Mira Murati’s Thinking Machines reportedly raising $2B in funding – News

Thinking Machines Lab Inc., an artificial intelligence startup led by former OpenAI Chief Technology Officer Mira Murati, is reportedly raising a $2 billion seed round.

Business Insider reported the company’s fundraising effort today. The publication cited two sources as saying that the round could value Thinking Machines at upwards of $10 billion. In March, Insider reported that the company was seeking to raise $1 billion round at a $9 billion valuation.

Thinking Machines launched in February to build multimodal AI models with reasoning features. During her time at the OpenAI, Murati (pictured) oversaw the development of ChatGPT and the DALL-E series of image generation models. 

The Thinking Machines leadership team also includes other former OpenAI executives. John Schulman, one of the latter company’s co-founders, holds the role of chief scientist. ChatGPT co-creator Barret Zoph is Thinking Machines’ CTO. 

On Tuesday, News reported that the company had quietly appointed two other former OpenAI staffers as advisors. One of them is Bob McGrew, who was ChatGPT developer’s Chief Research Officer until last September. The other new adviser, Alec Radford, was the lead author of the academic paper that presented the GPT language model architecture.

Thinking Machines stated in a February blog post that its AI will be “at the frontier of capabilities in domains like science and programming.” However, the company hinted that its models will be more flexible than existing algorithms optimized for such tasks. “We’re building AI that can adapt to the full spectrum of human expertise,” the company’s staffers wrote.

Thinking Machines plans to equip its models with multimodal features. The company hasn’t specified whether those features will focus solely on processing multimodal input, or whether they will also allow the models to generate media files such as images and videos. Thinking Machines did, however, detail that it plans to make its AI easier to customize than existing algorithms.

The company plans to open-source at least some of the software its engineers will develop. In the February blog post, Thinking Machines hinted that the open-source initiative will place particular emphasis on sharing technologies related to AI safety. “We plan to frequently publish technical blog posts, papers, and code,” the company’s staffers wrote.

Training frontier AI models requires a significant amount of infrastructure. The $2 billion that Thinking Machines is reportedly raising would make it considerably simpler for the company to source the necessary hardware. According to Insider, the round is “round is in progress” and the details could still change. 

Photo: OpenAI