AI Startup Roundup: Anthropic Set to Raise $750 Million

Every week, AI Business brings you the latest startup news.

Keep up-to-date by subscribing to the AI Business newsletter to get content straight to your inbox and listen to interviews with AI experts from Fortune 500 companies on the AI Business Podcast on Apple iTunes and Spotify.

AI startups: bright spot in U.S. investments

U.S. investments in AI startups rose by 150% in the third quarter from a year ago, according to The Information, citing figures from PitchBook. That compares with a decline of 21% for all startups in general over the same period.

However, it is not as easy to get a check these days, even for AI startups. While clear AI leaders such as OpenAI and Anthropic continue to attract the most attention, others have to work harder to get funding. Part of the issue is that many AI startups use the tech, such as LLMs, of OpenAI and tech giants – which target the same clients as the startups.

Funding news

Anthropic, OpenAI’s staunchest rival that was founded by two OpenAI executives, is in talks to raise $750 million that would value the startup by as much as $18 billion, according to The Information. Earlier this year, Amazon invested $4 billion in the startup while Google participated in a $450 million series C round and also poured in $300 million.

It plans to spend “billions” of dollars over the next few years renting servers from AWS and Google to train and power its Claude language models, the news outlet said.

Anthropic reportedly said its revenue run rate would top $200 million by the end of this year and soar to $500 million by the end of 2024.

Despite OpenAI’s latest turmoil, Anthropic’s unusual governance structure did not deter investors. Its board is overseen by a 5-member trust who have no financial interest in the startup and are tasked with ensuring the AI models benefit humanity in the long run. Anthropic also is incorporated as a public benefit company, meaning its board can balance its financial interests with the public good.

The startup made a point early on to emphasize safe AI in its language models.

-

01.AI

Chinese AI startup 01.AI is developing open-source foundational large language models. It was launched by Lee Kai-fu, a venture capitalist and formerly the president of Google China.

The startup’s first model is Yi-34B, with 34 billion parameters, which was the first Chinese LLM to top the leaderboard on Hugging Face, according to Reuters.

Lee said he plans to open-source 01.AI’s smaller models and keep larger ones proprietary. He believes current popular foundation models skew towards U.S. values, which does not fit other cultures.

“Each country should have a high-quality model tailored to its culture, values, religion and language,” Lee wrote in an op-ed for The Wall Street Journal.

Latest funding: Up to $200 million, according to Reuters

-

Harvey

Legal AI startup Harvey offers a generative AI platform to help law firms, professional services providers and private equity firms find and analyze internal and industry data. The startup said revenue has risen 10-fold since April and engagement on the platform is up “exponentially.”

It is working with OpenAI to develop foundation models that are specific to certain domains.

Latest funding: $80 million, series B

Lead investors: Kleiner Perkins, tech entrepreneur Elad Gil

Other investors: OpenAI Startup Fund and Sequoia

Funding plans: Expand custom model building, scale its team and build out its suite of product features