GE Vernova and Vertiv are giving us a glimpse into the future of the AI boom

There’s no shortage of CEOs willing to declare — with dead certainty — what the future holds.

Usually they’re of the Silicon Valley variety, laying out futuristic visions of humanoid robots and endless AI productivity gains.

That’s nice and all. But for the non-visionaries among us, the tendency of these guys (they’re mostly guys) to gloss over nuts-and-bolts business details makes their predictions less than bankable.

Do they have enough money to make their vision a reality? A sufficient strategy to take on incumbents and competitors? The right regulatory relationships to navigate hurdles?

Those tech visionaries can be tough to pin down on the details.

And then there’s what we got today: two earnings reports from the picks-and-shovel AI companies that actually support the massive build-outs. First, we got numbers from an honest-to-god metal-bending industrial company at the heart of the AI data center boom. Electrical equipment and gas turbine maker GE Vernova’s results provided a sturdy block of evidence that the AI boom has room to run.

Order backlogs at GE Vernova — one of three companies spun out of industrial icon GE just over two years ago — jumped 32% to a new high of $163.28 billion during the three months that ended March 31, underscoring the fact that the data center build-out could last for years.

“In the quarter, our Electrification segment booked $2.4 billion in equipment orders to support data centers, more than all of last year,” GE Vernova CEO Scott Strazik said in the company’s press release.

The surge in orders is why Chris Dendrinos, an analyst covering the company for RBC Capital Markets, thinks GE Vernova will be completely sold out of its production capacity for heavy-duty turbines through 2029 — and possibly into 2030 — by the end of this year.

“Backlogs are getting longer and prices are getting higher because the hyperscalers are willing to lock in the orders for three years out into the future,” Dendrinos said.  

To be sure, orders are not the same thing as actual sales.

And those warning about the risk of the AI boom going bust can point to previous episodes where panicky buyers doubled up orders of tough-to-get equipment from multiple suppliers, only to walk away when demand from their own customers softened. The result can be a sharp downturn in activity that leaves a glut of unprofitable inventory.

Dendrinos says GE Vernova’s order numbers should be taken seriously in part because they represent large amounts of money — up to 25% of the cost of a turbine — that utilities, hyperscalers, and other customers plonk down when they make firm commitments to buy. Such “firm” commitments accounted for 44% of the company’s equipment order book in Q1.

The other half of the orders (less than ironclad commitments known as a “slot reservations”) also require deposits from customers, he said.

What’s more, given the rush to secure power for data centers — GEV’s gas turbines are a key part of the on-site power plants data centers now often require — the company can mitigate the risk that when push comes to shove, buyers can’t or won’t meet their commitments.

“GEV’s got more demand than they have slots. They can can be choosy as to who they want to work with,” Dendrinos said. “They favor utility companies because utilities don’t default, or really high-quality large hyperscaler customers that are going to follow through.”

The run-up in GE Vernova shares suggests the market is convinced.

But Vertiv Holdings, another stock at the white-hot center of the data center boom, had a different experience.

It too reported better-than-expected results early Wednesday. But last quarter, the maker of power and cooling equipment for AI data centers told the market that it would stop offering quarterly backlog data, after reporting a massive 250% surge in remaining orders that sent its shares up sharply.

(Officials cited the lumpiness of such orders as a reason to cut the quarterly releases, though they will continue to release backlog data on an annual basis.)

Without that data, the market focused on Vertiv’s guidance for the coming quarter. Vertiv said it expects adjusted earnings per share of between $1.37 and $1.43 next quarter, mostly below the $1.43 consensus estimate. It also guided for net sales of $3.25 billion to $3.45 billion in Q2, compared to Wall Street’s call for $3.40 billion. The stock sold off, but after a run-up of 300% over the last year, it’s not the end of the world. Expectations were clearly pretty high.

So, what’s the lesson? Well, one takeaway may be that after the ride investors have had on some of these AI stocks, they’ll need almost constant reassurance that the boom still has legs.

GE Vernova’s report provided it; the stock went up. Vertiv didn’t. Shares fell.

Expect similar dynamics as we roll through the next few weeks of earnings.

After all, AI stocks are pretty important: almost all the S&P 500’s top performers this year are heavily reliant on the ongoing boom for both their revenues and the bullish sentiment that’s driven up their shares. That includes memory plays like Sandisk (the top performer in the S&P), Western Digital, and Seagate, as well as optical networking stocks like Lumentum, Ciena Corp., Corning, and Coherent.

In fact, Goldman Sachs analysts recently estimated that AI investment spending is likely to account for roughly 40% of all earnings growth for members of the S&P 500 this year.

So yeah, the outlook for that spending is kind of important. And any incremental glimpse of the future companies provide will be watched incredibly closely.