New York, March 23, 2026, 06:17 EDT.
- Early Monday saw Nvidia, Broadcom, Microsoft, AMD, and Micron all trading in the red, with oil prices breaking above $100 a barrel and investors trimming expectations for Federal Reserve rate cuts.
- OpenAI has been out pitching private-equity investors for new funding. On the other side, Elon Musk announced plans for fresh chip factories in Austin through Tesla and SpaceX—another sign that the AI build-out keeps picking up speed.
U.S. stocks tied to AI slipped again out of the gate Monday. Nvidia dropped 3.1% in early trading, with Broadcom off 2.8%. Microsoft and AMD each fell 1.9%, and Micron sank 4.8%.
Sellers moved quickly as the mood shifted against tech. By 4:43 a.m. ET, Nasdaq 100 E-minis showed a 0.72% drop. U.S. crude futures surged 3%, clearing the $100 mark. After last week’s energy shock, traders stopped pricing in any Fed easing for this year.
The stakes are high, with AI leaders still commanding valuations that assume ongoing high spending and ready cash. But Goldman Sachs reports hedge funds turned net sellers of tech last week, despite the sector’s continuing dominance in the market’s growth narrative.
OpenAI on Monday sweetened its pitch to private-equity shops, dangling a minimum 17.5% return alongside early access to its latest models. The move ratchets up pressure in its enterprise battle with Anthropic. “There’s a big race to lock in as much enterprise,” said Matt Kropp from Boston Consulting Group. Reuters
Elon Musk announced a day earlier that SpaceX and Tesla plan to set up two advanced chip factories in Austin—one to supply Tesla’s cars and humanoid robots, the other aimed at powering AI data centers in orbit. “We either build the Terafab or we don’t have the chips,” Musk said. Reuters
Nvidia remains the linchpin here. Just last week, the company laid out a jaw-dropping projection: its Blackwell and Rubin chips could together unlock at least $1 trillion in revenue by 2027. Amazon Web Services, for its part, is set to snap up 1 million of Nvidia’s GPUs over that span. Nvidia isn’t stopping there—it’s ramping up efforts around inference, the real-time engine behind AI responses.
Broadcom stands out as the key large competitor. Earlier this month, the company told investors it sees AI chip sales possibly reaching $100 billion by 2027, with custom AI chips—known as ASICs—picking up momentum. These are processors tailored for specific tasks, unlike general-purpose chips. According to Reuters, Broadcom’s projects with OpenAI and Anthropic are nearing the scale of Nvidia and AMD’s recent deals.
The risks aren’t quieting down. Ed Yardeni, head of Yardeni Research, flagged a possible dilemma for central banks: if oil and gas prices hold at these levels, policymakers might be forced to juggle the idea of rate cuts with the threat of hikes—a backdrop that tends to weigh on stocks priced for big future profits.
Investors haven’t been shy about voicing concerns over ballooning build-out costs. Shares of Micron dropped roughly 5% last Thursday, after the company bumped up its fiscal 2026 capital spending target by $5 billion, now topping $25 billion—even though it had just delivered a solid quarter. JonesTrading’s Mike O’Rourke pointed out that this only underscored expectations: the supply squeeze in memory is likely short-lived, and fresh capacity is on the way.
The long-run spending push isn’t off the table. Back in February, Bridgewater put out numbers: Alphabet, Amazon, Meta and Microsoft are slated to shell out around $650 billion on AI infrastructure this year, a jump from $410 billion penciled in for 2025. Still, Greg Jensen cautioned the boom had moved into what he called a “more dangerous phase,” with demand for computing power still outstripping supply. Reuters
Monday shapes up as more of a macro temperature check than any clear reset in the AI narrative. Right now, the trade is stuck needing both sustained demand and continued access to cheap funding—a tricky balance, especially now that oil has punched back above $100. Investors are left eyeing the bill for the next batch of chips, new data centers, and model launches.
