It was only a matter of time: Microsoft and the company specializing in artificial intelligence OpenAI announced, on Monday, January 23, the extension of their partnership, which started in 2019. The new agreement extends over several years and includes ” several billion dollars “Investments, says Microsoft’s press release, without giving further details.
The tech giant thus retains the exclusivity of OpenAI tools on its Azure cloud. In other words, it will be the only one – with OpenAI – to market the uses of the DALL-E image generator, the GPT-3 text generator, or even the Codex code generator. Microsoft therefore strikes a blow in the race for artificial intelligence and even manages to worry one of its main competitors, Google.
Microsoft soon majority shareholder of OpenAI?
Earlier in the month, US media, citing sources familiar with the matter, put the new investment at more than $10 billion. After a complex financial agreement, Microsoft would recover 49% of the capital of the commercial branch of OpenAI while the parent company of the same name, which operates under the status of non-profit company, would only keep 2%. The 49% would remain in the hands of the other shareholders, who are mostly big names in Silicon Valley like Peter Thiel (PayPal, Palantir…), Reid Hoffman (LinkedIn), and employees like CEO Sam Altman (ex-Y Combinator).
The idyll between OpenAI and Microsoft began in 2019 with an investment of 1 billion dollars. This financial transaction paved the way for the marketing of OpenAI tools, while the company created four years earlier was only dedicated to research.
In 2021, Microsoft backed down – for an unknown amount – and began integrating into its Azure cloud tools such as the powerful GPT-3 natural language model – from which ChatGPT is derived, the chatbot superpower on which the spotlight has been shone since November 2022.
Fuel for OpenAI
This partnership is essential to the ambitions of OpenAI, which must pay researchers among the biggest names in the world, and run artificial intelligence models that consume a lot of computing capacity. According to Fortune, the company would have lost more than 500 million dollars in 2022 alone, and for good reason: it has so far recorded practically no income. But the company now hopes to achieve 200 million dollars in turnover in 2023, in particular thanks to the upcoming marketing of ChatGPT, and the expansion of its offer on the Azure cloud.
However, Microsoft’s investment is not only financial, it is also technological: the tech giant is building custom supercomputers for OpenAI to train its artificial intelligences with several billion parameters on trillions of data. .
Towards “general artificial intelligence” augmented copy of the human brain
As a reminder, OpenAI remains above all a research organization, which ultimately aims to create an artificial general intelligence (or AGI in English) capable of replicating the functioning of a human brain, or even exceeding its capacities.
If some experts doubt that such a goal is achievable, it pushes the company to develop very ambitious and resource-intensive artificial intelligence models. Result, the support of a computer giant quickly became essential, and the declension of its work in marketable tools was done more by pragmatism to ensure its survival than by real initial will.