Keynote Address at HUAWEI CONNECT 2024
At HUAWEI CONNECT 2024, Zhang Ping’an, who serves as the Executive Director of the Board of Huawei and the CEO of Huawei Cloud, presented an engaging keynote speech entitled “Thrive with the Cloud: Reshaping Industries with AI.” He emphasised that enterprises must harness the opportunities presented by the intelligent era and leverage AI to create competitive advantages, advocating for the adoption of an AI-native mindset.
Embracing AI for Competitive Advantage
Zhang outlined several essential steps for enterprises to thrive in this new landscape. Firstly, he urged companies to embrace AI proactively, unlock industry-specific scenarios, and develop robust enterprise AI platforms that align with their core operations. Notably, Huawei Cloud’s Pangu Models are already implemented across more than 30 industries and 400 different scenarios.
The Importance of AI Computing Power
Secondly, he highlighted the significance of AI computing power. Enterprises must establish AI-native cloud infrastructure tailored to their needs. Zhang officially introduced CloudMatrix, a resource pooling solution that connects CPUs, NPUs, DPUs, and memory. This innovation represents a shift from traditional monolithic computing to a more flexible matrix compute model, enabling enterprises to access substantial AI computing resources efficiently.
Building Knowledge-Centric Data Foundations
Thirdly, Zhang stressed that the quality of data is pivotal for the effectiveness of AI models. He called for the creation of knowledge-centric data foundations to enhance AI capabilities. Huawei Cloud has undergone a comprehensive upgrade of DataArts, offering clients AI-oriented and knowledge-centric data solutions. The enhanced features include convergence engines for AI and data, as well as services for data governance and application enablement.
Tailoring AI Models for Business Applications
Fourthly, he encouraged businesses to develop suitable AI models tailored to specific applications. Zhang cautioned against the misconception that larger models are universally superior, asserting that a single foundation model cannot fulfil all requirements. The Pangu Models 5.0 series offers a diverse range of sizes, with parameters varying from billions to trillions, effectively addressing nearly all business application needs.
Advancements in Multimodal Generation
Furthermore, the Pangu 5.0 models enhance spatiotemporal controllable generation (STCG) within multimodal generation scenarios. For instance, in autonomous driving, the Pangu model can create driving scenarios that accurately replicate real-world conditions. This capability allows automotive firms to train their autonomous systems more efficiently, generating videos that depict normal driving situations, unpredictable road conditions, and instances of aggressive driving.
Introducing the Mainframe-to-Cloud Solution
Huawei Cloud also unveiled the Mainframe-to-Cloud Solution, which aims to assist customers in developing new core systems on the cloud, boasting high availability, simplified operations and maintenance, and enhanced agility, achieving an impressive 99.999% financial-grade availability. Many banks across China have opted for Huawei’s expertise to establish their new cloud-based core systems.
Huawei’s Digital Transformation Methodology
Tao Jingwen, Huawei’s Director of the Board and President of the Quality, Business Process & IT Management Department, shared insights into Huawei’s digital transformation journey. He described a methodology encompassing three layers, five phases, and eight steps. The layers include redefining intelligent business, AI model development and delivery, and continuous optimisation of intelligent applications. The phases encompass identifying suitable scenarios, reshaping processes, transforming organisations, optimising corporate data, and adopting AI applications.
Shaping the Future with AI
Bruno Zhang, CTO of Huawei Cloud, provided an overview of how Huawei Cloud is harnessing AI to reshape data centres, infrastructure, and cloud services. He introduced the “1+N” Pangu assistant system, marking a significant advancement in cloud service interaction. This system, featuring the Pangu Doer, encompasses numerous use cases across product research and development, data analytics, security, and office collaboration, where Pangu models are trained using scenario-specific data to enhance cloud services and boost operational efficiency.
