Vertiv Sells the Picks and Shovels of the AI Boom. It's Priced Like the Gold.

---

First, the case for Vertiv, stated at full strength, because it is genuinely one of the most impressive operating stories in the market and a forensic piece that downplayed it would be dishonest. Vertiv makes the unglamorous but essential physical infrastructure that keeps data centers alive: power distribution, thermal management, and increasingly the liquid-cooling systems that the latest generation of AI chips require because they run far too hot for conventional air cooling. As the hyperscalers pour hundreds of billions of dollars into AI data centers, they must buy this equipment, and Vertiv is among the handful of companies with the scale, technology, and global service network to supply it at the level required.

The results read like a momentum trader's dream. In the first quarter of 2026, Vertiv grew net sales 30% to $2.65 billion, expanded its adjusted operating margin by 430 basis points to 20.8%, grew adjusted operating profit 64%, and lifted adjusted earnings per share 83%. Its backlog reached $15.0 billion, up 109% year over year. In the prior quarter its organic orders grew approximately 252% against the year before, with a book-to-bill ratio near 2.9 — meaning that for every dollar of product it shipped, it booked nearly three dollars of new orders. The company guides to roughly $13.25–13.75 billion of net sales for 2026 with organic growth of 27–29%. These are not the numbers of a sleepy industrial supplier. They are the numbers of a company sitting in the fastest-moving current in the entire economy.

This essay is not an argument that the demand is fake or the backlog illusory. Both are real. It is an argument about a single, decisive question that the spectacular numbers are designed to make you stop asking: what should you pay for them? Because the entire logic of a picks-and-shovels investment is that you pay less for the shovel-seller than for the speculative miner — and the market has done the opposite.

The whole point of picks and shovels is that they're cheaper

To see why the valuation matters so much, you have to remember why the picks-and-shovels strategy is wise in the first place, because the wisdom contains the warning. In the California gold rush, the miners faced binary, catastrophic risk: most claims yielded nothing, and a prospector could dig for a year and go home broke. The merchant who sold the miners their picks, shovels, tents, and denim collected from all of them — the lucky and the doomed alike — and bore none of the claim-specific risk. He didn't need to know which mine would hit. He just needed the digging to continue.

That diversification across diggers is exactly why the shovel-seller is supposed to be a safer, lower-multiple business than any individual miner. He has steadier demand, less binary risk, and therefore deserves a more modest valuation — the price of safety is that you don't get the explosive upside of a miner who strikes the mother lode. This is the bargain at the heart of every picks-and-shovels trade: you trade upside for safety, and you pay a lower multiple to reflect it.

Vertiv has broken that bargain. At roughly 50 times forward earnings — more than double the median multiple of the industrial-products sector it belongs to — the stock is priced not like a safe supplier collecting from all the diggers, but like a speculative miner that has struck gold and will keep striking it for years. The market has granted Vertiv the explosive-upside multiple of the AI bet itself, while the business retains the underlying character of a cyclical equipment maker. You are being asked to pay gold-rush prices for the shovel company, which means you have surrendered the entire advantage — the safety, the lower multiple, the margin of error — that made selling shovels smart in the first place. If you pay the miner's multiple for the shovel-maker, you have taken on the miner's risk without the structural protection that justified buying the shovel-maker at all.

"Diversified across diggers" hides a brutal concentration

The picks-and-shovels framing rests on an assumption that deserves direct scrutiny, because in Vertiv's case it is largely false: the assumption that the shovel-seller is selling to many independent diggers. The nineteenth-century merchant sold to thousands of unrelated prospectors whose fortunes were uncorrelated. Vertiv sells to a customer base that is anything but.

Roughly 75% of Vertiv's revenue comes from data-center customers, and at the leading edge that means hyperscalers and large colocation operators — a small group of enormous buyers. The order strength Vertiv is celebrating is concentrated precisely among the handful of companies making the single largest correlated bet in the history of capitalism: Amazon, Microsoft, Google, Meta, and a few others, collectively spending toward half a trillion dollars a year on AI infrastructure. That is not a diversified base of independent diggers. It is a few giants, all digging the same mine, all on the same thesis, all capable of pausing their spending at roughly the same time if that thesis wobbles. Vertiv's $15 billion backlog is not a fortress wall of a thousand small contracts; it is a relatively concentrated set of commitments from a few customers whose capital-spending plans rise and fall together.

This is the hidden fragility inside the picks-and-shovels story. The strategy is safe only when the diggers are many and uncorrelated. When the diggers are four or five hyperscalers executing the same AI build-out, the shovel-seller's "diversification" is an illusion — its revenue is a leveraged bet on the continuation of a single, correlated capital-spending cycle. And capital-spending cycles, however powerful, are the most cyclical phenomenon in all of business. They do not grow to the sky. They overshoot, they digest, they pause.

The backlog is a bet on someone else's capex

Consider the $15 billion backlog more carefully, because it is the single most-cited reason the bulls believe the stock is safe even at 50 times earnings, and it is more conditional than it looks. A backlog is a record of orders placed, not revenue banked. It converts to revenue only if the customers take delivery and pay — and in an industry where the customers are hyperscalers whose own boards are beginning to face hard questions about the return on their AI spending, the durability of that backlog is precisely as solid as those customers' continued conviction.

Recall the dynamic documented elsewhere in this series: Amazon's free cash flow collapsed 95% under the weight of its roughly $200 billion AI capex plan; the hyperscalers are spending at a pace that is straining even their mighty balance sheets; and a growing chorus is asking whether the AI revenue will arrive fast enough to justify the infrastructure. Vertiv's backlog is the mirror image of that spending — it is the hyperscalers' capex, viewed from the supplier's side of the invoice. Which means Vertiv's fortress backlog and the hyperscalers' increasingly strained capex are the same money, and any pause, deferral, or cancellation in the hyperscalers' build-out flows directly into Vertiv's order book. The backlog does not protect Vertiv from a capex slowdown. The backlog is the capex that a slowdown would reduce.

This is why a 50-times multiple on Vertiv is, in substance, a leveraged derivative of the bull case on AI infrastructure demand. If the hyperscalers keep spending at the current frantic pace for years, the backlog converts, the orders keep coming, and even the rich multiple may prove justified. If the spending merely decelerates — not collapses, just slows from a 250%-order-growth boom to something normal — then the orders that drove the stock to a gold-rush multiple reverse, and a cyclical priced for permanent hyper-growth re-rates the way cyclicals always do at the top: fast, and far.

The last time the picks-and-shovels crowd got crushed

There is a precedent for exactly this trade, and it is worth remembering because it ended badly for the people who believed the infrastructure suppliers were the safe way to play a technology boom. In the late 1990s, the consensus "smart money" position on the internet was not to buy the dot-coms — those were obviously speculative — but to buy the infrastructure that the internet would run on: the fiber-optic component makers, the networking- gear suppliers, the companies selling the physical guts of the buildout. This was the picks-and-shovels logic in its purest form: don't bet on which website wins, sell the plumbing to all of them.

It was a catastrophe. When the telecom and dot-com capital-spending boom reversed in 2000–2001, the infrastructure suppliers did not prove safer than the dot-coms. They crashed harder, and took longer to recover, because their revenue was a leveraged function of capital spending that had been pulled forward into a frenzy and then collapsed. Companies that had been celebrated for record backlogs and triple-digit order growth watched those backlogs evaporate into cancellations as their customers — themselves over-extended — slammed the brakes on spending. The "safe" infrastructure play turned out to be a high-beta bet on the durability of someone else's capex, dressed in the comforting language of picks and shovels, and when the capex stopped, the comfort vanished and the leverage was revealed. The investors who paid premium multiples for those suppliers at the peak, reasoning that the demand was real and the backlogs proved it, were the ones most thoroughly destroyed.

The parallel to Vertiv is not exact — Vertiv is more profitable, better managed, and serves a transition with more genuine staying power than much of the late-1990s fiber bubble. But the structure of the trade is identical: a richly valued infrastructure supplier whose order book is a leveraged derivative of a correlated customer-capex boom, valued by a market that has decided the boom is permanent. The history of that exact structure is not encouraging, and "this time the demand is real" was said loudly, and sincerely, in 1999 too.

The competition the margins will summon

There is one more pressure that a 50-times multiple ignores, and it is the most reliable force in all of capitalism: a 20.8% operating margin in a booming market is a magnet for competition. Vertiv's profitability and its order book are advertisements, broadcast to every well-capitalized industrial company on Earth, that there is enormous money to be made supplying AI data-center power and cooling. Those companies are responding. Electrical and power-management giants like Eaton and Schneider, specialized cooling firms, and a wave of liquid-cooling startups are all pushing into the same market Vertiv dominates today, drawn by exactly the margins that make Vertiv's stock so beloved.

In the near term, demand so exceeds supply that everyone can win and pricing holds. But that is the characteristic condition of the early phase of a capacity boom, and it does not last. As competitors add capacity and the supply-demand balance normalizes — which it always eventually does — the pricing power that underpins those 20.8% margins comes under pressure, and a business valued at twice its sector's multiple on the assumption of sustained premium margins is acutely exposed to even modest margin compression. The bull case implicitly assumes Vertiv holds both its growth and its margins against a rising tide of well-funded rivals chasing the same hyperscaler dollars. That is a great deal to assume, and the assumption is embedded in a price that allows no room for it to be wrong. Add the operational risk of ramping production fast enough to convert a $15 billion backlog without component bottlenecks or execution stumbles, and the list of things that must go right to justify the multiple grows longer still.

What the bulls get right

In fairness, the bull case is strong on its own terms, and several of its pillars are genuinely durable rather than merely momentum.

The liquid-cooling transition is real and structural. As AI chips grow hotter, air cooling becomes physically inadequate, and liquid cooling shifts from exotic to mandatory — a genuine technological transition that expands Vertiv's addressable market for years regardless of the exact pace of any single year's capex. One forecast has the data-center liquid-cooling market growing from about $4 billion in 2026 to nearly $28 billion by 2033, and Vertiv is positioned at the front of it, reinforced by acquisitions like ThermoKey and Strategic Thermal Labs that deepen its thermal capabilities. The company also has a real moat that pure-play upstarts lack: a global installed base and service network, the kind of operational scale that hyperscalers require and that takes years to replicate. And the orders are not vaporware — a 2.9 book-to-bill is a hard, audited number reflecting demand that has already been committed. Vertiv is a genuinely excellent company executing flawlessly into a genuine boom.

The honest synthesis is that none of this essay's skepticism is about Vertiv the business, which is superb. It is entirely about Vertiv the stock at this price. A wonderful company supplying a real, multi-year transition can still be a dangerous investment if you pay a speculative multiple for it at the peak of a capex cycle — because the multiple, not the business, is what determines your return from here. The bulls are right that Vertiv will likely sell a great deal more cooling equipment over the next decade. They are answering a question — is the demand real? — that is not the question the 50-times multiple poses.

The multiple is the whole trade

Strip everything else away and the Vertiv decision reduces to a single proposition: you are paying roughly twice the industrial-sector multiple for a company whose revenue is concentrated among a few hyperscalers executing a correlated, cyclical capital-spending boom that is currently running at a pace no one believes is the permanent state. Every part of that sentence is the risk. The premium multiple means there is no margin of safety — the good news is fully priced, so only better-than-expected news moves the stock up, while any disappointment moves it down hard. The customer concentration means the "diversified picks-and-shovels" safety is illusory. And the cyclicality of capex means the current order growth, by definition, cannot continue at its current rate, because nothing growing 250% a quarter does so for long.

The cruelest feature of cyclical stocks priced as secular growth stories is that they look cheapest exactly when they are most dangerous — at the peak, when trailing earnings are highest and the multiple on those peak earnings looks almost reasonable, right before the cycle turns and both the earnings and the multiple fall together, the dreaded double compression. Vertiv may be years from that turn, or quarters. But a 50-times multiple leaves no room for the turn to arrive even a little early, and capex cycles have a long history of arriving early.

The kicker

The merchants who got rich in the gold rush did so because they bought their inventory cheap and sold it dear to desperate miners, pocketing a steady margin while bearing none of the claim risk. The ones who went broke were the latecomers who paid peak prices for their shovels and tents right before the diggings played out, left holding expensive inventory no one needed. Vertiv is selling real shovels into a real rush, and the rush may run for years. But the market is no longer offering you the merchant's bargain — steady demand at a humble price. It is charging you the prospector's premium for the privilege of supplying the prospectors, which means you have taken on the gold's risk while the company keeps the shovel-maker's fundamentally cyclical nature. When the digging slows — not stops, just slows — you will discover which one you actually bought. Picks and shovels are a wonderful business at a fair price and a dangerous one at a gold-rush price, and the only thing that has changed about Vertiv is which of those two prices the market is now asking you to pay.

The backlog is fifteen billion dollars and the orders are up two hundred and fifty percent and the earnings are up eighty-three, and every one of those numbers is real, and none of them tells you what the shovels are worth. Somewhere a hyperscaler's board opens a spreadsheet and asks, quietly, whether the spending can really go on forever — and the answer to that question, not the size of the backlog, is the whole of the trade.