From July 17th to 20th, the 2026 World Artificial Intelligence Conference (WAIC) is being held in Shanghai.
At the LENOVO GROUP exhibition area, popular displays such as embodied intelligent robots and a World Cup AI technology experience zone attracted numerous visitors for interaction.
Showcasing a theme of "Hybrid AI Makes AI a Productive Force for Innovation," Lenovo systematically presented the complete chain from token production to value creation, spanning from super nodes and intelligent computing centers supporting large model training and inference, to enterprise super agents and city agents deployed in real-world scenarios, and a diverse range of innovative AI terminal products.
The exhibition demonstrated how AI is genuinely integrating into industries and personal life, transforming into measurable productivity outcomes.
WAIC 2026 Lenovo Exhibit: Heterogeneous Intelligent Computing Platform 5.0+ and Super Node, Building the Token Production Foundation
As artificial intelligence accelerates its move into industrial applications, the continuous development of large model training, inference, and agent applications has led to an explosion in computing power demand.
At WAIC 2026, LENOVO GROUP systematically demonstrated its full-stack, software-hardware capabilities in token-driven computing infrastructure, building the foundation for token production in the age of intelligent agents.
Prominently displayed were the recently launched Wanquan Heterogeneous Intelligent Computing Platform V5.0 and Super Node solution, which combine software and hardware to maximize every bit of computing efficiency.
The former, relying on nine differentiated core technologies including continuously iterated cluster training/inference acceleration and chip-model compilation optimization, achieves full-scenario coverage from hundreds to tens of thousands of accelerator cards.
The Super Node architecture, a "new species" in the server field, integrates computing, communication, and storage into a logically unified node, significantly boosting computing power and application efficiency to meet the demands of large-scale training, high-concurrency, and long-duration inference task execution.
The Lenovo Wentian Super Node solution breaks through bottlenecks via architectural innovation, with a single node capable of housing 40 GPUs, and possesses four core capabilities: powerful computing, low latency, easy deployment, and scalability.
LENOVO GROUP's Wanquan Heterogeneous Intelligent Computing Platform and Super Node cluster were on display.
Zhou Tao, General Manager of the Server Business Unit of Lenovo China's Infrastructure Business Group, stated in an interview that this year is poised to be a significant year for AI application deployment, and the company is adopting a dual-track market strategy.
"On one hand, we are seizing the high-growth opportunities in the internet business; on the other, we are deepening our presence in traditional non-internet industries," Zhou said.
He noted that as AI drives up the cost of infrastructure and storage components, industry unit prices and sales scale continue to rise.
Driven by these two strong trends, Lenovo China's Infrastructure Business Group is confident in advancing its business toward a revenue target of 100 billion yuan next year and establishing a leading position in the fierce market competition through its comprehensive AI product portfolio.
Zhou Tao explained that to address the challenge of fragmented computing power forms caused by the increasing diversity of GPU and CPU combinations, Lenovo has introduced an innovative "1+3+N" universal architecture system.
This architecture aims to use a unified foundational platform to be compatible with various mainstream CPUs and different forms of GPUs (including 8-card, 16-card, and Super Node configurations), thereby avoiding redundant product development for single combinations.
This strategy not only significantly reduces R&D costs and shortens development cycles but also gives customers the flexibility to replace hardware components, markedly enhancing product iteration agility.
Zhou Tao noted that in traditional GPU servers, cards are only interconnected within a server; connecting to the next server requires network and CPU connections, which introduces latency and reduces efficiency.
However, all GPUs within a Super Node are interconnected, which can greatly improve GPU efficiency.
"Our (Super Node solution) can improve (efficiency) by more than 30% compared to original servers," Zhou claimed.
Regarding the recent intensive industry focus on Super Nodes, Zhou Tao commented that over the past two years, large model training has begun transitioning towards inference.
Super Nodes, with their contributions to computing power enhancement and energy consumption reduction, will gain more attention and adoption from customers in the next year or two.
More critically, the launch of Super Node products by major manufacturers this year marks the beginning of a new era.
At the event, LENOVO GROUP also showcased the world's first 2U rack server equipped with the Blue Core LX500 processor—the Lenovo Wentian WR5219G5.
This also marks Lenovo's first public display of a server product based on an open-source architecture.
Building on this solid computing foundation, Lenovo China's Solution Services business uses its Qingtian AI as an engine to deliver visible AI value to customers.
Among these offerings, the Intelligent Computing Center (AIDC) solution provides customers with a comprehensive package for AI data center investment, design, construction, and operation and maintenance.
Lenovo's xCloud intelligent cloud optimizes through integrated modeling and computing, along with dynamic tuning of token pathways, ensuring every token is used effectively.
To integrate tokens into customers' actual business processes, Lenovo created the FDE (Front-loaded Development and Delivery Integration) model and launched the FDE Workshop, embedding intelligent agents into the entire enterprise workflow of "R&D, Production, Supply, Sales, and Service" for full lifecycle management of agent design, development, delivery, and operation.
The reusable intelligent assets and templates accumulated in this process are deposited into an intelligent agent asset library.
Through Qingtian AI, Lenovo has developed Lenovo Baiying, Lenovo Qingtian, and Lenovo's full-cycle AI services.
At WAIC, Lenovo used real-world examples to demonstrate how these comprehensive intelligent capabilities act on actual business scenarios.
Furthermore, on the afternoon of the 17th, at the "Artificial Intelligence Terminal Industry Evolution and Collaborative Development Forum" held concurrently with WAIC, the first batch of test results for the national standard series "Artificial Intelligence Terminal Intelligence Grading Standard" were released.
LENOVO GROUP had a total of 42 products, including PCs, tablets, and mobile phones, pass the L3-level tests, demonstrating its leading advantage in AI terminal technology and product deployment.
Comments