Jay Goldberg stands alone with the only Nvidia sell rating as AI mania peaks

Jay Goldberg stands alone on Wall Street, holding the only sell rating on Nvidia while nearly everyone else keeps cheering the company higher.

Jay works from a cluttered home office in a three‑story Victorian near Haight‑Ashbury in San Francisco. According to him, the problem is simple. “There’s a lot more that can go wrong with Nvidia than can go right,” Goldberg said. Out of 80 analysts covering the company, 73 have it rated as a buy. Six call it a hold.

Only Jay says sell. Nvidia’s stock has jumped more than 3,000% since the beginning of 2020, making it the strongest performer in the S&P 500.

The demand for its graphics processing units, or GPUs, has helped fuel both the U.S. economy and market records. Yet Goldberg does not see this as a sign of stability. He sees too much blind confidence and too little evidence of sustainable return.

Going against Wall Street’s direction

Goldberg admits he does not mind being the outlier. “I’m probably a little cantankerous by nature, so I’m skeptical of all of the hype around AI right now,” Jay said. “This is not my first bubble.” He covers twelve companies in total.

Nvidia is the only one he rates as a sell. He has buy ratings on Apple, Netgear, Broadcom, and Arm Holdings. All of those companies are exposed to the AI boom, but Goldberg argues the real foundation of AI spending lies with six firms: Microsoft, Alphabet, Amazon, Meta, Oracle, and OpenAI.

Their spending has driven Nvidia to a valuation of about $4.5 trillion. The five publicly traded firms in that group are expected to spend nearly $400 billion in capital expenditures this year, up more than 67% from the prior year.

OpenAI has plans to spend more than $1 trillion. But with four of those firms reporting earnings soon, investors are asking what actual returns justify these costs. Jay compares the situation to the telecom buildout during the dot‑com bubble.

Cisco Systems soared on infrastructure spending back then, only to collapse when the traffic failed to appear. Cisco still has not regained its peak from 2000.

“That feels very strongly like the pattern we’re seeing now,” Goldberg said. “We’re going to build up all this AI stuff for what are largely psychological reasons. At some point, the spending will stop, and the whole thing will tumble down, and we’ll reset.”

Bulls push back, but questions remain

Since Jay initiated coverage on April 30, Nvidia shares have risen more than 70%. The average analyst target is now $220, about 18% above the recent close of $186.26. Bulls argue that demand is still in early stages.

Frank Lee at HSBC recently raised his target to $320, calling for broader adoption of AI accelerators.

But Jay questions where upside remains if Nvidia is already close to fully sold out. His target is $100, the lowest on the Street. He also raises concerns about electricity availability for new data centers and the leverage tied to their construction.

“Once you trace down where all these GPUs are going, you get into the weeds of the neoclouds and all these electricity and property deals that are taking place,” Goldberg said. “It’s easy to see how some obscure company fails and that cascades down the rest of the supply chain.” Jay keeps shelves of old phones, camcorders, and a Qualcomm server that never made it to market, a reminder of how fast technology leaders can turn into history.

Even bullish investors are starting to voice caution. Goldman Sachs CEO David Solomon compared current AI enthusiasm to the dot‑com era. A growing share of global fund managers now call AI stocks a bubble. When asked if there is an AI bubble, OpenAI CEO Sam Altman answered, “Yes.”