AI Capital Cycle Reshapes Corporate Capex: Is the Hype Machine Running on Fumes?

There’s a certain poetry to watching the AI capital cycle go full Ouroboros. Hyperscalers pour $415 billion into data centers, chips, and the kind of infrastructure that makes the cloud look like a quaint ‘90s dial-up service, and Wall Street responds with a standing ovation, until someone asks what all this spending is actually buying. The latest Seeking Alpha report puts the number at $250 billion added to US GDP, a figure that looks impressive until you realize it’s less about productivity and more about a market narrative that’s become self-referential. The AI hype machine is now so loud that even President Trump, in his latest SOTU, couldn’t resist name-dropping stock market highs as if Nvidia’s capex budget is a leading economic indicator.

The facts are straightforward: capital expenditures by US tech hyperscalers have ballooned from $160 billion to $415 billion in just a few years, with AI as the ostensible driver. This is not just a rounding error in the GDP tables. It’s a full-blown investment supercycle, reminiscent of the shale boom or the dot-com fiber buildout. The difference this time is the speed. AI-related spending is now the single largest driver of incremental S&P 500 earnings growth, and it’s not just the usual suspects. Industrials, utilities, and even REITs are getting in on the action, chasing the digital gold rush.

But here’s the rub: the market’s faith in the AI capital cycle is bordering on religious. The S&P 500 is up 34% in 2025, and seven of the top ten contributors to the MSCI EM Index were AI-related names, accounting for over 40% of the return (Seeking Alpha, 2026-02-24). That’s not a rotation, that’s a regime change. The Nasdaq 100, however, has started to wobble, down 5% since January 28, while the broader S&P 500 is only off 2%. Under the hood, sector rotation is picking up steam, with money leaking from mega-cap tech into more cyclical names. The market is starting to ask whether the AI capex bonanza is sustainable, or if we’re just front-loading future returns into a single, glorious melt-up.

The macro backdrop is a study in contradictions. Inflation is sticky in Australia (WSJ, 2026-02-24), and central banks globally are still in a data-dependent holding pattern. US GDP is getting a sugar high from AI investment, but productivity gains remain elusive. The AI narrative is now so dominant that it’s crowding out every other macro theme. Even commodities, usually the canary in the coal mine for late-cycle excess, are flatlining ($DBC at $24.675, +0%). The market is pricing in a world where AI solves everything, from labor shortages to margin compression. But as any trader who lived through the fiber-optic bubble can tell you, capital cycles have a nasty habit of overshooting.

The real story here is about capital discipline, or the lack thereof. Hyperscalers are spending like drunken sailors, betting that AI will deliver the kind of productivity gains that justify a $415 billion capex tab. But history suggests that when everyone builds at once, returns on invested capital tend to collapse. The market is already sniffing this out. The Nasdaq’s underperformance relative to the S&P 500 is a warning shot. If AI spending starts to disappoint, or if the ROI on all this infrastructure turns out to be less than promised, the unwind could be brutal.

Strykr Watch

Technically, the AI trade is still intact, but the cracks are showing. The S&P 500 is flirting with key support at 4,900, while the Nasdaq 100 is struggling to reclaim 16,500. Breadth is deteriorating, with fewer stocks making new highs. The rotation into cyclicals is picking up, but the leadership is narrow. Watch for a break below 4,850 on the S&P 500 as a signal that the AI capital cycle trade is losing steam. Relative strength indicators are rolling over, and the 50-day moving average is now a battleground. If hyperscaler earnings disappoint next quarter, expect a swift repricing.

The risk is that the market’s faith in AI-driven capex is overextended. If the narrative cracks, the unwind could be sharp, with tech multiples compressing and capital-intensive names leading the way down. On the flip side, if productivity gains finally materialize, the upside could surprise even the bulls. But for now, the risk/reward is skewed to the downside.

Opportunities exist for nimble traders. Long cyclicals against tech on a pairs basis has started to work. Short the AI hype basket if the S&P 500 breaks 4,850. Look for mean reversion in utilities and REITs that have lagged the capex rally. If you’re a true believer, buy the dip in hyperscalers on a retest of the 50-day moving average, but keep stops tight.

Strykr Take

The AI capital cycle is the market’s favorite story, but every cycle ends. The risk is that we’re closer to the top than the bottom. Stay nimble, watch the technicals, and don’t drink the Kool-Aid. When everyone is building data centers, it’s time to ask who’s actually getting paid.

Strykr Pulse 58/100. The AI trade is tired, but not dead. Threat Level 3/5. Rotation risk is rising, and the unwind could be violent if the narrative cracks.