In September 2024, Jim Covello became Wall Street’s most recognizable AI skeptic. While Silicon Valley executives, venture capitalists, and technology analysts competed to make increasingly ambitious claims about artificial intelligence, Covello was asking a far less fashionable question: Where are the profits?

His critique was straightforward. The AI industry was spending hundreds of billions of dollars building infrastructure, data centers, chips, and models without demonstrating that customers would ultimately generate enough economic value to justify the investment. The concern was not technological. It was economic. And at the time, it was one of the most intellectually serious critiques of the AI boom. Nearly two years later, Covello appears more skeptical than ever.

In a recent Goldman Sachs discussion, he argued that the economics of AI may actually be more questionable today than when he first raised concerns in 2024. The amount of money being spent has increased. The infrastructure buildout has accelerated. And the commercial returns necessary to justify that spending remain elusive.

If anything, Covello’s central argument has become more pessimistic, not less. Which raises a fascinating question: Why is Goldman Sachs simultaneously publishing increasingly detailed research on how investors should profit from the very AI spending cycle Covello continues to question?

That is where the story becomes interesting. Not because Jim Covello changed his mind. But because he appears to have positioned himself to win regardless of which side of the AI debate ultimately proves correct.

If AI disappoints, Covello can point to years of warnings. He questioned the economics. He questioned the monetization. He questioned whether the industry’s spending was rational. History will record that he saw the risks long before many of his peers.

But if AI succeeds, Goldman Sachs has already provided investors with a roadmap. The firm’s research increasingly identifies the beneficiaries of the AI buildout:

• the hyperscalers,
• the infrastructure providers,
• the networking companies,
• the optical suppliers,
• the companies selling picks and shovels into the AI gold rush.

In that scenario, Goldman participated in the upside while preserving the ability to claim skepticism throughout the cycle. The result is an unusually comfortable position: the skepticism protects against being wrong, the investment thesis protects against missing out. Heads, the skeptic wins. Tails, the investor wins.

To be fair, there is a legitimate defense of this approach. An analyst can believe an industry is overvalued while still identifying individual companies that stand to benefit from the spending. An investor can question whether a gold rush is rational while still profiting from selling shovels. Those positions are not inherently inconsistent. But they do create a tension that deserves scrutiny.

Because Covello’s recent public commentary increasingly sounds more bearish than the investment positioning emerging from the broader Goldman Sachs research machine. The economics remain questionable. The profits remain uncertain. The spending continues. Yet the investment recommendations continue as well. At some point, investors are entitled to ask a simple question: If the economics are still broken, what exactly are we investing in?

The answer appears to be momentum. Not necessarily technological momentum. Capital momentum. Money continues flowing into AI infrastructure. Markets continue rewarding AI-linked spending. And Wall Street continues searching for ways to participate.

That may ultimately prove to be the correct strategy. But it is not quite the same thing as conviction. It is something closer to optionality. And optionality is one of the most valuable assets on Wall Street. Especially when it allows you to be right no matter what happens next.