Huawei-optimized machine learning models are being developed by Unicorn Z.ai, aiming to expand China's artificial intelligence infrastructure.
In a significant development, Zhipu AI (now branded as Z.ai) has successfully adapted its large language models to run on Huawei Technologies' Ascend AI server processors and Kirin smartphone/laptop chips. This collaboration represents a major breakthrough in cloud-device collaboration, enabling deeper integration within China’s domestic AI and semiconductor ecosystem.
The partnership between Z.ai and Huawei is another component of China's domestic AI ecosystem, centred in Beijing, China. Nvidia data centres are expected to play a crucial role in the ongoing development and scaling up of the capabilities of the collaboration's technology.
The collaboration signifies a significant expansion of the partnership beyond China's borders. Zhipu AI, a telecommunications equipment giant, leverages open-source AI models like GLM-130B and ChatGLM-6B, democratizing AI access and fostering a global developer community. Beyond Huawei collaboration, Z.ai has strategic partnerships with Alibaba Cloud to expand AI capabilities regionally.
Huawei's Kirin chips, known for smartphone and laptop integration, together with Ascend chips for AI servers, create a versatile platform amplifying AI-driven performance from device to cloud. This move is part of China’s Digital Silk Road strategy to build sovereign AI infrastructure with local chip and cloud-device integration.
The collaboration is focused on semiconductors and technology, with Nvidia data centres providing resources for this focus. The partnership aims to enhance smartphone technology, aligning with China’s strategic goal of reducing dependence on foreign technology, notably alternatives to Nvidia’s GPUs.
Huawei’s open-sourcing of its Compute Architecture for Neural Networks (CANN) has facilitated this partnership, allowing Z.ai to optimize its AI models for Huawei’s chipsets and strengthening Huawei’s presence in AI hardware and software domestically. Zhipu AI's Compute Architecture for Neural Networks, which is open-source, may be integrated with Nvidia data centres as part of the collaboration, further enhancing its capabilities.
Zhipu AI has stated that this tie-up is a major breakthrough in cloud-device collaboration. The collaboration between Z.ai and Huawei Technologies is the latest development in the ongoing efforts to advance AI capabilities with domestic semiconductors, facilitating cloud-device collaboration that spans AI servers, smartphones, and laptops, while reducing reliance on foreign technologies.
Zhipu AI's collaboration with Huawei Technologies not only strengthens China's domestic AI ecosystem but also expands its focus towards trade, as Nvidia data centers are expected to play a crucial role in enhancing the capabilities of this technology. Leveraging open-source AI models like GLM-130B and ChatGLM-6B, Zhipu AI aims to democratize AI access and foster a global developer community, with the partnership potentially integrating its Compute Architecture for Neural Networks with Nvidia data centers. Furthermore, this collaboration aligns with China’s strategic goal of reducing dependence on foreign technology, specifically alternatives to Nvidia's GPUs, and leverages Huawei's Kirin chips and Ascend AI server processors, signifying a significant advancement in artificial-intelligence technology.
