Chinese Giant Alibaba Now Has an American Rival, Mutual Claims of Being the Most Advanced

Jakarta, CNBC Indonesia - Chinese technology giant Alibaba has introduced its next-generation Xuantie C950 5-nanometre processor. The AI chip is claimed to be the most advanced in the world.

The 3.2 GHz server chip is built with open-source RISC-V chip architecture. Alibaba describes its new chip as the highest-performing in the world and three times faster than its predecessor, the Xuantie C920.

“The open standard nature of RISC-V allows chip designers to customise instruction sets and accelerate specific AI workloads with low or no licensing costs,” the company’s blog post stated, quoted from Reuters, Wednesday (25/3/2026).

“This is important for the development of AI agents,” Alibaba added.

The launch of the new chip signals Alibaba’s shift towards the AI market. This rapidly growing industry in recent years is believed to be able to increase the company’s revenue.

Alibaba’s Xuantie series itself focuses on high-performance cloud systems and AI agents.

Additionally, Alibaba launched an AI platform called Wukong last week. The platform optimises AI agent workflows, at a time when many in Chinese society are adopting OpenClaw.

The platform is believed to be able to automate complex business organisations for small and medium-sized enterprises.

Alibaba has also reorganised several teams focused on AI earlier this month. The new team is under the Alibaba Token Hub to focus on building AI work platforms for companies.

Previously, Nvidia announced its latest generation chip named Vera Rubin. CEO Jensen Huang claimed that Vera Rubin is capable of significantly multiplying artificial intelligence (AI) performance.

The chip is said to provide AI computing capabilities up to five times greater than the previous generation, particularly for chatbots and large-scale AI applications.

The Vera Rubin platform consists of six separate Nvidia chips. Its flagship server will carry 72 GPUs and 36 latest-generation CPUs. According to Huang, these chips can be assembled in large systems or “pods” containing more than 1,000 chips, enabling an increase in efficiency for creating “tokens”, the basic unit in AI systems, up to ten times.

“This is how we can deliver such a large performance leap, even though our transistor count is only 1.6 times higher,” Huang stated, quoted from Reuters, some time ago.