OpenAI is inscrutable. At least that's what it's seemed like since 2015, when this non-profit company appeared as if by magic, ready to develop something that, in just under nine years, has changed the world: artificial intelligence. Since Sam Altman and 10 other leading tech figures joined forces, a lot has changed and very little remains of that “foundation? NGO? that sought to “advance digital intelligence in order to benefit humanity as a whole as much as possible”.
The first years of the technology company's life passed quietly until 2019, when, seeing all the possibilities that AI could offer, OpenAI changed course and became a limited profit company, known as OpenAI LP. This corporate form allowed the company to attract significant investment to raise capital. They needed money, and the total donations fell far short of expectations.
Historic round
And so, with this dual approach, the company achieved the largest funding round in history, $6.6 billion, bringing the value of Altman's company to $157 billion, with a daily spend of $700,000. Despite this funding round, the company announced the formalization of a credit line of up to $4 billion from a dozen US financial institutions, including giants such as JPMorgan Chase, Goldman Sachs, Citi and Morgan Stanley.
According to The Information, the company spent around $7 billion on training models and $1.5 billion on its workforce.
In this historic round, Thrive Capital is OpenAI's largest backer, injecting approximately $1.3 billion, with an exclusive option to invest up to another $1 billion in the same valuation by 2025, according to The New York Times. In addition to Thrive, Microsoft, Nvidia, SoftBank, Khosla Ventures, Altimeter Capital, Fidelity and MGX also participated.
OpenAI has a few tricks up its sleeve: investors don't receive shares or equity in the company, but a contract that guarantees them a share of the company's future profits, if any. The second trick is the most impressive because, according to the Financial Times, Altman's company asked its new investors not to invest in rivals such as xAI or Anthropic. “During the negotiations, the company made it clear that it expected an exclusive investment agreement,” explains the business daily. Despite these conditions, the technology company had no problems carrying out this round and even managed to choose its investors.
So what now?
We enter the controversy. After completing this funding round a few days ago, the doubts in the Californian company haven't stopped, nor have the questions about what to expect from the creators of ChatGPT. The speculation is that they will close down this business hybrid and become a proper, non-altruistic, for-profit company. A B Corp, as they are known in the US.
This change involves many things and is based on the fact that OpenAI needs more and more money to be able to train models and develop what Altman wants, superintelligence.
The new structure will mean that the non-profit board of directors - the non-profit side of OpenAI - that controlled OpenAI until now will no longer do so, and investors will no longer be able to influence the company's decision-making. This doesn't mean that the non-profit side of the company will disappear, as it will continue to have a stake in the company, but will no longer control it as before.
There will be new shareholders and, with them, new shares and stakes. This will attract venture capitalists or large technology companies that don't compete with Altman's, who will also receive a stake of around 7%, according to various media reports.
And the operation will be as expected in a large company, since the profits that OpenAI makes will be distributed among investors and employees without any limit and proportionally to whoever owns the most shares in the company. Unlike in the past, OpenAI will also be able to reinvest part of its profits more aggressively in the company itself. And all this to achieve what Altman has been hinting at for a long time: going public.
The changes
But to get this far, Altman has lost founders and important people in the development of what was supposed to be a company that would put “artificial intelligence at the service of society”. Something that, a priori and if all goes according to plan, will not happen.
The exits highlight the company's internal conflicts, and the move to profit has caused friction among workers, who question whether it is ethical to launch their technology on the market, given the potential for misuse.
Many left the project when it was not yet the giant it is today, such as Vicky Cheung, who left in 2017 to join ride-hailing company Lyft after two years as head of infrastructure. She is currently CTO of the machine learning company Gantry, which she co-founded with another former OpenAI researcher.
Trevor Blackwell also left his position at OpenAI in 2017 to become a partner at startup accelerator Y Combinator until 2020. His personal website says he is working on “better ways to program robots”.
But some of these employees have recently been leaving for companies they consider to be more in line with their values. Anthropic has been a particularly popular destination. This year, it hired Jan Leike, former director of security at OpenAI; Schulman, co-founder and chief architect of ChatGPT; and Durk Kingma, another co-founder who also tried his luck at Google. It's important to remember that Dario Amodei, founder of Anthropic, was also vice-president of OpenAI.
“Anthropic 's approach to AI development is significantly in line with my own beliefs; I look forward to contributing to Anthropic's mission of developing powerful AI systems responsibly,” Kingma said earlier this week. This indicates that OpenAI's change of direction could have - and is having - serious consequences.
In this change and unraveling of the creators of Dall-E and ChatGPT, Ilya Sutskever, who left OpenAI to found Safe Superintelligence Inc. to create safe AI, has been left behind, now that it seems that this technology can lead to anything. Also announcing their departure recently was Mira Murati, Chief Technology Officer (CTO), who announced her departure after six years to focus on personal projects, as well as Andrej Karpathy, who joined Altman's in 2016 as a research scientist and founding member.
Karpathy's story goes back and forth, because in 2017 he went to Tesla, founded by Elon Musk - also ex-ApenAI - only to return in 2022 to Altman's company and leave in February this year. And the brain drain continues because this week he announced his departure from Sora, Tim Brooks who returns to Google DeepMind after working for Nvidia and OpenAI, to strengthen the video creation of the company that created the most famous search engine on the planet.
With this panorama, some see the end of OpenAI approaching, but others cling to the words of the company's president, Greg Brockman, who, despite having announced in August that he was taking a break until the end of the year, “the mission is far from complete; we still have to build a secure AGI”, so it seems that he will return to work on the project.
