Inside the $1.4 Trillion AI Capex Supercycle Reshaping Wall Street
Hyperscaler spending is rewriting the S&P 500's earnings story. Here's where the next leg of the trade is hiding.
Hyperscaler spending is rewriting the S&P 500's earnings story. Here's where the next leg of the trade is hiding.

Illustration by IMF Alpha editorial · Reviewed by Marcus Hale
The four largest U.S. hyperscalers are on track to deploy more than $1.4 trillion in AI infrastructure by the end of 2026 — a capital cycle that now rivals the buildout of the U.S. interstate highway system in inflation-adjusted terms.
What's different this time isn't the size of the number. It's the velocity. Microsoft, Meta, Alphabet and Amazon have together raised AI-related capex guidance in seven of the last eight quarterly prints, with the marginal dollar increasingly flowing to custom silicon, liquid cooling and grid-scale power contracts rather than off-the-shelf GPUs.
For equity investors, that rotation matters. The first wave of the AI trade was a pure compute story, and Nvidia captured almost all of the upside. The second wave looks broader: networking, memory, power, real estate, and the long-tail of software companies that can actually monetize inference at scale.
We mapped the 38 names with the highest revenue beta to hyperscaler AI capex and stress-tested them against a 20% pullback in 2027 spending. Eleven of them still compound earnings double-digits in that scenario. Those are the names worth owning through the cycle.

Michael Burry, known for 'The Big Short,' has publicly voiced concerns about a potential AI sector bubble, drawing parallels to past market exuberances.

The round, led by a sovereign wealth fund, doubles the company's valuation in nine months.

Inference costs are collapsing faster than pricing. That changes the entire moat conversation.