$650 Million for Casetext & $1.3 Billion for MosaicML + Massive New AI Funding Rounds Mean Great Week for Generative AI
I talked with the CEO of MosaicML and investors in this week's big acquisitions
Two major AI startup acquisitions and three large AI funding rounds — all announced this week — only amplified the excitement for generative artificial intelligence startups.
“This is not crypto. This is real,” Naveen Rao, the CEO of MosaicML, who just sold his startup to Databricks, told me this week.
In case you missed it:
The generative AI platform company Mosaic announced an agreement to sell to Databricks Monday in a $1.3 billion deal, including retention packages. The Information reported that Databricks is paying 65X Mosaic’s $20 million annual recurring revenue.
The legal technology company Casetext, which used ChatGPT to build a co-pilot-style service for lawyers, announced that it was selling for $650 million to Thomson Reuters.
The empathetic large language model startup Inflection raised $1.3 billion at a $4 billion valuation in a round that included Microsoft, Nvidia, Bill Gates, and Eric Schmidt participated in the round. Inflection co-founder Reid Hoffman also contributed to the round.
The AI-powered video editing company Runway raised $141 million from Google, Salesforce Ventures, and existing investors at a $1.5 billion valuation.
The AI content creation company Typeface raised $100 million from Salesforce and Alphabet at a $1 billion valuation.
While the large fundraising rounds for artificial intelligence companies have become par for the course, the acquisitions are something new.
I talked to some key investors at Casetext and MosaicML, along with the companies’ CEOs.
Cold Hard Cash for Casetext
Perhaps the most important development this week was Thomson Reuter’s decision to pay $650 million in cash to buy the legal technology company Casetext.
Everyone in startup world has been burned too many times at this point to believe in sky-high private valuations. Investors are willing to attach all sorts of reckless valuations to startups. But actually purchasing a company for $650 million in cash is something else altogether. That’s a real credible evidence that old-guard companies see a real value in generative artificial intelligence.
The other thing exciting about the acquisition of Casetext’s is that generative AI was truly essential to the startup’s recent success. The company had a long journey toward this exit, with the company’s partnership with OpenAI fueling much of its recent product improvements and customer growth.
As Casetext CEO Jake Heller told me a few weeks ago, his company had come out of the same Y Combinator Summer 2013 batch along with DoorDash and Webflow. At first, Casetext thought it would ride the user generated content trend that was fueling the likes of Quora and Reddit.
That’s part of what excited Union Square Ventures about investing in the startup.
“The original idea was to crowdsource legal annotation — legal research and annotation,” says John Buttrick at Union Square Ventures. “We like backing big ideas which don’t necessarily have a business model at the time we invest.”
But crowdsourcing didn’t take. The company instead leaned into machine learning and automation to build its tools to help lawyers search for relevant caselaw.
Casetext’s decision to partner with OpenAI and implement ChatGPT into its product led to explosive growth. The company announced a co-pilot for lawyers called CoCounsel just a few months ago.
For anyone who was arguing that building on top of ChatGPT can’t create great businesses — Casetext is a sign to think again. For Thomson Reuters — a competitor to my old employer Bloomberg and to legal data companies like LexisNexis — Casetext’s technology and experience working with ChatGPT could give it a real edge over the incumbents in legal research technology.
After Union Square led the Series A round in November 2016, Rebecca Lynn at Canvas Ventures led the $12 million Series B the following year, according to PitchBook.
Buttrick estimated that his firm invested about $13 million and owned about 14% of Casetext. At closing the investment should generate a more than 6X return for USV.
MosaicML + Databricks = IPO Vehicle for AI Hype?
Some people think the $1.3 billion price tag for MosaicML — a startup that’s making it easier for companies to build their own foundation models — is ludicrous. The company had just $20 million in annual recurring revenue, according to The Information.
After talking to investors, I’ve actually started to think it makes a lot of sense.