A.I. Unicorns Have Good Reasons Not to Go Public
Unicorns, or startups valued at $1 billion or more, are hardly a rare sight in generative A.I. In fact, leading players like OpenAI and Anthropic are seeing their valuations soar past tens of billions of dollars, leading tech financiers to wonder if—and when—these A.I. darlings will go public. Experts say A.I. startups may hold off on considering IPOs because of unpredictable revenue prospects, generous private investments, and an evolving regulatory landscape amid a challenging market.
As of May 2024, 12 of the 28 startups that had reached the unicorn status this year were A.I. companies, according to BestBrokers. Perplexity AI, an A.I.-powered search engine, is reportedly in talks to raise $500 million in a round that could valued the just two-year-old company at $8 billion or more. The Jeff Bezos-backed startup is part of a growing batch of A.I. companies dominating the unicorn list, including the $80 billion OpenAI, the $30 billion Anthropic and the $7 billion Scale AI.
It’s typical for a private company to consider an IPO when it becomes unicorns. Going public has the potential to significantly increase a company’s access to capital by allowing it to sell equity to the public. IPOs can also boost brand awareness and give founders and investors an opportunity to cash out their shares. However, IPOs have slumped in recent years due to market headwinds like higher interest rates and inflation, leading to a backlog of IPO registrations.
In the first half of 2024, only 37 companies went public, generating $28.4 billion in exit value, according to Pitchbook. That’s significantly less than 2021 when 310 companies went public and generated $588.7 billion in IPO gains during that same time period.
Generative A.I. startups, in particular, are even more inclined to stay private compared to firms in other industries because of the rapid, unpredictable development of A.I. products, according to Aron Bohlig, founder and managing partner of the investment banking firm ComCap. A.I. companies, Bohlig said, don’t have “good forecasts” for how much revenue they’re generating, which could make it “very difficult” to have quarterly financial models that project the progress of their businesses to their shareholders even if they are making money.
“This is something that’s very specific to A.I.,” Bohlig told Observer. “I wouldn’t want to be the CFO of OpenAI and on a quarterly basis and need to have disclosures to investors and change my strategy every six or nine months because the industry is evolving so quickly.”
After all, generative A.I. unicorns have already raised substantial private capital that are comparable to post-IPO gains, reducing the urgency to go public, according to Ray Wu, managing partner at Alumni Ventures who oversees the VC firm’s A.I. fund. Startups have also striked lucrative partnerships with tech giants. Microsoft, for instance, has poured $13 billion into OpenAI, and Amazon and Google have invested $4 billion and $2 billion, respectively, in Anthropic, lowering the pressure to go public for additional capital.
“Strategic partnerships further reduce the need for an IPO, as these investors seek value beyond financial returns, including collaboration and shared expertise,” Joe Endoso, president of Linqto Capital, a fintech platform for pre-IPO investing, told Observer. “This dynamic enables A.I. companies to access substantial capital and strategic resources without the pressures of public markets.”
In addition, the evolving regulatory landscape around A.I. poses risks to an IPO. California’s SB1047, a vetoed A.I. safety bill that would’ve required companies in the state to conduct additional safety tests on their A.I. models if it passed, drew the ire of tech giants like Meta and OpenAI claiming it would stifle product development. Introducing stricter controls on A.I. development could make going public less attractive for those companies, Endoso said.
Legal concerns around data privacy may also delay an IPO. News Corp (NWSA), the media conglomerate that owns the New York Post and The Wall Street Journal, sued Perplexity AI over copyright infringement, alleging the startup trained its model on their publishers’ work without permission. OpenAI has been hit with similar lawsuits from The New York Times and the actress Scarlett Johannson.
As policies around these issues evolve, A.I. companies may be better off focusing on responsible A.I. development instead of short-term gains. After all, companies still need to address risks like the lack of transparency, bias and discrimination, and security risks before they can go to market, said Gayle Jennings O’Byrne, CEO of Wocstar Capital, a New York-based venture capital firm.
“Companies like OpenAI and Anthropic are deeply invested in ensuring their technologies are developed and deployed ethically and safely, which requires careful research and sometimes slower decision-making,” Alumni Venture’s Wu told Observer. “Remaining private allows them to prioritize these long-term goals without the constant pressure to deliver immediate financial results.”
Ultimately, the path to an IPO will depend on how the A.I. market evolves. Bohlig speculates there will be more IPO filings in 2025 from A.I. companies after a new U.S. president gets elected. New leadership means new regulations that could set the pace for when firms decide to go public. The end of rate hikes could also free up cash from fixed-rate investments like bonds into the stock market. But for an IPO to be successful, A.I. companies, at the very least, must show signs of profitability and be able to accurately project steady returns.
“When we start to see real value created as a result of real profitability and real customers and not oversubscribed, over-hyped funding rounds, we’ll start to see some exits,” Byrne said.