Huawei Cloud CEO Zhou Yuefeng: We must cultivate a "silicon-based black soil" for intelligent agents.
Author | Huang Yu
“Four years ago, we believed that doing AI meant buying a bunch of computing cards; three years ago, many companies thought doing AI meant developing their own large models. But today, we see that doing AI is about developing and utilizing intelligent agents.”
On June 5th, about half a year after becoming CEO of Huawei Cloud, Zhou Yuefeng deeply discussed his views on current AI development during a media briefing at the Huawei Cloud INSPIRE Creators Conference.
He pointed out that computing power and model technologies have now moved to the background, while the technological iteration of models and computing power is happening quickly.
Therefore, for many industries, whether we can build a dedicated public cloud that ensures data security and enables fast iteration and sharing of AI computing resources and model resources is crucial.
In addition, Zhou Yuefeng stated that the new paradigm of infrastructure in the Agentic AI era equals an efficient token factory + continuous learning + integrated intelligent scheduling + secure autonomy.
To seize this industry opportunity, Huawei Cloud's strategic positioning has shifted from the “black soil of computing power” in 2025 to the “silicon-based black soil.”
Zhou Yuefeng noted that Huawei Cloud’s direction in the next few years is to leverage soft and hard chip collaboration to focus on intelligent agents and develop the “silicon-based black soil.”
At the conference, four major layers and more than ten new Agentic AI products and ecosystem zones were launched in one go.
Zhou Yuefeng stated that this is the most AI product releases Huawei Cloud has ever had at a single event since 2025.
Specifically, at the Agentic Infra layer, Huawei Cloud introduced four new products: AICS Lingqu Intelligent Computing Cluster, AMS Agentic Memory Storage Solution, CCE VolcanoNext Integrated Intelligent Scheduling, and Agent runtime environment AgentSphere.
Zhou Yuefeng emphasized that only by achieving extremes in token cost, integrated scheduling, RL (reinforcement learning) efficiency, and secure autonomy can we gain the initiative in developing infrastructure for the Agentic AI era.
At the Agentic Model Service layer, Huawei Cloud released the new-generation ModelArts Next model training and inference platform, building four core capabilities: RL reinforcement learning services, confidential inference, model routing, and model matrix.
At the enterprise-level intelligent agent platform layer, Huawei Cloud launched the brand new Huawei Cloud portal “Smart Orchard,” fully skill-ified and CLI-ified, designed specifically for agent services rather than human users.
Aiming at different industries, Huawei Cloud also announced the official launch of the “Industry AI Dream Factory,” with smart healthcare, embodied intelligence, intelligent manufacturing, and scientific computing as four major zones. This, in some ways, reflects the industry markets Huawei Cloud now intends to focus on.
It is worth mentioning that for the embodied intelligence industry, Huawei Cloud launched the world’s first full-process embodied intelligence development platform, CloudRobo. The platform is expected to enter public beta on June 30, and features a PB-level data foundation and development pipeline, a cloud-native embodied model production engine, and the first fully domestic Real-Sim data production and model evaluation system.
Additionally, Zhou Yuefeng revealed that more Industry AI Dream Factory zones will continue to be created in the second half of the year.
Since 2025, competition among major cloud providers in the AI battleground has intensified. Alibaba Cloud, Byte Volcano Engine, and Baidu Intelligent Cloud have all sought to increase their presence at airports, advertising their “number one” status in different AI cloud market segments.
When asked how Huawei Cloud will differentiate itself in the fiercely competitive AI cloud race, Zhou Yuefeng put forward three distinctions. First, Huawei Cloud’s core goal is to use AI technology to enhance productivity for enterprises, especially those in critical national sectors, rather than blindly pursuing total token volume and revenue, valuing the actual value tokens bring rather than just emotional value.
Second, Huawei Cloud adheres to a dual strategy of balancing both public and private clouds, catering to some needs of key national sectors. Third, Huawei Cloud insists on localization, aiming to build a second computing power plane.
Risk Warning and DisclaimerThe market is risky, investment requires caution. This article does not constitute personal investment advice, nor does it take into account individual users’ special investment objectives, financial conditions, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their specific circumstances. Invest accordingly at your own risk.