An AI bubble or a psychological domino? Why investors mustn’t confuse OpenAI’s health with that of the whole AI industry

It wasn’t, of course, and some of the initial shock losses were pared once OpenAI had struck a reassuring tone in a statement that said it was still “firing on all cylinders.”

Still, the knee-jerk reaction was an indication of how closely the fate of OpenAI is tied to perception of the AI industry’s future. Circular financing has made these conclusions understandable, but OpenAI’s execution struggles shouldn’t be allowed to speak for the sector as a whole.

After the Journal reported that OpenAI’s chief financial officer was worried its revenue growth couldn’t support its expansion plans, the shares of its partners tumbled. Oracle, which is building OpenAI’s data centers, fell about 3%. AMD, which has a deal to provide OpenAI with GPUs, was down 5.5%, while OpenAI’s chip-making partner, Broadcom, was down by more than 4%.

But the skittishness escaped containment. Intel doesn’t have any direct deals with OpenAI but was down as much as 5%, for example. Power companies linked to the AI buildout also fell. The big slate of tech earnings this week—with four main hyperscalers reporting on Wednesday—will now be considered in the context of OpenAI’s performance issues.

Investors would be wise to consider the struggles of CEO Sam Altman’s company as, first and foremost, an OpenAI problem. The company is not synonymous with the AI industry.

The OpenAI bubble was inflated thanks to the company’s first-mover advantage from the almost accidental success of ChatGPT. That launch, OpenAI’s first huge viral moment, made it the fastest-growing consumer tech product in history—100 million users within two months. Extraordinary sums of venture capital followed, and the company is now worth $852 billion. Quickly placed on its shoulders was the fate of the industry.

OpenAI isn’t a public company yet, but tech stocks are being assessed through its lens. What better way to assess the health and future prospects of the new industry than by looking at the ‘leading’ AI company?

The litmus test is outdated. OpenAI has changed. It has contended with internal drama, departure of key talent, complications in acquiring computing power and a competitive landscape that has all but eroded ChatGPT’s moat. Anthropic has out-executed OpenAI on enterprise use cases in deep-pocketed industries such as cybersecurity and law.

Bigger competitors, which OpenAI humbled at first by beating them to market, have regrouped. They are now turning the screw on their individual, formidable strengths that have been built up over decades—Google with its colossal consumer business, cloud infrastructure and in-house chips; Amazon with the same in different forms; Meta with its legacy businesses to lean on if things take longer than expected, as they invariably will.

OpenAI called the Journal’s report “ridiculous” and “clickbait.” There are reasons for optimism if you want to look for them: Codex, OpenAI’s answer to Anthropic’s Claude Code, is being well received, and on Wednesday OpenAI made further progress on broadening its enterprise customer base with a $50 billion deal with Amazon’s AWS.

Still, OpenAI’s turf has clearly been invaded, and that’s what’s driving narratives of stalled growth. That’s no grounds for panic.

ChatGPT’s daily active user share of the US chatbot market is 38.3%, according to Apptopia, down from 55.4% this time last year. Google’s Gemini has been the benefactor. So, too, has Anthropic’s Claude, which received a publicity boost in the wake of its battle with the Pentagon and the backlash caused by Altman’s cynical and opportunistic move to step in as an alternative for classified military use.

Consumers are not turning their backs on generative AI; they just have more choice, as do many of the companies whose shares were punished because of their links to OpenAI’s fortunes.

No doubt, any genuine pullback from OpenAI on computing spend would have widespread repercussions for those waiting on the hundreds of billions of dollars that have been promised.

Even so, if the AI revolution is real, and the promised productivity gains materialize, other companies will be more than eager to buy up what computing power is available. The so-called neocloud company CoreWeave got it right when, in a statement responding to the market’s nerves on Tuesday, it pointed out that “OpenAI is a terrific partner, but not our only one.”

And if the AI revolution isn’t real, well then that’s a different question, one we’re no closer to answering than we were at the start of the week.

The author is Bloomberg Opinion's US technology columnist.