NEW YORK, March 27, 2026, 08:09 EDT.
Nvidia, after dropping 4.16% to $171.24 on Thursday, faces more pressure as Friday’s session approaches. The chip giant’s market cap is still around $4.53 trillion. U.S. stock index futures remained in the red ahead of the open.
Nvidia now stands in for AI spending across the market, so the latest swing is significant. On Thursday, a wave of selling hit, dragging the Nasdaq down 10.7% from its October 29 peak—a clear correction. Add in a spike in oil prices and traders ditching hopes for Fed rate cuts this year, and the mood turned fast.
Sources told Reuters Huawei’s 950PR chip—designed as a domestic rival to Nvidia—performed well in customer tests, with ByteDance and Alibaba preparing to place orders. Mass production starts next month, and broader shipments are slated for the second half.
The next battleground is inference—the point where AI models actually respond to questions or perform tasks instantly. Nvidia, for its part, projected earlier this month that its AI chip revenue might hit $1 trillion or more through 2027. CEO Jensen Huang, speaking to investors, declared: “the inference inflection has arrived.” Reuters
The race isn’t just in China anymore. Arm told Reuters this week its AGI CPU might hit $15 billion in annual revenue within five years. That sent Arm shares up 20% on Wednesday, while AMD leapt more than 5% as investors wagered that next-gen AI—capable of managing multistep jobs with little human input—will demand more CPUs.
Nvidia’s numbers remain impossible to ignore. Back in February, the chipmaker reported a 73% jump in quarterly revenue to $68.1 billion, with full-year revenue up 65% to $215.9 billion. CEO Jensen Huang described “Enterprise adoption of agents is skyrocketing” and said customers are scrambling to pour money into AI compute. NVIDIA Investor Relations
Yet, analysts keep flagging potential challenges for Nvidia as inference picks up and clients shift toward developing their own chips. “Nvidia is definitely going to see more competition compared to a year ago,” KinNgai Chan at Summit Insights Group said to Reuters this month. Even so, he noted Nvidia continues to command “close to over 90% market share” in both training and inference right now. Reuters
Policy risk could hit first. Reuters reported that four Chinese universities—including a pair with ties to the People’s Liberation Army—have acquired Super Micro servers equipped with restricted Nvidia chips. Two U.S. senators this week pressed Commerce Secretary Howard Lutnick to look into halting export licenses for advanced Nvidia systems destined for China or routed through Southeast Asia. According to Jacob Feldgoise at Georgetown’s Center for Security and Emerging Technology, these transactions might support improvements to China’s weapons design, planning, and autonomous platforms.
There’s also a more immediate issue. According to a Reuters report on Thursday, U.S. utilities and grid operators are now pushing data centers to scale back electricity usage in periods of peak demand. PJM is warning of possible supply gaps as soon as next year. Meanwhile, Nvidia rolled out a partnership with Emerald AI aimed at shifting power draw when the grid is under pressure. So for investors, it comes down to a more basic question: how quickly can all this AI infrastructure translate into actual, running compute?
On Wall Street, the focus has shifted. It’s less about whether AI demand is real—everyone’s on board there—and more about how long Nvidia can keep its edge as AI infrastructure expands. Following February’s numbers, Jacob Bourne at eMarketer told Reuters that ROI questions are heating up for enterprises, especially with hyperscalers now targeting at least $630 billion in capital spending by 2026.
