Key Takeaways:
- CPU market projected to grow 40%+ annually to $2.2 trillion by 2030
- Nvidia, Qualcomm, Arm and others challenge Intel-AMD duopoly
- China's Hygon expected to capture 35% of domestic server CPU market by 2028
返回
AI inference boom drives CPU market toward $2.2 trillion by 2030
The CPU market, long considered a mature duopoly, is being reshaped by AI inference workloads that demand a radically different balance of computing power.
The global CPU market is projected to grow at more than 40 percent annually through 2030, reaching as much as $2.2 trillion, as AI inference and agent workloads shift computing demand away from graphics processors toward central processors.
"CPU attach rates are skyrocketing. You simply cannot buy CPUs — they are all sold out," Tony Pialis, who leads Qualcomm Inc.'s new data center business, told investors in New York last month.
Bernstein Research estimates the data center CPU market will expand from $370 billion in 2025 to $2.23 trillion by 2030. Bank of America's forecast is more conservative at $1.7 trillion but still represents nearly fivefold growth, implying a 37 percent compound annual rate. The shift is driven by a changing CPU-to-GPU ratio: early AI training workloads used one CPU for every eight GPUs, but inference and agent-based tasks now require ratios approaching one-to-two or even one-to-one, according to industry executives.
The expansion has drawn a wave of new entrants into a market long dominated by Intel Corp. and Advanced Micro Devices Inc. Nvidia Corp., the GPU giant, has begun selling its Vera CPU as a standalone product — Chief Executive Officer Jensen Huang said the CPU business alone could generate as much as $200 billion in annual revenue. Arm Holdings Plc, backed by SoftBank Group Corp., released its first physical chip — a data center server CPU — and targets $150 billion in annual CPU revenue by 2030.
New Entrants Reshape a Duopoly
Qualcomm Inc., MediaTek Inc., Google, Amazon.com Inc. and Microsoft Corp. are all developing or expanding CPU offerings, many based on Arm architecture rather than the x86 standard that Intel and AMD have championed for decades. Arm-based designs offer power efficiency advantages — critical in data centers where GPU clusters already consume enormous amounts of electricity.
"Arm-based CPUs from Nvidia and Arm itself have an edge in energy efficiency, which matters when GPUs are already power-hungry," said Brady Wang, associate director at Counterpoint Research. "Intel and AMD CPUs, on the other hand, have historically excelled at handling complex, latency-sensitive tasks."
TechInsights estimates Arm-based data center CPUs will capture about 20 percent of the global market by 2026, even as the overall market expands. Arm told Nikkei Asia that its technology now powers more than 50 percent of CPUs deployed by the world's leading hyperscale data center operators.
Intel, which still holds more than 50 percent of the data center CPU market, is seeing a resurgence in demand. "I've had several calls in the past few weeks asking for more CPUs," Chief Executive Officer Lip-Bu Tan said at the Computex trade show in Taipei last month. AMD CEO Lisa Su projected the CPU market will grow at least 35 percent over the next five years — a stark contrast to the 3 percent to 4 percent growth rates of the past.
China's Homegrown Push Gains Traction
Chinese chip makers are also capitalizing on the CPU boom. Hygon Information Technology Co., which uses x86 architecture, is expanding its market share as Chinese cloud providers accelerate domestic procurement. Bernstein estimates Hygon's share of China's server CPU market will rise from about 20 percent this year to roughly 35 percent by 2028, though foreign suppliers are still expected to hold the majority.
"Hygon will benefit from strong x86 CPU demand and gain further share in the Chinese market," said Bernstein analyst David Dai, citing product improvements and better interoperability with domestic AI accelerators.
Other domestic players are also expanding. Shanghai Zhaoxin Semiconductor Co. is preparing for a Shanghai STAR Market listing, while Huawei Technologies Co. continues developing its Kunpeng server CPU as part of a broader push toward self-sufficient computing. Loongson Technology, once seen as China's best hope for a homegrown Intel rival, has posted steady revenue growth but remains unprofitable.
Because CPU computing power is relatively lower than GPUs, US export controls on CPUs are less stringent than those on advanced AI chips. That leaves the CPU market as one of the few segments where American chip makers can still compete in China, even as local rivals gain ground. Nvidia's Huang and AMD's Su have both acknowledged the importance and competitiveness of the Chinese market. Qualcomm CEO Cristiano Amon told Nikkei Asia the company plans to design chips specifically for Chinese data center customers that comply with US export rules.
For investors, the CPU renaissance creates both opportunity and disruption. Nvidia's push into standalone CPUs threatens Intel and AMD's traditional stronghold, while Arm's data center ambitions could reshape the processor value chain. Intel trades at a discount to semiconductor peers, and a sustained CPU demand recovery could narrow that gap. AMD's growth projections suggest it is well-positioned to capture share in both x86 and AI-adjacent workloads. The competitive landscape is shifting faster than at any point in the past two decades, and the winners will be determined by execution on production timelines and customer adoption over the next 12 to 18 months.
This article is for informational purposes only and does not constitute investment advice.
