Investor frenzy will overvalue AI tech start-ups, says early OpenAI backer

Stay informed with free updates

Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.

Artificial intelligence start-ups are overvalued and most will fail to make money, according to Vinod Khosla, an early OpenAI backer, in a warning to investors who are pouring billions into the buzzy sector.  

Since the launch of OpenAI’s ChatGPT chatbot a year ago, investors have rushed to put their money in AI start-ups, including rival chatbot makers Inflection, Anthropic and Cohere, sending valuations soaring.

“Most investments in AI today, venture investments, will lose money,” Khosla said this month at a technology conference in Laguna Beach, California.

Khosla, speaking to the Financial Times on the sidelines of the Wall Street Journal’s Tech Live event, drew a parallel between the hype around AI and last year’s frenzy of investment into cryptocurrency start-ups, including failed exchange FTX. 

Many later entrants are “investing because everybody else is investing, that is what’s happening in AI”, he said. This year, venture capitalists have invested $21.5bn into AI companies globally, compared with $5.1bn in all of 2022, according to PitchBook.

Many hope that bets on highly valued companies will pay out because others will invest later at an even greater valuation — a phenomenon known as the “greater fool theory”, Khosla told the FT. 

Khosla is confident AI will radically change the world. He believes AI has the potential to take on 80 per cent of the workload in 80 per cent of all human roles over the next two decades, and will create huge economic value.

Khosla Ventures was one of the first venture firms to bet on OpenAI, investing $50mn in the young company in early 2019 in a round valuing the start-up at $1bn, according to people with knowledge of that investment. OpenAI is now seeking a valuation of around $86bn. 

Khosla Ventures declined to comment on OpenAI’s current valuation, noting that OpenAI had not confirmed that target.

The valuations of rival AI companies have also soared this year. Anthropic was valued at $5bn in a deal earlier this year, before it raised a larger round from Amazon, while Cohere was valued at $2.1bn this summer and Inflection reportedly at $4bn. Fledgling businesses have also secured eye-catching sums. In June, French start-up Mistral AI raised €105mn in Europe’s largest ever seed round when the company was just a month old.

Khosla suggested he would sit out later-stage funding rounds where expectations risked exceeding the underlying value of the business. “We are very, very active but very selective,” he said. 

His firm’s other investments include online grocery delivery companies Instacart and DoorDash as well as financial technology companies Stripe, Block and Affirm.

“We invest in the fundamentals . . . which is different from investing in momentum, which is [what] most of the management community [does],” said Khosla. 

He also highlighted concerns that powerful AI technology could be used to undermine next year’s US presidential election. “There will be millions of bots interfering with our election next year . . . nation states — which means China mostly — influencing the election next year and attempting to make democracy dysfunctional,” he said. 

He lauded efforts by the Biden administration and Congress to staunch the flow of US capital and expertise to Chinese companies working on AI, semiconductors or quantum computing. 

President Joe Biden issued an executive order in August that would limit the ability of US venture capitalists to invest in those sectors. A congressional committee has been probing historical investments by prominent US firms that had a presence in China, including Sequoia Capital. 

The US should use all the tools at its disposal to win the AI race with China, including importing talented researchers from Russia, pulling in capital from the Middle East and creating an attractive regulatory environment, said Khosla.

Homegrown AI tools could provide free doctors or AI tutors to children around the world, he suggested. “That kind of ability to bring services to people who desperately need it will result in whether western values win . . . or China wins. I think that’s the key.”