When four leading artificial intelligence researchers left Google this year to create a start-up called Mobius AI, they weren’t sure what their product might be — just that it would involve A.I. technology that could generate its own photos and videos.
Within about a week, two of Silicon Valley’s top venture capital firms, Andreessen Horowitz and Index Ventures, had swooped in with a funding offer, three people with knowledge of the matter said.
Suddenly, Mobius — little more than more than four guys and a laptop — was valued around $100 million, an usually high number for a start-up that was just a week or so old, the people said. When word of the deal leaked out, other investors descended to urge Mobius to take their money too, they said.
Over the past few months, a gold rush into start-ups working on “generative” artificial intelligence has escalated into a no-holds-barred deal-making mania. The interest has mounted so rapidly that A.I. start-up valuations are soaring beyond that of 2021’s “everything bubble,” with investors trawling the rosters of companies like Google, Meta and OpenAI for A.I. experts who may have an itch to start their own company.
Sam Altman, the chief executive of OpenAI, left, and Kevin Scott, Microsoft’s chief technology officer. The funding race has heated up ever since ChatGPT, the chatbot made by OpenAI, went viral last year.Credit…Ruth Fremson/The New York Times
The funding race has heated up ever since ChatGPT, the chatbot made by OpenAI, went viral last year by showing the power of A.I. to generate its own tweets, emails, articles, answers and ideas. Even as investors expect last week’s failure of Silicon Valley Bank, an institution that many tech start-ups relied on, to cast a pall over start-up funding, there is still a mismatch between the number of opportunities in artificial intelligence and the money available to fund them.
That’s because of the scarcity of A.I. companies and the potential of the technology. With few experts in the field, and most of them working at a handful of big tech companies, only a few generative A.I. start-ups — such as Stability AI and Jasper — have broken out. Investors desperate for the next big thing are competing fiercely to invest in these companies, offering some A.I. entrepreneurs nine-figure valuations for little more than an idea and a résumé.
“We’re in that phase of the market where it’s, like, let 1,000 flowers bloom,” said Matt Turck, an investor who specializes in A.I. at the venture firm FirstMark. He added that the deal-making stood out in an otherwise dreary moment for tech marked by layoffs, cost-cutting and a drought of initial public offerings.
Andreessen Horowitz did not respond to a request to comment. Index Ventures declined to comment on Mobius’s funding.
Stability AI, the start-up behind the Stable Diffusion image generator, is one of the few generative A.I. companies to have broken out.Credit…Omar Marques/SOPA Images, via Getty Images
The blooming flowers include Dust, a start-up founded by former employees of OpenAI. Dust is nearing a $5 million funding round led by Sequoia Capital that will value it at $30 million to $40 million, two people with knowledge of the situation said. The round was competitive, with term sheets offering valuations as high as twice that, one of the people said.
Perplexity AI, a start-up created by former employees of OpenAI, Google and Meta, is raising $20 million to $25 million, led by NEA, that values the company at about $150 million, two people familiar with the situation said. And LangChain, a start-up working on software that helps other companies incorporate A.I. into their products, has raised funding from Benchmark, a person with knowledge of the matter said.
Those follow the $13 billion that OpenAI raised from Microsoft, including $10 billion in January, and $300 million raised this year by Anthropic, another A.I. start-up.
Dust and LangChain declined to comment.Various aspects of the funding rounds were earlier reported by Business Insider, The Information and Newcomer.
At Y Combinator, the start-up incubator, at least 50 of the 218 companies in the current program are working on generative A.I., according to a tally taken by Truewind, an A.I. bookkeeping start-up that is part of the program. Alex Lee, Truewind’s chief executive, said ChatGPT had helped investors, potential customers and potential employees understand the possibilities of the technology.
“Before, you’d go in and say, ‘We’re doing something with A.I.,’ and it’s hard to picture exactly what that looks like,” Mr. Lee said. “Now, they say, ‘Oh, I’ve played with ChaptGPT, and I can imagine how I could use this in my world.’”
He declined to comment on his company’s fund-raising ahead of Y Combinator’s demo day, when companies pitch investors, in April.
Even though more mature A.I. start-ups have already raised large sums, they can’t afford to ignore the latest overtures from investors, said Mike Volpi, an investor at Index Ventures who sits on the board of the A.I. start-up Cohere.
Mike Volpi, an investor at Index Ventures who sits on the board of the A.I. start-up Cohere.Credit…Christopher Goodney/Bloomberg
That’s partly because A.I. technologies like ChatGPT, which learn by analyzing vast amounts of digital data, require a lot of computing power, which is expensive. Mr. Volpi estimated that start-ups needed at least $500 million to develop their own large language model, the technology that underpins ChatGPT.
At a tech conference in Los Angeles organized by the investment firm Upfront Ventures this month, A.I. was inescapable. The event began with a goofy video skit about a venture capitalist giving a ChatGPT-generated speech. It quickly went off the rails, with the investor confidently regurgitating incorrect information from the bot as the punchline.
Speakers — including Al Gore; Marc Benioff, the chief executive of Salesforce; and V Pappas, a top executive at TikTok — weighed in from there, mostly by hailing the technology’s transformative potential. A panel of A.I. specialists said they welcomed the sudden attention, with Phil Blunsom, head of science at Cohere, noting that he has worked on language modeling for 20 years and that, until recently, “absolutely no one was interested in it.”
Some investors debated whether A.I. start-ups would get run over by more established players with deeper pockets. Given the steep computing costs, some said big players like Microsoft and Google’s parent company, Alphabet, had too much of an advantage. Google bought the A.I. lab DeepMind, which is developing a wide range of technologies as an Alphabet subsidiary, in 2014 for $650 million.
Last week, Salesforce announced a $250 million fund for investing in generative A.I. start-ups, alongside new investments in Cohere, Anthropic and others.
“There are a few times in technology where you really see a generational leap forward with revolutionary technology,” said John Somorjai, who leads Salesforce’s venture investments. “These companies are the next trillion-dollar opportunities in software.”
Sam Lessin, a venture capitalist at Slow Ventures, said he didn’t think the tech advances in A.I. translated to opportunities for start-ups. The best way to invest in A.I., Mr. Lessin said, is to buy the publicly traded stocks of big tech companies.
“The absolute vast majority of the spoils will go to the incumbents,” he said.