Remember when everyone was convinced AI would destroy the world? Now they’re convinced it’ll destroy their portfolios. Plot twist: they’re probably wrong on both counts.
Last week, the market threw a tantrum over an AI selloff. Broadcom got blamed for kicking it off, but here’s the thing—the company just posted nearly 50% revenue growth, 80% semiconductor revenue growth, and a $30 billion backlog. That’s not weakness; that’s a company drowning in demand. If that’s a red flag, I’m a flamingo.
So what actually spooked everyone? Three things, really. First, Iran tensions pushing oil toward $110-$120, which could reignite inflation. Second, political noise that shouldn’t matter until 2028. Third, the creeping sense that the market got too euphoric. Fair concerns, maybe. But zoom out and they look way more manageable than the panic suggests.
Here’s where it gets interesting: the spending keeps accelerating. China’s dropping nearly $300 billion over five years on AI data centers. Nebius is investing 1.7 billion euros in the U.K. AMD’s committing 2 billion pounds. SK Telecom is building a gigawatt-scale AI cloud in South Korea. Meanwhile, OpenAI just raised $122 billion and filed for an IPO. SpaceX is raising roughly $75 billion. This isn’t a slowdown—it’s a floodgate opening.
The Uber-Anthropic token budget story is hilarious if you think about it. Everyone’s treating it like proof AI isn’t working. But that’s like saying a kid’s new toy is useless because he played with it constantly until his parents set limits. Companies aren’t abandoning AI; they’re moving from “max out the tokens” to “budget the tokens.” That’s institutionalization. That’s maturity. That’s bullish.
Now for the real centerpiece: we’re looking at roughly $5 trillion in new market cap hitting public markets in a single year. SpaceX near $1.75 trillion, OpenAI and Anthropic each tracking toward $1.5 trillion. A lot of people see that as a top signal. History disagrees. The dot-com era’s biggest IPOs came in 1998-1999, and smaller companies trickled through afterward. The bulls open the gates first.
Here’s a contrarian take: last week’s strong jobs report (170,000+ added) got people claiming the AI labor apocalypse was overblown. But Challenger, Gray & Christmas data tells a different story—nearly 100,000 job cuts in May, the largest May figure since 2020, with AI cited in 40% of them. AI-driven cuts have already hit 88,000 this year, about 60% more than all of 2025 with five months to go. The disruption is real and accelerating.
Where does all that fresh IPO capital go? Straight back into compute. More networking, more memory, more chips, more cooling, more power. The cycle feeds itself.
Bottom line: the selloff got the fundamentals wrong. The data says the AI boom is accelerating, not peaking. When SpaceX clears the IPO gates, expect the summer of AI to resume.
