August 18, 2026, (Inside AI) — Chinese cloud provider UCloud has deployed servers running Hygon‘s Tianxi series AI accelerators on its public-cloud platform. This marks the first large-scale public-cloud rollout of the domestic chip series, according to a report from Yicai.
The move gives UCloud customers access to locally developed computing hardware for AI workloads. Neither company has disclosed the initial deployment scale or the specific services supported. The announcement signals a shift in China’s cloud infrastructure strategy as domestic alternatives gain traction.
Hygon, a Shanghai-based chip designer, has positioned the Tianxi line as a competitor to imported accelerators from NVIDIA and AMD. The chips are built on x86 architecture and target inference and training tasks. UCloud’s adoption suggests that Chinese cloud providers are actively testing and deploying domestic silicon to reduce reliance on foreign technology.
The rollout comes amid tightening US export controls on advanced semiconductors. Chinese cloud companies have faced restrictions on acquiring high-end NVIDIA GPUs, pushing them toward domestic alternatives. Hygon’s Tianxi accelerators are seen as a viable option for certain AI workloads, though performance benchmarks remain limited.
Read: DeepSeek Plans 160,000-Chip Huawei Cluster in Inner Mongolia
Domestic chips gain ground in cloud infrastructure
UCloud is not the first Chinese cloud provider to explore domestic AI hardware. Alibaba Cloud and Huawei Cloud have integrated chips from Cambricon, Biren, and Huawei‘s Ascend series. However, UCloud’s deployment is notable for its scale and public-cloud availability.
Hygon’s Tianxi chips are designed to handle deep learning frameworks like TensorFlow and PyTorch. The company has partnered with software vendors to ensure compatibility. UCloud’s platform likely offers these accelerators as an alternative instance type for AI developers.
Industry analysts note that domestic chips still lag behind NVIDIA in raw performance and software maturity. But for many inference workloads, the gap is narrowing. Cost and supply chain security are driving adoption among Chinese enterprises.
What the rollout means for AI developers
For developers using UCloud, the Tianxi instances provide a new option for running AI models without relying on imported hardware. The pricing and performance details remain unclear, but early adopters may benefit from competitive rates as UCloud seeks to attract customers.
UCloud has been expanding its AI infrastructure offerings. The company operates data centers across China and Southeast Asia. Adding domestic accelerators aligns with its strategy to serve customers with data residency and compliance requirements.
The lack of disclosed scale raises questions about capacity. A large-scale rollout suggests significant investment, but UCloud may be testing demand before expanding. Hygon’s production capacity and yield rates are also factors to watch.
Read: China’s AI Giants Still Rely on Nvidia Chips Despite Self-Sufficiency Push
The partnership could influence other cloud providers to accelerate their domestic chip integrations. If UCloud’s deployment proves successful, it may set a precedent for broader adoption across the industry.
