Huawei Cloud launches Agentic AI portal '智果园'

🔑 Enhanced Key Takeaways

•Huawei Cloud's '智果园' (ZhiGuoYuan) is a core component of its broader "Silicon-Based Black Soil" strategy, which aims to build a full-stack AI ecosystem encompassing domestic chips, software toolchains, cloud infrastructure, model training, and industry applications.
•The portal is built upon a new paradigm for Agentic AI infrastructure, focusing on creating efficient token factories, enabling continuous learning, integrating general-purpose and intelligent computing scheduling, and ensuring secure autonomous operations.
•Beyond general-purpose agents, '智果园' features four dedicated "Industry AI Dream Factories": Smart Healthcare (which includes the RuiPath large model developed with Shanghai Ruijin Hospital), Embodied Intelligence (showcasing CloudRobo, a full-stack embodied intelligence development platform), Intelligent Manufacturing, and Scientific Computing.
•Huawei Cloud is strategically differentiating itself by avoiding the prevailing "token price war" in the Chinese AI cloud market, instead prioritizing the enhancement of productivity and the development of robust Agentic Infra.
•The '智果园' portal is specifically designed for AI agents, not human users, offering a "Skill-based" and "CLI-based" approach that allows agents to control cloud services through simple commands and orchestrate complex workflows autonomously.

🛠️ Technical Deep Dive

Infrastructure: Utilizes CloudMatrix384 supernodes with MatrixLink interconnect to support hybrid compute and scale inference clusters.
MoE Parallelism: Employs expert parallelism in Mixture-of-Experts (MoE) models, enhancing single-PU throughput by 4–5 times.
Memory-Centric Storage: Features AMS Agentic Memory Storage, providing PB-level memory space with KV Cache layered pooling to reduce inference costs and overcome long-term memory bottlenecks for day-long agent tasks.
Unified Scheduling: Integrates CCE Volcano Next for unified scheduling across compute, AI, and storage resources, aiming for over 30% improvement in resource utilization.
AI Computing Cluster: The AICS Lingqu AI Computing Cluster supports a 100,000-card cluster scale with a total computing power of 200 EFLOPS, reducing token generation latency to under 10 milliseconds and achieving 5 million tokens per second throughput per thousand cards.
Runtime Foundation: Includes AgentSphere Secure Autonomous Runtime Foundation, enabling sub-100-millisecond startup and the creation of tens of thousands of batch agents per minute using lightweight sandboxes.
Model Service Platform: The upgraded ModelArts Next platform offers enhanced model matrices, semantic routing, RLaaS (Reinforcement Learning as a Service), and hardware-level trusted environments, supporting over 15 SOTA models and providing various model routing strategies.
AI Security: Incorporates a comprehensive AI security protection system across cloud, network, edge, model, and application layers, including auditing, rollback, model version control, sandboxing, and anomaly detection.

🔮 Future ImplicationsAI analysis grounded in cited sources

Huawei Cloud's emphasis on industry-specific "AI Dream Factories" will significantly accelerate the practical adoption of agentic AI in specialized enterprise sectors.

By providing tailored solutions and collaborating with industry partners, such as the Smart Healthcare zone with Shanghai Ruijin Hospital, Huawei Cloud addresses domain-specific challenges and fosters measurable productivity gains, moving beyond general-purpose AI applications.

The "Silicon-Based Black Soil" strategy, which prioritizes domestic chips and a full-stack AI approach, will bolster Huawei Cloud's competitive standing in regions that emphasize data sovereignty and local technological ecosystems.

This strategy offers an alternative to globally dominant computing paths, appealing to governments and enterprises seeking nationalized resources and an integrated, secure, and self-reliant AI infrastructure.

The '智果园' portal's agent-centric design, featuring Skill-based and CLI-based interactions, will drive a notable shift towards more autonomous cloud operations with reduced human intervention.

By enabling AI agents to directly control cloud services and orchestrate complex workflows through simple commands, the platform facilitates a new paradigm of automated decision-making and execution in enterprise environments.

⏳ Timeline

2005

Huawei's cloud division was formed.

2011

Huawei established the Noah's Ark Laboratory for basic AI research.

2017-03

Huawei Cloud officially became a top-level business unit.

2017-09

Huawei Cloud launched its Enterprise Intelligence (EI) platform.

2020

Huawei Cloud became the fifth-largest cloud provider globally in revenue and market share (IaaS layer).

2026-06-05

Huawei Cloud launched '智果园' and a suite of Agentic AI products at its INSPIRE conference.

Huawei Cloud launches Agentic AI portal '智果园'

⚡ 30-Second TL;DR

Key Points

🧠 Deep Insight

🔑 Enhanced Key Takeaways

🛠️ Technical Deep Dive

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline

📎 Sources (9)

👉Related Updates

Glow emerges from stealth at $1.2B valuation for AI security

Kunlun Tech Mureka V9.5 Enhances AI Music with Reflective Reasoning