The Rise of Domestic GPUs in China: IP Licensing, Imagination Technologies, and Market Dynamics

Domestic GPU development in China has surged, with companies such as Chip‑Dynamic Technology launching the "Fenghua 1" chip and Moore Thread achieving mass production of a full‑function GPU within 18 months.

Imagination Technologies, a UK‑based chip design firm known for its PowerVR IP used in Apple’s A‑series SoCs, has become a focal point of discussion, especially after its 2017 split with Apple and subsequent acquisition by the Chinese‑backed Canyon Bridge fund.

Chinese‑backed chip design companies

Imagination’s IP licensing model, which historically generated revenue from major customers like Intel, Samsung, and Huawei, faced a steep decline after Apple reduced its purchases, prompting a strategic pivot toward desktop, automotive, and high‑performance computing markets.

How IP licensing works

IP (Intellectual Property) in the semiconductor industry refers to reusable, verified circuit modules called IP cores, which can be soft (HDL source), firm (gate‑level netlist), or hard (mask layout) cores.

Suppliers such as ARM, Imagination, and VeriSilicon provide these cores, allowing chip designers to reduce cost and time‑to‑market, though the level of integration determines flexibility and required downstream work.

Complex GPU architectures, like Nvidia’s Hopper, integrate dozens of IP cores (e.g., shader engines, tensor units, ray‑tracing units) along with memory, video decode, power management, and I/O, making full‑chip integration challenging.

Why Chinese GPUs are emerging now

The convergence of AI‑driven chip startups, national “Xinchuang” initiatives promoting domestic alternatives, and rising demand from gaming, metaverse, and autonomous driving has accelerated GPU projects at companies such as Moore Thread, Wall‑Ray Technology, and Chip‑Dynamic.

Key factors include the adoption of Imagination’s PowerVR B‑series IP in Fenghua 1, the need for high‑performance graphics in AR/VR and digital‑human applications, and the broader market pressure from cryptocurrency mining and global chip shortages.

Despite progress, Chinese GPU firms still face hurdles such as securing comprehensive IP portfolios, achieving competitive performance, and building robust software ecosystems comparable to Nvidia’s CUDA or AMD’s ROCm.