Huawei Cloud Targets Enterprise AI Agents with New Hybrid Stack
Huawei Cloud has unveiled a new hybrid architecture designed to support the rapid rise of autonomous AI agents, promising to streamline complex business operations through unified infrastructure and data management.
At the Huawei Cloud Summit in Shanghai, the company presented a new hybrid cloud architecture tailored for the emerging era of AI agents. The system, known as Huawei Cloud Stack, aims to address the shifting dynamics of business execution where digital workers are increasingly collaborating with humans and other agents. The initiative focuses on integrating compute, data, and models into a cohesive platform rather than managing them as separate, siloed resources.
The company argues that traditional IT management is no longer sufficient for the demands of agentic AI. As these digital workers take on tasks in coding, customer service, and design, the way resources are provisioned must change. Instead of layer-by-layer assembly, the new approach delivers integrated capabilities organized around specific agent-driven tasks, aiming to reduce the complexity that often slows down enterprise adoption.
Shifting from Human-Driven to Agent-Driven Workflows
Antonony Gu, President of Huawei Hybrid Cloud, emphasized that AI agents are redefining the fundamental actors in business processes. The focus is moving away from humans manually using applications toward a model of human-agent and agent-to-agent collaboration. This shift requires a deterministic hybrid cloud architecture that can handle the continuous evolution of these digital employees.
The new architecture includes an AI-native capability hub that connects data, compute, and tokens to support this transition. By treating agents as a new class of digital employees, the platform seeks to provide the necessary infrastructure for them to operate at scale. This move is intended to help enterprises navigate the next three to five years of accelerated AI adoption.
Infrastructure Upgrades Boost Compute Efficiency
A key component of the new stack is the development of Agentic Infra, which supports the latest A5 SuperPoDs. By utilizing technologies such as NPU pooling and memory snapshots, the system claims to increase compute utilization from 30% to 70%. This improvement aims to make the infrastructure more agile and elastic, capable of handling both general-purpose and specialized AI workloads.
The platform also introduces an observable and governable operations framework for AI agents. This turns traditional cloud management platforms into intelligent hubs that can oversee the entire lifecycle of these digital workers. End-to-end protection is provided for infrastructure, models, and agents to ensure security and reliability in this new operational model.
Data and AI Integration Drives Productivity
Beyond infrastructure, the strategy focuses on unlocking the value of enterprise data. The platform offers end-to-end data lifecycle capabilities, including a decoupled storage and compute architecture. Tools like AI DataLake and DataArts are designed to manage multimodal data and support intelligent, data-driven decision-making, turning data into a new source of productivity.
Huawei Cloud is also investing in foundation models integrated with industry know-how to drive automated model iteration. This includes domain-specific models for vision-language processing, scientific computing, and prediction. The goal is to bring AI into production across various industries, enabling continuous evolution and optimization of business processes through these advanced tools.
