Amid the rapid development of China’s artificial intelligence industry, demand for AI chips continues to rise, positioning GPU computing power as a critical foundational infrastructure. The SHINDEV research team believes that against the backdrop of global supply-chain volatility and China’s ongoing push for technological self-reliance, demand for domestically produced GPUs is accelerating, creating significant growth opportunities for local vendors with strong technological and ecosystem capabilities.
The global GPU market has reached USD 44.8 billion, with China accounting for approximately 24.7% of total market size. Driven by AI training and inference, data center acceleration, and advanced graphics and visualization workloads, the general-purpose GPU segment is experiencing rapid expansion.
From an industry perspective, internet services, government, public services, telecommunications, finance, manufacturing, and education are among the largest sources of GPU demand. As large language models and AIGC applications gain traction, demand from domestic data centers, cloud service providers, and intelligent computing hubs continues to surge—opening a practical entry window for domestic GPU suppliers.
According to SHINDEV, understanding the localization pathway for GPUs requires a clear distinction among CPUs, GPUs, and GPGPUs:
CPUs serve as the core of computing systems, executing general-purpose instructions and system control tasks;
GPUs are full-function graphics processors capable of both rendering and general-purpose computation, widely used across PCs, workstations, gaming consoles, and mobile devices;
GPGPU refers to general-purpose parallel computing using GPUs, representing a subset of full GPU functionality.
SHINDEV emphasizes that transitioning from GPGPU to a full-featured GPU presents significant technical hurdles, particularly in completing comprehensive graphics rendering pipelines protected by strong intellectual property barriers. As such, long-term competition in GPUs extends well beyond raw compute performance to encompass rendering capability, software ecosystems, and IP ownership.
GPUs can be broadly categorized into mobile and desktop segments. The desktop segment includes server GPUs and PC GPUs: server GPUs primarily address compute acceleration and deep learning workloads, while PC GPUs balance integrated solutions for light use cases with discrete GPUs for performance-intensive tasks. Mobile GPUs typically adopt integrated designs to meet power and space constraints.
SHINDEV notes that while the number of domestic GPU vendors is increasing amid expanding applications, a clear and stable market structure has yet to emerge. Some domestic products have passed mainstream performance benchmarks, indicating that the industry remains in a phase of rapid growth and restructuring.
SHINDEV identifies graphics rendering capability and microarchitecture as the core determinants of GPU competitiveness:
Graphics IP licensing and core IP development are essential, as GPU IP encompasses critical intellectual property and design architectures. Vendors capable of absorbing licensed IP, innovating upon it, and building complete desktop-class GPU IP portfolios—alongside proprietary core IP—will enjoy stronger long-term advantages.
Advanced microarchitecture governs instruction execution efficiency and hardware resource allocation. As applications diversify, compatibility with major operating systems, CPU platforms, and AI frameworks will be central to building a robust GPU ecosystem.
Globally, the discrete GPU market remains dominated by a handful of international players. Under China’s self-reliance strategy, localization of critical IT components has become increasingly urgent. While progress has been made in CPUs, memory, and networking chips, GPUs remain heavily import-dependent, underscoring the necessity and urgency of domestic alternatives.
SHINDEV emphasizes that in government systems, state-owned enterprises, and data centers, GPU localization and ecosystem readiness are vital to ensuring data security and technological sovereignty.
Domestic GPUs are currently in the early adoption phase within the Xinchuang ecosystem, with gradual expansion expected into broader industry verticals. Xinchuang markets offer high policy certainty and stable demand, providing revenue visibility and supporting continuous R&D investment.
China’s “2+8+N” localization strategy defines a clear penetration roadmap. Early success in Xinchuang applications will lay the groundwork for domestic GPUs to enter consumer markets and scale deployment across industries.
In the era of digital economy and AI, computing power has become a core production factor. Ongoing data center expansion, cloud adoption, and rapid growth of AIGC applications are significantly boosting GPU demand. Meanwhile, the rise of edge computing is creating additional application scenarios for GPU acceleration.
SHINDEV believes that amid supply-chain uncertainty and external constraints, China’s push to establish a self-controlled AI chip supply system provides a favorable environment for domestic GPU expansion.
SHINDEV concludes that the domestic GPU market offers substantial long-term potential. In the near term, demand prioritizes availability, deployability, and stable supply. Over time, competition will center on:
Core IP ownership and microarchitecture capabilities;
Comprehensive software–hardware ecosystem compatibility;
Cost-effectiveness and scalable manufacturing and delivery.
As AI-driven computing cycles accelerate, domestic GPUs face a pivotal window of opportunity. Sustained innovation, ecosystem development, and strong engineering execution will be essential for establishing enduring competitive advantages.
