. Nor that the fate of a good portion of the co-founders is another AI company, Anthropic itself, created by them in the belief that OpenAI has strayed from its initial mission. Or that one of them, Elon Musk, the very same one who invested most of the money that allowed it to take off, is suing him for the same reason.
OpenAI was founded in 2015 as a kind of dream team of engineers and researchers specializing in artificial intelligence. A technology lab. “As a non-profit organization, our goal is to create value for all, not for shareholders. “Researchers will be strongly encouraged to publish their work, whether in the form of articles, blog posts or code, and our patents (if any) will be shared with the world,” he said in his cover letter.
However, the direction of the organization, especially since Altman took over, has been different. OpenAI created a private company, also called OpenAI, dedicated to raising capital for the lab. The result has been that the project behaves like any other company, patenting and selling its technology to users and partners. “OpenAI has become a de facto closed subsidiary of the largest technology company, Microsoft,” Musk protested in his lawsuit.
AI needs money, lots of money
Musk eventually withdrew the request after OpenAI published messages in which he himself acknowledged that to fulfill its mission, the organization that would create ChatGPT years later would need money. Lots of money. “We need to start with a number well above $100 million to avoid appearing desperate… I think we should say we are starting with a $1 billion funding commitment. “I will cover what others don’t,” the mogul wrote in one of them.
“We all understood that we would need much more capital to succeed in our mission: billions of dollars a year, which was far more than any of us, particularly Elon, thought we could raise as a nonprofit organization. Elon and we recognized that a for-profit entity would be necessary to obtain these resources,” OpenAI explains.
Getting resources is exactly what the company has been dedicated to since then. Altman has become a great specialist in this work. “The best thing he does, by far, is raising funds. That is his real talent. Knowing that at the moment, OpenAI’s business model is based on losing your shirt, because it requires constant assaults on the capital markets,” explained Javier Recuenco, CEO of Singularsolving, in a report on elDiario.es.
But this ability to raise money has eventually become a problem for OpenAI itself, which has taken to burning other people’s money to operate. The organization does not yet have a stable business model, nor is it able to generate profits on its own. According to an analysis by The informationSilicon Valley-based research media company OpenAI could post losses of around $5 billion this fiscal year, adding to the additional $8.5 trillion it has accumulated in previous years.
A very complicated economic model
The question of how OpenAI will manage to start generating revenue is common to most AI startups. Multinationals like Google or Microsoft are adding this technology to their products already on the market as an added value. But it does not seem that subscriptions to AI systems like ChatGPT are an economically viable model for startups that had no place in the sector.
The problem lies in the high computational cost of AI, which means that each new subscription involves an additional cost. “This circumstance determines a very clear economic structure: companies like OpenAI charge more and more over time and the use of their algorithms becomes popular, but their costs also increase uncontrollably, which puts them in a very complicated survival perspective,” explains Enrique Dans, professor of innovation and technology at IE Business School.
But the doubts do not end there. In these two years, a large number of AI companies have emerged whose systems do not seem too different from each other. OpenAI continues to be a leader in research, but it seems that this advantage is due to the significant financial resources it has and the support of Microsoft. ChatGPT is not a system that others will have difficulty matching, as Google’s search engine or Facebook’s social network used to be. On the contrary, in these two years, others have emerged, such as Gemini (Google), Claude (Anthropic) or Perplexity, comparable to its predecessor. An attempt by Wall Street Journal found few differences between them.
All this has not gone unnoticed by investors. The Elliot Management fund has already openly declared to its clients: AI is “overrated” and many of its uses “will never be profitable, will never work very well, will consume too much energy or will be unreliable,” it warns in a report to which the Financial Times had access.
These are the types of funds that finance this type of development. But your money is no longer circulating as before, the voices that speak of a “bubble” are multiplying and AI companies are generating more uncertainty. OpenAI, which needs the most money to operate, could be the most exposed to a crisis in the sector.
OpenAI could end up being seen as the WeWork of AI
Voices like AI expert, data scientist, and writer Gary Marcus see this explosion coming ever closer. “Generative AI itself isn’t going away, to be sure. But investors could stop spending money at the rate they used to, enthusiasm could wane, and many people could lose their shirts. Companies currently valued in the billions of dollars could go bankrupt or be broken up,” he warns.
“By the end of 2024, things could be radically different than they were just a few months ago,” continues Marcus, who has been one of the most critical voices regarding the development of AI startups and their business model since the beginning.
“I’ve said it before, and I’ll say it again: OpenAI could end up being seen as the WeWork of AI. Huge operating losses, huge payroll costs, apparent morale problem, no GPT-5 yet, competitors playing catch-up, Meta spinning off similar technology, Microsoft making deals with competitors,” he said after learning of Brockman’s departure. “Will they make enough to justify their $80 billion valuation? Will they raise another round at an even higher valuation when they run out of money? The outlook doesn’t look as strong as it used to.”
