Just a few months ago, the AI trade was somewhat polarizing amongst retail investors. Markets spent much of March and early April getting punched in the face by tariff fears, recession chatter, and growing skepticism around whether the AI boom had simply gotten too euphoric, too expensive, and too detached from reality.

All major indices had taken a certified beating, and investors were simultaneously asking whether AI spending had become unsustainable. Then…April 1st hit, and the market basically hit a giant red “lol never mind” button. Since the late-March lows, the S&P 500 has surged roughly 16%, while the Nasdaq has exploded about 26%, dragging markets back toward record highs. And once again, the entire rally appears increasingly tied to AI.

At this point, the S&P 500 is starting to look less like a diversified index and more like a giant AI infrastructure fund wearing a trench coat. The top 10 companies in the index now account for nearly 40% of its total weight and roughly $25 trillion in combined market value. Nearly all of them — NVIDIA, Microsoft, Amazon, Alphabet, Apple, Meta, Broadcom, Tesla — are now either direct AI plays or aggressively repositioning themselves around AI infrastructure, automation, cloud computing, or machine learning.

And the money flowing into this thing is getting genuinely absurd. Semiconductor companies alone have added roughly $3.8 trillion in market capitalization over the past six weeks, according to The Wall Street Journal. Intel is up 239% this year, Sandisk has surged 558%, and leveraged semiconductor ETF SOXL is up roughly 1,200% over the past year. The richest companies in the world are now panic-buying computing power like civilization depends on it. Because, increasingly, they think it does.

Visual: WSJ

“Hyperscalers,” as some call them, are expected to spend something like $755 billion on capex this year alone as they race to build data centers, buy GPUs, secure memory chips, expand networking systems, and lock down electricity supply. BlackRock has estimated that AI-related infrastructure spending could eventually reach $7 trillion to $8 trillion globally over the coming years. At this point, comparisons to the Manhattan Project and wartime industrial mobilization are no longer sounding entirely unserious.

So yes, it’s well worth asking as an investor: Is this… a bubble, or a real thing? Then again, FOMO is real. As one retired investor interviewed by the Journal described the current environment: “the party is best about a half-hour before the police shut it down.” But unlike many past bubbles, this one comes with something awkwardly real underneath it: the profits. Some of these companies are generating real earnings growth so extreme that even after massive stock rallies, parts of the sector still don’t look especially expensive by traditional valuation metrics. That’s part of why the market keeps chasing it.