“In this new era, AI will replace energy as the most important commodity, driving the daily work of all industries. The vast majority of AI capabilities will be generated and delivered in the form of Tokens on the cloud computing network. Tokens are the electricity of the future.”

At the main forum of the 2025 Yunqi Conference, this statement by Wu Yongming, the CEO of Alibaba Group, outlined the essence of AI more clearly. When artificial intelligence becomes the “primary productive force”, the core element behind it – computing power – is highlighted. Whether it is stable and sufficient directly determines whether AI can step out of the laboratory and truly enter industries and society.

This transformation has sparked a “computing power boom”. Currently, enterprises and institutions are starting to pay more attention to how to obtain sustainable, green, and scalable computing power support. This is not only a technical issue but also a real – world test of whether the industry can cross the next threshold.

The Yunqi Conference is precisely one of the best windows to observe the development path of the AI industry. As a long – term partner of Alibaba Cloud, ZTE comprehensively demonstrated the innovative achievements of both parties in the fields of intelligent computing power, extreme computing power, lossless switching, intelligent computing centers, and global cooperation at the Yunqi Conference, centered around the theme of “Technology for Good, AI Empowering the Real Economy”.

By providing an open and integrated intelligent foundation, a large – scale intelligent computing engine, and a green and low – carbon flexible solution, ZTE attempts to answer a core proposition in the current AI computing power industry – how to systematically build a powerful and inclusive computing power supply system to enable AI to truly move from “demonstrable” to “producible”.

Three New Answers to the “Computing Power Question”

If the core proposition of the AI industry is “how to make computing power more inclusive” and drive the industry from a “technology competition” to “industrial application”, then ZTE presented three targeted answers at this Yunqi Conference.

Firstly, Full – stack Intelligent Computing – An Open and Integrated Intelligent Foundation
In the practices of many enterprises, AI projects often get stuck at the threshold of “incompatible hardware and difficult – to – switch architectures”. The full – stack open intelligent computing solution proposed by ZTE covers the hardware system and the AI platform, and is compatible with mainstream GPUs and CPUs, allowing enterprises to make flexible choices according to their business needs. On the hardware side, it includes self – developed switching chips, intelligent computing servers, super – node servers, and the AiCube intelligent computing all – in – one machine. In terms of software, relying on process optimization and framework upgrades, the new platform can achieve efficient training and inference.

When enterprises need to deploy different models in a short period, the minute – level application development capability of the Co – Sight Super Intelligent Agent Factory can turn experimental explorations into products that can be quickly launched, accelerating the inclusive implementation of AI.

Secondly, Super – node Servers – A Large – scale Intelligent Computing Engine
As large models enter the long – term operation stage, the intensity of single – point performance is no longer the focus. Whether it can be stably expanded to the “ten – thousand – card level” is what the industry really cares about. At this Yunqi Conference, the super – node server exhibited by ZTE can support 64 GPUs per machine, which can be vertically expanded to meet extreme performance requirements and horizontally expanded into a large – scale cluster.

This means that this solution can support tens of millions of user requests from Internet and e – commerce platforms or complex prediction models of energy enterprises, so that AI training and inference are no longer restricted by the scale ceiling. The full – sequence coverage of intelligent computing servers from 8 – card to 64 – card also means that enterprises do not need to “go all in” at the beginning and can gradually upgrade the computing power scale as their business evolves.

It is understood that currently, ZTE’s intelligent computing servers have widely entered the core business scenarios of several leading domestic Internet enterprises and achieved large – scale commercial use in the international market, gradually building a global intelligent computing ecosystem. This also marks that ZTE is continuously breaking through the underlying core technologies in the three major fields of “computing power, network, and terminal”, forming a full – stack advantage.

Thirdly, Intelligent Computing Centers – A Green and Low – carbon Flexible Solution
If the super – node server answers the question of “whether the computing power is sufficient”, then the intelligent computing center focuses on “whether the energy consumption can be sustained”. According to data from People’s Daily, as of the end of June this year, the total scale of racks in domestic computing power centers in use reached 10.85 million standard racks, and the intelligent computing power scale reached 78.8 EFLOPS. The storage capacity scale exceeded 1680 EB, a year – on – year increase of about 40% compared with 2023.

In this general trend, ZTE proposed the concept of “intelligent and energy – saving” in the construction of intelligent computing centers. The new solution features a modular and prefabricated architecture, which can shorten the delivery cycle by 40% and significantly reduce the energy consumption of liquid cooling and high – density deployment.

For operators, this can fully relieve the pressure of electricity costs. For manufacturing and government customers, it means that even under the constraint of the “dual – carbon” goal, they can still safely expand the scale of data centers. This flexible solution provides a more balanced solution to the long – standing “performance vs. energy efficiency” dilemma in the industry.

Thirteen Years of Collaboration, Opening a New Chapter in Full – stack Intelligent Computing

In 2013, ZTE built the first OTN transmission network for Alibaba Cloud. Since then, the cooperation between the two parties has continuously expanded, from the Chinese market to the global layout. ZTE’s hardware and engineering capabilities have been continuously integrated into Alibaba Cloud’s business landscape.

At this year’s Yunqi Conference, ZTE also actively participated in multiple forums such as “Special Cloud Intelligence and Internationalization”, “Hardware Engineering Practice of the Super Intelligent Computing Interconnection System (ALS)”, and “Alibaba Cloud Tongyi Xiaomi Special Session”, sharing the latest practices of cloud – AI integration. This clearly signals that for the AI industry to reach a higher level, it must rely on the collaborative construction of upstream and downstream partners.

Under the conference theme of “Integration of Cloud and Intelligence, Symbiosis of Carbon and Silicon”, a clearer consensus is emerging. The industrialization of AI no longer depends on single – point technology advantages but on whether software and hardware, supply and demand, and domestic and overseas markets can be truly woven into an ecosystem. For any manufacturer, this means shifting from a “product – oriented” approach to a “system – oriented” one, supporting partners to move from “local optimization” to “global collaboration”.

When Chinese Computing Power Projects Become “Global Solutions”

In 2025, the Chinese cloud computing market showed an obvious differentiation trend. Cloud service providers chose completely different paths in terms of capital expenditure and market strategies. Behind this differentiation is the different understanding of the essence of computing power in the industry. Some may regard GPU resources as tickets to seize the market, believing that economies of scale will bring returns. Others think that in – depth exploration of computing power efficiency and industry depth is the long – term solution.

ZTE and Alibaba Cloud demonstrate a third possible path – ecological symbiosis. The two parties have in – depth collaboration in the fields of public cloud, hybrid cloud, and private cloud.

On the special cloud side, the two parties have launched industry – specific solutions based on the integration of ZTE servers and other hardware with Alibaba Cloud’s operating system/platform, covering more than a dozen industries such as government affairs, finance, energy, power, and transportation. These solutions have been implemented in more than 30 provinces, serving more than 200 customers. Behind this is a profound understanding of compliance, security, customized delivery, and long – term operation and maintenance.

On the public cloud side, ZTE has been deeply involved in the JDM model, truly implementing the supply model of “standardized production capacity + on – demand customization”. This model not only improves the supply efficiency but also enhances the resilience of the industrial chain.

At this year’s Yunqi Conference, Alibaba Cloud proposed to build a “globally leading full – stack artificial intelligence service provider”, which requires not only large – model and platform capabilities but also global infrastructure.

In the process of exploring the international market, ZTE and Alibaba Cloud have jointly built cloud capability centers in Indonesia, Algeria, Singapore, Malaysia, and other places, relying on their respective cloud network, channel, and ecological resources, forming replicable and expandable regional models.

The key to this cooperation mode lies in “joint exploration + local delivery”. Alibaba Cloud provides the platform and ecosystem, while ZTE provides hardware and engineering capabilities. For enterprises going global, this means they can enjoy the world – leading cloud platform capabilities and receive customized services that meet local regulations and market demands.

These projects not only support enterprises to go global and operate locally but also promote the transformation from “pilot to scale”, transforming the “Chinese – style computing power project” into globally applicable solutions.

From the transmission network to the computing network, from the hardware foundation to full – stack collaboration, ZTE and Alibaba Cloud are moving forward hand in hand on a path of closer collaboration. If “Tokens are the electricity of the future”, then what ZTE is doing is to build the “infrastructure” for this digital energy revolution.

Today, computing power is not only a competitive resource for enterprises but also a key variable in the global industrial chain restructuring led by AI. Those who can truly implement the inclusive computing power in industries and society with a faster speed and a more stable foundation will be able to take the lead in this global competition.

The cooperation between ZTE and Alibaba Cloud is part of the answer to the question of “where AI is heading in the future”. It means that upstream and downstream partners in the computing power industry can form a system through ecological co – construction, find a balance through ecological symbiosis, and create models through global cooperation. That’s why ZTE is qualified to stand at the center of the industrial stage in the AI era and become one of Alibaba Cloud’s closest allies.