Huawei Cloud: Revolutionizing AI Computation and Industry Solutions
Huawei Cloud: Revolutionizing AI Computation and Industry Solutions
Huawei Cloud is leading the forefront of artificial intelligence (AI) innovation, providing dynamic solutions tailored to various industries. At the HUAWEI CONNECT 2025 event, Zhang Ping'an, the Executive Director of Huawei's Board and the CEO of Huawei Cloud, presented the groundbreaking concept of "All Intelligence Empowering AI Pioneers for Industries." This initiative seeks to deliver unprecedented support for companies to harness AI's capabilities effectively.
In 2025, Huawei Cloud announced its latest service, AI Compute Service, powered by the already notable CloudMatrix384 super-node. This upgrade will dramatically enhance computational power from 384 GPUs to an astounding 8,192 GPUs, allowing for expansive cluster operations capable of managing between 500,000 to one million GPUs. This capability offers a robust AI computing resource, essential for businesses navigating the challenges of today's intelligence-driven landscape.
Additionally, Huawei Cloud introduced the Elastic Memory Service (EMS), an innovative storage solution that expands video RAM resources, thereby reducing delays during complex conversations in base models. This development significantly enhances user experience when utilizing AI applications.
Opposed to traditional data center structures, Huawei has implemented fully liquid-cooled AI data centers across provinces in Guizhou, Inner Mongolia, and Anhui in China. These facilities exhibit high-efficiency heat dissipation (up to 80 kW per cabinet) and achieve a remarkable Power Usage Effectiveness (PUE) of 1.1. This innovation ensures that businesses no longer need to remodel or build new data centers, merely requiring fiber optic connections to tap into high-performance AI compute capabilities and a comprehensive suite of cloud services.
Zhang Ping'an emphasized the capabilities of the AI Token Service that simplifies complex technological processes for users, ensuring that they harness AI's computing power in the most effective manner. The CloudMatrix384 facilitates total resource consolidation—computing, memory, and data storage—and transforms serial operations into parallel distributed tasks, significantly boosting the inference performance of systems. In various inference tasks, whether online, near-line, or offline, it delivers an inference performance that is approximately three to four times greater than traditional systems.
Going beyond infrastructure, Huawei Cloud is refining its Pangu models, tailoring them to industry-specific scenarios and collaborating with clients to tackle pressing challenges. This includes leveraging openPangu to promote best practices for AI training and inference, thereby equipping developers to maximize the AI computing power at their disposal. Alongside this, Huawei is developing a closed-source version of Pangu, with ongoing investments aimed at enhancing this model and supporting customers in creating bespoke industry-specific models.
Currently, Pangu models have been integrated into over 500 scenarios across more than 30 industries, including government, finance, manufacturing, healthcare, coal mining, metallurgy, rail transport, autonomous driving, and meteorology.
Innovating Beyond Endpoints
In 2025, Huawei Cloud launched the CloudRobo Embodied AI platform, which capitalizes on advanced algorithms and intelligent cloud logic to produce lighter and more efficient robots. This cloud intelligence transcends previous limitations, enabling broader applications of robotics in varied scenarios. To facilitate seamless communication between robots and cloud services, Huawei introduced the Robot to Cloud (R2C) protocol, and the inaugural 20 partners have officially adopted this standard.
Additionally, Huawei Cloud is committed to enhancing synergy between hardware and software through its ARM-powered Kunpeng Cloud services, noted for their efficiency, security, and reliability. Over the past year, the number of Kunpeng compute cores has surged from 9 million to 15 million—a 67% increase—and compatibility with over 25,000 applications has been achieved to ensure familiar user experience.
The GaussDB databases further bolster performance, featuring layered connectivity that supports multiple simultaneous read/write operations across nodes, liberating users from traditional architectures that were limited to primary nodes. With the ability to process 5.4 million transactions per minute within distributed setups, GaussDB enhances parallel transaction processing effectiveness significantly.
Lastly, Huawei Cloud’s distributed cloud solutions—including CloudOcean, CloudSea, CloudLake, and CloudPond—provide centralized services across diverse geographic areas, ensuring cohesive operations for clients.
Huawei Cloud has also unveiled Versatile, a corporate-grade agent platform allowing clients to create AI agents rapidly tailored to specific use cases with ease and efficiency. Users only need to prepare operational documents and flow diagrams, with agent generation achievable in just two steps, heralding a new era for AI development.
With various forums and roundtable events scheduled during HUAWEI CONNECT 2025, Huawei Cloud aims to showcase pioneering technologies and practices across cloud infrastructure, large models, databases, AI agents, and embodied AI, leading the charge for digital and intelligent transformation across industries. Huawei is ensuring that the latest technological advancements will help businesses navigate the complexities of the digital age effectively.
Topics Consumer Technology）
【About Using Articles】
You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.
【About Links】
Links are free to use.