How Cloud HPC Is Redefining Data+AI: Insights from Alibaba Cloud’s VP

Amid the digital‑transformation wave, companies are seeking data‑driven growth, and the rapid evolution of AI is accelerating this shift. Wu Jiesheng, Vice President of Alibaba Cloud Intelligence Group and head of Elastic Computing and Storage product lines, highlighted at the 20th CCF HPC China conference that data has become an indispensable corporate asset and that cloud‑AI integration is emerging as a new development trend.

Diversified HPC Workloads

High‑performance computing (HPC) now faces increasingly varied workload demands, ranging from large‑model training and autonomous driving to life‑science research, industrial manufacturing, and semiconductor design. Wu categorises HPC loads by compute‑coupling and data intensity into three types: extreme‑coupling, tight‑coupling, and loose‑coupling.

Alibaba Cloud’s HPC Infrastructure

To address these needs, Alibaba Cloud has built a complete HPC infrastructure with specialised services:

Lingjun ZhiSuan – serves extreme‑tight‑coupling workloads such as large‑model training.

E‑HPC – supports tight‑coupling HPC tasks.

E‑HPC Instant – caters to loose‑coupling workloads.

Elastic Capability of Cloud HPC

Wu emphasised that the greatest advantage of Cloud HPC lies in its elasticity. By leveraging a pooled cloud resource pool and elastic scheduling, resources can be created and released on demand, improving utilisation and reducing customer costs.

Heterogeneous Computing and One‑Click Deployment

Cloud HPC also supports heterogeneous computing (GPU, FPGA) and offers one‑click deployment and automated management, providing flexible, high‑efficiency solutions for AI‑driven workloads.

CIPU Architecture

Alibaba Cloud’s self‑developed Cloud Infrastructure Processor (CIPU) integrates CPU, GPU, and accelerator capabilities, delivering differentiated performance for big data, HPC, and AI training. Since 2017, the architecture has evolved, with CIPU 2.0 delivering enhanced security, stability, and performance, including higher‑throughput elastic RDMA for E‑HPC.

Industry Applications

Examples of real‑world impact include:

Automotive manufacturers using E‑HPC for simulation and optimisation, achieving a 25% efficiency boost and significant cost savings.

Life‑science firms accelerating drug‑computing tasks with E‑HPC Instant, cutting costs to one‑third and speeding up new‑drug development.

AI start‑up “Moon’s Dark Side” leveraging Alibaba Cloud’s large‑scale, stable AI platform for model training and application expansion.

Future Outlook

Wu envisions a future where every enterprise becomes a “data + AI” company, with high‑performance computing providing the foundational compute power for AI, cloud computing, big data, IoT, and other emerging technologies. Alibaba Cloud aims to collaborate with industry players to continuously advance HPC‑driven AI applications.