Microsoft said to be in talks to invest more in OpenAI
Software giant is already in deep, giving the startup $1 billion in 2019
Microsoft is reportedly in advanced talks with OpenAI to invest more in the seven-year-old research company.
Both The Wall Street Journal and The Information, citing unnamed sources, said no deal has been reached and no funding amount settled on, but the two companies have met in recent weeks to discuss the possibilities.
According to WSJ, OpenAI uses a huge amount of compute power to run its AI products on Microsoft Azure. Funds from any new investment could be used to help offset those costs.
The money would come on top of the $1 billion Microsoft invested in OpenAI in 2019 in a deal that included OpenAI agreeing to jointly develop new supercomputing technologies for Azure based on its AI products and making Azure its exclusive cloud provider.
"We'll be working hard together to further extend Microsoft Azure's capabilities in large-scale AI systems," OpenAI officials wrote in a blog post announcing the 2019 investment.
OpenAI was founded in 2015 by Elon Musk and several other investors and, according to The Information, is now valued at almost $20 billion. That was based on an undisclosed sale of OpenAI stock to existing shareholders, with the price of the shares suggesting to unnamed sources the startup's overall valuation.
The Register has asked both Microsoft and OpenAI for comment. We will update the story when either or both respond.
The AI space is expanding rapidly, with IDC analysts this year predicting the global market will grow an average of 18.6 percent a year through 2026, when it will hit $900 billion. OpenAI is focusing on what's called artificial general intelligence (AGI), which is aimed at creating cognitive capabilities in software so that an AGI-based system could find a solution to an unfamiliar intellectual task in a fashion similar to humans.
Microsoft already offers OpenAI technology in Azure and last year introduced the Azure OpenAI Service, which combines OpenAI's models and Microsoft capabilities like security, compliance, and regional availability in Azure.
Microsoft unveiled Azure OpenAI Service at its Ignite 2021 show as a new product within Azure Cognitive Services, which itself is part of the Azure AI portfolio. At that time, it was available only by invitation.
At its Build 2022 conference in May, Microsoft announced the Azure OpenAI Service was now available in a limited access preview, with organizations able to apply for access.
- AI programming assistants mean rethinking computer science education
- How GitHub Copilot could steer Microsoft into a copyright storm
- Weird robot breaks down in middle of House of Lords hearing on AI art
- Just $10 to create an AI chatbot of a dead loved one
The service has been expanding its range of capabilities, with features including filtering out harmful content, detecting abuse, and access to more models, such as GPT-3 (natural language), Codex (code generation and plain language translation), and embedding models (for semantic search and other tasks).
More recently, OpenAI rolled out DALL∙E 2, which allows users to use strings of text or images as input to generate custom images. It initially was open as a limited beta, but in late September the company removed the waitlist for the technology, enabling anyone to sign up for it.
At last week's Ignite 2022 show, Microsoft said that DALL∙E 2 would be available in Azure, initially by invitation only to select Azure OpenAI Service customers. The company is integrating Dall-E 2 into its Edge browser and Bing search engine with its Image Creator tool, which soon will be coming to Edge.
The company is also integrating it into Microsoft Design, a graphic design app for presentations, posters, and other graphics.
"The rollout of DALL∙E 2 across Microsoft products and services reflects how the company's investment in AI research is helping to thoughtfully infuse AI into everything it builds, produces and delivers to help everyone boost productivity and innovation," Liat Ben-Zur, corporate vice president of modern life, search, and devices at Microsoft, wrote in a blog post this week.
Other vendors have created such generative AI models, including Google Cloud with Imagen and Stability AI – which this month raised $101 million in seed round investments – with Stable Diffusion. Such tools can be used for everything from text, image, and music generation to data augmentation and algorithm invention.
In announcing on September 28 that it was removing the waitlist for DALL∙E 2, OpenAI said more than 1.5 million users were creating more than 2 million images a day with the technology. In addition, more than 100,000 users were sharing their creations in the company's Discord online community. ®