A Changing of the Guard in Profitability
We are witnessing what amounts to a major changing of the guard in the technology sector. The era of the historically super-profitable big-tech franchise is giving way to a structural decline in profitability, and the cause is straightforward: capital expenditure at the current scale will eventually do that to any company, no matter how dominant. When you commit the kind of dollars that the AI buildout demands, the math eventually catches up, and even the most celebrated names begin to look strained.
Not every firm in this race actually has the checkbook to keep writing the checks. Some have the cash flow to absorb the cost of staying in the game indefinitely; others are stretching to keep up. Meta sits firmly in the second camp. The wave of layoffs we are seeing across the less-profitable members of big tech is a tell — a signal that the emperors do not always have the clothes the market assumes they do, particularly when measured against the balance sheet and the ability to keep spending at these elevated levels. That is precisely why a stock like Meta is vulnerable to the kind of selling pressure that turned the NASDAQ from green to red on a recent earnings morning.
The Hidden Side of the Balance Sheet
Part of the reason the strain is hard to see is that companies are getting creative about hiding it. Oracle, for example, carries roughly $250 billion in off-balance-sheet debt. Meta has effectively invented an entirely new category of off-balance-sheet obligation: the "residual value guarantee," in which the company commits to a data-center investment but also accepts the downside if the project does not work out. These are new mechanisms designed, in effect, to keep observers from realizing how much these firms are stretching to keep pace with the poor underlying economics of the AI buildout.
And the punchline is that, despite the staggering infrastructure spend, we still do not have particularly good data underpinning these AI models. Ask any model why it lies so frequently and it will happily list seven thoughtful reasons. The race is going to last a long time, and Oracle, Meta, and Amazon — the names currently in it — are likely to struggle to stay in it.
Apple and Meta: A Study in Opposites
The contrast between Apple and Meta has become one of the most instructive in the market. Meta is spending and spending. Apple is sitting back, watching the horse race, and waiting to play kingmaker once a winner emerges. Both approaches are revealing.
Apple can afford patience because of the strength of its network and services. Whenever a useful capability matures, Apple can plug it in and monetize it immediately without having to invent it in-house. It also has an enormous cash hoard and a continuous river of cash flow that will let it pay whatever it must to acquire what it eventually needs. Meta, by contrast, is the weakest sister in the group, which explains why its leadership is throwing haymakers — whether that means buying up scarce talent at premium prices or pouring money into data centers and compute. Meta is doing what it must to stay in the club. Apple is enjoying the spoils of a wide moat and a deeply profitable business model, and it can comfortably do exactly what it is doing.
The Incestuous Web Behind the Buildout
Behind the headline names, a less-discussed set of beneficiaries quietly collect the checks: former Bitcoin miners pivoting to compute, power companies, and nuclear plays, all riding the wake of what could fairly be called exorbitant spending by the hyperscalers. The risk is that the entire ecosystem has become uncomfortably interconnected. There is a great deal of incestuous investing going on — borrowing from one firm to pay another, chip vendors taking equity stakes in companies that then become customers, deals layered on top of deals — and the residual-value guarantees on data-center buildouts are part of the same pattern.
If even one major participant turns out to be unable to cash the checks it has been writing, the structure can unravel quickly. With on the order of $2 trillion being deployed into AI investment, this is by definition a high-stakes game, and many firms — including those not in the headlines — would feel the fallout if any of the main players had to push back from the table. Worth noting too: Oracle is funding its push largely on credit rather than free cash flow, which only adds to the fragility.
Where the Value Sits Outside the Hype
The interesting opportunities now sit well away from the names that dominate the AI conversation. A handful of stocks are priced as if their profits will permanently decline, and that pessimism looks misplaced.
- NPLX is an attractive "middleman" in the energy industry. Energy demand is structurally going up — partly because data centers will require enormous amounts of power, and partly because as standards of living rise globally, energy consumption inevitably follows. That trajectory looks inescapable, yet NPLX trades as though its profits will decline.
- CVR Partners (UAN) sits in the fertilizer space, where there is a genuine structural limit on supply. Margins are very high, profitability is strong, and yet the stock price implies a permanent profit decline.
- Autoliv (ALV) supplies airbags, seat belts, and other automotive safety features. Whether vehicles are being driven by humans or by software, safety only ever moves in one direction — no one ever brags about being less safe than they used to be — which makes it a quietly counter-cyclical opportunity.
The common thread across these names is valuation. After moving away from tech for some time — with the exception of a handful of cheap semiconductor names that happened to work out — the discipline is the same: do not chase what is expensive. The stocks worth owning are the ones priced for decline when the underlying business is anything but.
The Takeaway
The AI capex race is reshaping the fundamental profitability of big tech, exposing balance-sheet engineering that had previously gone unscrutinized, and concentrating risk in a tightly interlinked group of firms and their dependents. The companies with genuine cash flow and patience — Apple chief among them — will end up owning the spoils. The ones spending to stay in the club will continue to suffer on days when reality reasserts itself. And meanwhile, the more durable opportunities for investors may sit not in the AI narrative at all, but in the unfashionable corners of the market where prices already assume the worst.