Could Microsoft be about to steal a competitive advantage over Google in machine learning?
In 2019, Microsoft invested some $1 billion in OpenAI LP, the for-profit division of nonprofit OpenAI Inc., which was created in December 2015 by Sam Altman, Elon Musk, Reid Hoffman, Peter Thiel, Amazon Web Services and other investors, who between them contributed another billion dollars to develop open-source machine learning. Elon Musk then withdrew from the board, remaining with the company as a potential donor, with no active involvement in its management.
Microsoft’s investment was split between a small part in cash and another in Azure usage bonuses, enough for the company to live rent-free on Microsoft’s cloud until now (despite its opening to the public and the drop in the price of its API, which sharply increased its activity), rather than on Amazon’s. This form of investment, widely used by Microsoft with promising startups, limits its financial exposure but gives it access to a lot information about the company and the option to increase its stake in it.
Under the terms of the deal, Microsoft could now get rights to 75% of the profits generated by OpenAI LP until it has recouped its investment, and then take a 49% stake in the company, with another 49% divided among other potential investors, and 2% held by the non-profit parent, OpenAI, Inc. The company expects to earn revenue of $200 million by 2023, and to reach $1 billion by 2024.
That $1 billion is now looking like one of Microsoft’s best investments, prompting it to put a further $10 billion more into OpenAI, which with a valuation of $29 billion is one of the most valuable startups of the moment. The company is now looking at incorporating the features of GPT3 and later versions to products such as its search engine Bing, as well as to Word and its mail manager Outlook, thus providing them with potentially very useful writing assistants.
Could incorporating a machine learning assistant into its products give Microsoft a competitive advantage? People who have tested Chrome’s ChatGPT extension highlight the potential of a search engine able to answer questions, and that this could change how we use the web, seriously challenging Google’s hegemony.
Could the company whose stated advantage lay in its artificial intelligence being smarter than its competitors now be overtaken on the inside lane, in part for having failed to incorporate its LaMDA chatbot, into its products? And could the relatively limited application of machine learning through large language models give Microsoft the edge in this fast-growing area?
(En español, aquí)
