Nvidia isn’t a bubble, but is it a buy?

The recent plunge in Nvidia shares – it broke another record after shedding $406 billion in market capitalisation in a week – has prompted talk of a bubble bursting. Is it?
No, says a recent note by JPMorgan strategist Michael Cembalest. He characterises Nvidia’s financials as “phenomenal” and the “antithesis of the dot-com era”, saying Nvidia “bears no resemblance” to dot-com leaders such as Cisco.
Besides “high and rising” profit margins, Cembalest points to projections that Nvidia will maintain more than 90 per cent market share in AI chips for the next two years. Thus, even if shipments of Nvidia’s next-generation Blackwell chip are delayed, others are not positioned to take advantage.
So Cembalest is bullish? Not so fast. Whatever about the next two years, “competition is coming”. Competition partly explains why market leaders in past decades, such as GM, IBM, Altria, Cisco, GE and Exxon, underperformed in the future (Microsoft and Apple are exceptions in this regard).
Cembalest also cites MIT research suggesting AI’s boost to productivity will be much smaller than many believe, and asks why many ChatGPT users access it infrequently. Cembalest is himself an infrequent user, saying ChatGPT got half his questions right but the other half wrong in prior tasks. Its mistakes are unpredictable, necessitating he check every answer, and decidedly limiting any productivity benefits.
Similar doubts increasingly crop up in strategist reports. Is AI over-hyped? Will companies scale back spending on a technology that isn’t generating big revenues? What about competition? Will regulators challenge Nvidia’s near monopoly?
Nvidia isn’t a bubble, but that doesn’t mean it’s a buy. For now, investors are focused on the uncertainties.