Breakingviews: Nvidia’s AI mini-bets set an example not to follow

The stock-market hype around artificial intelligence turns out to be contagious. SoundHound, a company that makes AI-powered voice assistants for cars and customer-service lines, got a 76% share price boost on Thursday thanks to a newly disclosed investment from $1.8 trillion chipmaker Nvidia. It’s sensible for Nvidia boss Jensen Huang to spread his bets, even if some prove to be duds. But those hoping to ride his company’s coattails in pursuit of AI riches can’t say the same.

Until this endorsement from what recently became the United States’ third-most valuable company, things didn’t look so great for SoundHound. Its shares were trading at less than one-fourth of the price at which it went public in April 2022 through a merger with a special-purpose acquisition company. As is commonplace among SPACs, SoundHound laid out projections ahead of its debut that now look unattainable. Its investor presentation forecast $110 million in revenue and $84 million in gross profit in 2023. While SoundHound hasn’t yet reported its full-year figures, analysts polled by LSEG are expecting less than half those levels.

Sure, more hands-on support and capital from the colossal Nvidia, which SoundHound named as a strategic investor before its public-market debut, could help get it closer to its goals. Already SoundHound says that one-fourth of all automotive units produced globally use its voice technology, and it recently added restaurant chains Jersey Mike’s and White Castle to its client roster. So the upside could be big.

And if it isn’t? Nvidia, armed with over $18 billion in cash as of last October, won’t much care. In this week’s filing alone, it disclosed much larger stakes in biotechnology firm Recursion Pharmaceuticals and fellow chipmaker Arm. After making 20 different investments in AI companies last year alone, more than even technology giants Alphabet and Microsoft, Nvidia will barely feel the sting of one bet gone bad. Investors with shallower pockets, however, should be more cautious before jumping aboard the AI hype-by-proxy train.