For months, seemingly everyone with any authority has been talking about the possibility of a stock-market bubble driven by record spending on AI. I say “possibility” because, whether or not we’re in a bubble—and what kind it might be—has been the subject of vigorous debate.

Once you listen to enough interviews or read enough of our deep coverage on the subject, you notice there are really only two issues at the heart of the debate.

The first: How are we going to build all the data centers and related physical stuff required to make AI happen, even if funding is no object? Let’s not forget that anytime you hear of supply-chain issues, chip shortages, energy price hikes or other constraints on raw materials, there’s a direct relationship between that thing and the systems that train and deploy AI.

The second: Who is going to pay enough for the resulting AI products or achieve enough of a productivity boost to justify all this increased spending? Is there an AI tool worth you spending, say, $35 more a month to subscribe?

I put both questions to a half dozen top analysts, all of whom had recently authored deep dives on these subjects. Usually, trying to get consensus from a group like this is like herding cats. Unusually, this time their answers to my questions all more or less converged.

For a detailed look at what they’re thinking now, and how they arrived at their conclusions, check out this week’s Keywords column, which I worked on with WSJ data visualization maestro Nate Rattner.