Huawei unveils an AI Data Platform designed to accelerate inference, improve retrieval accuracy and help enterprises turn AI models into real business value.
Huawei introduced a new AI Data Platform at Mobile World Congress 2026, aiming to help enterprises bridge the gap between artificial intelligence models and practical business applications.
Announced by Yuan Yuan, president of Huawei’s Data Storage Product Line, the platform integrates technologies for knowledge generation and retrieval, inference acceleration, memory management and unified data caching. The company says the system is designed to address persistent challenges in enterprise AI deployment, including slow response times, hallucinations and limited throughput during inference.
Moving Beyond AI Training
While significant investment has gone into training large AI models, Huawei argues that the next stage of adoption depends on improving inference performance—the process by which models deliver real-time results.
The company’s new platform introduces a “3+1” architecture to strengthen data processing and memory capabilities.
Among the components are:
- Knowledge generation and retrieval, which converts multimodal data into structured knowledge through lossless parsing and token-level encoding, delivering retrieval accuracy above 95 percent.
- KV cache acceleration, which uses historical memory data to expand context windows and reduce redundant computation, cutting time to first token by as much as 90 percent.
- Memory extraction and recall, enabling models to store and reuse contextual knowledge over time.
- Unified Cache Manager (UCM), which manages memory data across multiple caching layers to coordinate knowledge bases, inference caches and memory banks.
Also Read: Cutting Through the Noise of SaaS Buying
Flexible Deployment Options
Huawei said the platform can be deployed in two ways.
An appliance mode built on the company’s OceanStor A800 system provides high-performance infrastructure for new AI deployments. An independent mode allows enterprises to upgrade existing environments by combining AI data engine nodes with OceanStor Dorado storage, enabling incremental adoption while protecting prior investments.
The approach is designed to allow organizations to scale AI workloads while integrating them into existing infrastructure.
“Huawei will continue to deepen technological innovation,” Yuan said during the launch. “With the AI Data Platform serving as a bridge, we aim to transform model capabilities into real business value.”
