Recursive Superintelligence raises $500 million in funding at $4 billion value

In short: A new AI research start-up called Recursive Superintelligence has raised at least $500 million and was valued at about $4 billion.

What happened

Recursive Superintelligence is a four-month-old company founded by former engineers and researchers from Google DeepMind and OpenAI, according to the Financial Times. People familiar with the deal said the funding round was led by GV, which is Google’s venture capital arm, and included Nvidia.

The start-up’s valuation was put at $4 billion, not counting the new money raised. The FT reported that after the initial $500 million, investor demand was so high that the company could raise as much as $1 billion in total.

The founders include Richard Socher, a well-known AI researcher who previously worked as chief scientist at Salesforce, and Tim Rocktäschel, an AI professor at University College London who recently worked at DeepMind. The company reportedly has about 20 staff, including former researchers from OpenAI, as well as people who previously worked at Google and Meta.

Recursive’s long-term goal is to build an AI system that can keep improving itself without people constantly guiding it. Think of it like a student that not only learns from lessons, but also writes its own new lessons and keeps testing itself. The FT noted that this idea is still at the research stage and has not been proven to work for long periods.

Why it matters

Big funding rounds like this show that investors are still putting large amounts of money into AI research labs, even when the work is uncertain and may take years. If self-improving AI ever works as promised, it could speed up how quickly new software is created, but it could also raise harder questions about safety and control.

Source: Financial Times