The wheels are starting to come off the AI infrastructure gravy train. Oracle, Broadcom, and CoreWeave shares tanked Monday as investors got cold feet about the massive debt loads these companies are piling on to finance data center buildouts. The sell-off raises a critical question that's now dominating Wall Street: do the actual returns from AI infrastructure justify the capital-intensive model? That anxiety crystallized after Oracle announced an additional $15 billion in capex for the current fiscal year.
U.S. tech stocks have been jittery lately, but the real bloodletting Monday was concentrated in one corner of the market: AI infrastructure. Oracle, Broadcom, and CoreWeave all got hammered as investors suddenly started questioning whether the race to build AI data centers actually makes financial sense.
The catalyst was pretty straightforward. Oracle announced Wednesday that it would need to raise capital expenditure by an additional $15 billion for its current fiscal year and significantly increase its lease commitments for data centers. The company is turning to debt to finance all that spending. That revelation triggered a market recalibration about what happens when companies chase growth at any cost.
Oracle took a 2.7% hit on Monday. CoreWeave, the upstart competing in the same AI data center space, dropped around 8%. Broadcom, which supplies chips and infrastructure to these companies, slipped about 5.6% on concerns about margin compression. The declines aren't catastrophic in isolation, but the unified message across all three was clear: the street is worried about balance sheets.
The question investors are now grappling with is whether this capital intensity is sustainable. Every single one of these infrastructure plays is borrowing at scale to deploy billions in computing hardware before actually knowing if the economics will work out. It's a classic venture capital mentality applied to publicly traded companies with quarterly earnings obligations. That mismatch between speculative deployment and fiscal accountability was always going to create tension.
Matt Witheiler, head of late-stage growth at Wellington Management, offered the bull case during an appearance on CNBC's "Money Movers" Monday. "It definitely requires the ROI [return on investment] to be there to keep funding this AI investment," he said. "From what we've seen so far that ROI is there." The optimistic reading is that every AI company on the planet is essentially saying the same thing: give us more compute, and we'll generate more revenue. That demand dynamics argument is hard to ignore.
