According to CNBC, the selloff in AI infrastructure stocks continued sharply on Monday, hitting Broadcom, Oracle, and CoreWeave. This follows a rough week where both Broadcom and Oracle reported quarterly earnings that actually beat revenue forecasts and showed soaring AI demand. Still, investors sold off their shares. Matt Witheiler, head of late-stage growth at Wellington Management, commented that while every AI company says more compute means more revenue, the ROI needs to be there to justify the spending. He believes, so far, that return on investment is present. The three stocks are still up for the year, but the recent trend highlights growing Wall Street skepticism.
When Beating Earnings Isn’t Enough
Here’s the thing that’s really interesting. Broadcom and Oracle didn’t miss. They beat expectations and still got punished. That’s a huge signal. It tells you the market’s mood has shifted from “show me growth” to “show me profitable growth at this scale.” It’s one thing to sell AI chips or cloud capacity. It’s another to convince Wall Street that the trillions being poured into data centers and silicon will generate a proportional return. Basically, the easy money phase might be over. Now we’re in the “prove it” phase, and that’s a much harder game.
The Multi-Trillion Dollar ROI Question
So what’s the real worry? Look, building AI infrastructure is a capital-intensive nightmare. We’re talking about a global scramble for power, chips, and real estate. Companies like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs, see this demand firsthand in manufacturing and logistics, where compute meets the physical world. But on a macro scale, investors are starting to sweat. Will all these AI projects actually generate enough cash flow to cover their insane costs? Witheiler’s point about companies wanting more compute is valid, but it’s also a trap. Endless demand for a service that burns cash isn’t a business model; it’s a money pit. The market is suddenly asking if we’re building a sustainable industry or just the world’s most expensive science project.
What Does This Mean for the Trajectory?
I don’t think this is the end of the AI buildout. Not even close. The demand is too real. But it probably is the end of the “buy anything with AI in the name” rally. We’re going to see a brutal separation between the companies that are truly enabling profitable AI workloads and those just riding the hype wave. Infrastructure players will need to demonstrate clearer paths to margin expansion and free cash flow. And honestly, a little cooling off might be healthy. It forces discipline. The next few earnings calls won’t just be about revenue growth percentages; they’ll be deep dives into capital efficiency. Can these companies turn the AI gold rush into a lasting, profitable empire? That’s the multi-billion dollar question Wall Street is now asking.
