TAIPEI, June 2, 2026, 19:03 (UTC+8)
- Arm’s CEO said demand for its new AI data-center CPU is beating the company’s initial forecast.
- ByteDance and Oracle are customers, another sign that CPUs are getting pulled further into the AI buildout.
- Arm shares finished at an all-time high as Nvidia’s fresh Arm-based PC chip put more pressure on Intel, AMD and Qualcomm.
Arm Holdings could hit its $15 billion annual revenue goal for AI chips sooner than it expected, CEO Rene Haas said, as demand for data center central processors is picking up. Haas told Bloomberg Television he was “very confident” about the target, according to Seeking Alpha. Reuters also reported demand for Arm’s chips is higher than eight weeks ago. Seeking Alpha
Timing is in play. Arm wants to get away from its old model of licensing chip designs and taking royalties. The shift could mean a larger cut of AI infrastructure spending if big cloud and internet firms start buying Arm’s finished chips in bulk. Arm reported in May that fiscal 2026 revenue hit $4.92 billion. A $15 billion-a-year chip business would be a big move, not a side bet.
At Computex in Taipei, Haas said that ByteDance and Oracle are using Arm’s AGI CPUs. A CPU, or central processing unit, runs most tasks inside a computer. In AI servers, CPUs are paired with GPUs, which handle parallel math.
Arm is calling its chip the AGI CPU, targeting agentic AI — software that does more on its own without much user direction. The company said this is its first production-ready silicon built just for that space and flagged customer demand over fiscal 2027 and 2028 topping $2 billion, more than two times what it put out during the launch in March.
Arm shares jumped 15.73% to $408.85 at the close on Monday, according to Google Finance, and were set to open higher Tuesday with a premarket indication at $413.50. Nvidia’s launch of RTX Spark, a PC superchip built on Arm’s design, fueled expectations that Arm might pick up share from Intel and AMD in the PC and server markets.
Nvidia introduced RTX Spark, which combines a Blackwell RTX GPU, a 20-core Grace CPU and as much as 128GB of unified memory. It’s aimed at local AI agents, content creation and gaming. CEO Jensen Huang said, “The PC is being reinvented,” describing it as a reinvention of the PC. NVIDIA Newsroom
Arm is in a tricky spot. Nvidia counts as both a customer and partner, but also a competitor. At the same time, Nvidia’s moves are bringing more market focus to Arm-based computing. Qualcomm is moving on Windows-on-Arm PCs too. Intel and AMD are still big names in x86, with strong holds in corporate PCs and servers.
Barron’s quoted Mizuho’s Vijay Rakesh saying the move isn’t a winner-take-all. Rakesh keeps Arm and AMD at Outperform, Intel at Neutral, and thinks AI demand will prop up both Arm and x86 chips instead of taking out either side fast.
Retail sentiment is getting a boost as well. Jim Cramer said Arm could still go higher and connected its recent gains to showing up with Nvidia’s Huang at Computex. Earlier, Cramer said AI agents would need “a ton of advanced CPUs,” a trend he said could be good for Intel, AMD and Arm. Insider Monkey
Risks remain. Arm needs to turn demand into chips shipped, lock in capacity, and keep its longtime licensees from getting nervous. After May earnings, reports flagged supply limits, despite projections that new chip sales would double by early 2028.
Export controls create more hurdles. Haas told Reuters that stopping the export of AI-ready CPUs to China would be tough for Washington, given their wide use. “CPUs are kind of like oil relative to the application space,” he said. He said regulators “would have to limit everything” to effectively target these chips. Reuters
Right now, Arm is trading more like an AI infrastructure name than the usual royalty play. The question for the market is if Arm can keep investors focused on real orders and revenue growth, not just on a hot stock price.
