Huawei Launches AI DC Data Infrastructure Full-Stack Solution to Accelerate Industry Intelligence Leap

At the 2026 Huawei Innovation Data Infrastructure Forum held in Paris on May 21, under the theme "Data Awakening, Storage Leap," Huawei officially unveiled its AI DC (Artificial Intelligence Data Center) full-stack data infrastructure solution. The announcement was made by Yuan Yuan, Vice President of Huawei and President of the Data Storage Product Line, in a keynote speech aimed at accelerating the intelligent transformation of industries.

Yuan Yuan emphasized that AI is fundamentally reshaping enterprise operations. The proliferation of AI agents is turning them into a new form of digital productivity, gradually becoming regular "digital employees" in enterprises. Meanwhile, as AI applications deepen, enterprise token consumption is experiencing explosive growth. To accelerate AI implementation, enterprises need to evolve their existing IT architectures toward AI DC data infrastructure, focusing on key areas such as data lakes, knowledge and memory platforms, computing power, models, agent frameworks, and data resilience.

The AI DC solution comprises five core components. The AI Data Lake leverages the OceanStor Pacific all-flash distributed storage, which offers industry-leading high capacity density of 11PB per 2U, enabling optimal total cost of ownership for storing massive datasets. The DME Omni-Dataverse creates a unified data space, allowing multimodal and cross-site data to be ingested into the lake in real time with global visibility and manageability. It also supports billion-scale, thousand-dimensional vector data retrieval in seconds, ensuring high-quality data aggregation and supply.

For knowledge and memory, Huawei introduced the industry's first context memory storage (CMS) supporting heterogeneous computing for ultra-large-scale inference clusters. CMS enables KV semantic direct access or semantic offloading using dedicated DPUs, expandable to a PB-level shared KV Cache pool, reducing first-token inference latency by 90%. For enterprise AI inference, Huawei pioneered the "3+1" AI data platform, integrating a knowledge base with over 95% retrieval accuracy, KV Cache acceleration, and an evolving memory base, all managed by UCM technology to improve inference accuracy by 30%.

Model engineering and resource scheduling are handled by Huawei ModelEngine, which provides out-of-the-box model capabilities and a model gateway, supporting zero-code adaptation for new models and one-click deployment. Combined with fine-grained computing resource slicing and intelligent scheduling, it achieves up to 1:10 XPU card slicing, enabling "one card for multiple uses" and significantly improving resource utilization.

The Agent framework, powered by Huawei ModelEngine Nexent, allows users to generate agents directly through natural language interaction, drastically lowering the development barrier and shortening the agent deployment cycle by 80%. It also supports automatic optimization of skills, prompts, and memory, enabling agents to continuously evolve and become more intelligent over time.

To address potential data security risks across agents, models, platforms, and infrastructure, the solution includes a data resilience platform. It provides end-to-end protection against abuse, poisoning, tampering, and ransomware, safeguarding AI data assets comprehensively.

"AI brings infinite opportunities to the IT industry," Yuan Yuan stated. "The second half of AI lies in data. Huawei Data Storage will persist in technological innovation, continuously accumulate industry-specific AI deployment experience, and work with the entire industry to help customers accelerate toward an intelligent future."