How ARM CPUs Are Transforming China's Server Market – An In‑Depth Industry Insight
This article examines the rise of ARM and other CPU architectures in China, outlines three paths to domestic CPU development, profiles key players such as Huawei, FeiTeng, and Amazon, and evaluates the emerging ARM server ecosystem and its competitive stance against x86.
Server CPU Architecture Landscape
Modern data centers evaluate CPU architectures based on performance, power efficiency, and scalability. Selecting the appropriate instruction set and process technology is critical for meeting workload demands.
Domestic CPU Development in China
China’s ARM‑based ecosystem has grown rapidly, driven by low‑power, multi‑core advantages and geopolitical pressures to replace imported silicon. Major domestic vendors include Huawei HiSilicon, Phytium (FeiTeng), Qualcomm, Cavium, and Amazon, while home‑grown CPU firms such as Huawei, Phytium, Tianjin Haiguang, and Loongson target niche markets in government, finance, and power sectors.
CPU Localization Paths
IP‑core licensing : Acquire a licensed core and integrate custom peripherals. Lower technical barrier but limited control over the ISA.
ISA licensing : Obtain rights to modify or shrink an existing instruction set, offering higher autonomy at increased design complexity.
Independent ISA development : Design a proprietary instruction set, providing maximum sovereignty but requiring extensive R&D resources.
Key Architectures and Domestic Players
x86 – Dominated by Intel and AMD; closed to external licensing. Tianjin Haiguang and Loongson attempt self‑development via IP‑core licensing.
ARM – Open architecture owned by Arm Ltd.; licenses are available for both IP cores and the ISA. Huawei Kunpeng and Phytium have secured 64‑bit ARMv8 (AArch64) licenses.
MIPS – Fully open‑sourced by Wave Computing in 2018; Loongson purchased a perpetual MIPS ISA license in 2008.
ALPHA – Legacy ISA from DEC, now discontinued; China’s ShenWei obtained the ISA for supercomputing and military use.
ARM Server Ecosystem
Support for major operating systems (Linux, Windows Server) and virtualization/container platforms (Docker, VMware, KVM, Kubernetes) is mature on ARM. GPU frameworks such as CUDA (via NVIDIA’s ARM support) and Java runtimes also run on ARM, making it the most viable challenger to x86 for server workloads.
FeiTeng ARM Server Processors
Phytium released its first ARM‑based server chip, the FT‑2000, in 2016. Key specifications:
Up to 64 cores
Clock speed 1.5 – 2.0 GHz
28 nm manufacturing process
Power envelope 100 W
Performance comparable to Intel Xeon E5 series
Product variants include:
FT‑1500A/16 – entry‑level server
FT‑2000+/64 – high‑end server
FT‑1500A/4 – desktop workstation
FT‑2000A/2 – embedded control
Huawei Kunpeng and Amazon Graviton ARM Servers
Huawei launched the second‑generation Kunpeng 920 in 2019. Specifications:
64 cores, ARMv8 (AArch64) ISA
2.6 GHz base frequency
7 nm process technology
~30 % lower power consumption than comparable x86 chips
Integrated PCIe 4.0, 8‑channel DDR4, and 100 Gbps RoCE networking
The Kunpeng family powers TaiShan servers, which provide a full stack of hardware and software partners for a self‑reliant ecosystem.
Amazon’s Graviton 2, released in December 2019 after the 2015 acquisition of Annapurna Labs, is also based on an ARMv8 design. It delivers a 45 % cost reduction versus the first‑generation Graviton while offering comparable performance for cloud workloads.
Both Huawei and Amazon continue to expand their ARM‑based server portfolios, integrating custom silicon with a broad ecosystem of hardware OEMs, operating systems, databases, and virtualization solutions to reduce reliance on foreign technology.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
Architects' Tech Alliance
Sharing project experiences, insights into cutting-edge architectures, focusing on cloud computing, microservices, big data, hyper-convergence, storage, data protection, artificial intelligence, industry practices and solutions.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
