HomeArticle

WAIC 2026 Guide | World Expo, how many surprises have you been hiding?

未来一氪2026-07-13 11:43
2026WAIC Expo Area Visit Guide: An Overview of Pavilions, Forums and Interactive Experiences

                                              Expo Venue Guide

The 2026 World Artificial Intelligence Conference is about to kick off soon.

As the core highlight of this year's event, the World Expo Center and World Expo Exhibition & Convention Center feature an extremely high "AI density": on the hardware exhibit front, 108 chip models, 261 large models, 208 embodied intelligent terminals, and over 300 physical devices are showcased together; the forum lineup is even more impressive, with more than 140 industry thought-sharing sessions gathering 1500 high-profile guests, among whom 1030 are outstanding young AI talents from home and abroad under 35, plus 9 top award winners in attendance. The most significant forums, cutting-edge technologies, vibrant robot displays, and various hands-on interactive experiences can almost all be found here.

Here comes the question: the venues are too large, there are too many booths, and the forums are too densely scheduled — how exactly should you plan your visit?

Don't worry, this Expo Area Guide will help you sort out all the forums, pavilions, and experience activities in one go. Even if it's your first time attending WAIC, you can avoid detours and spend your time on what's truly worth seeing.

1. Smooth Pavilion Tour: Unlock H1-H4 One by One, Cover the Entire AI Industry Chain

The World Expo Exhibition & Convention Center has four main pavilions, H1 to H4.

In simple terms:

H1 shows how AI integrates into industries, H2 focuses on computing power and hard technologies, H3 is dedicated to robots, and H4 showcases startups and innovative projects.

Together, these four pavilions form a complete panorama of the AI industry ecosystem.

                                        H1 Application & Ecology Pavilion

                              Application&Ecology Pavilion

Pavilion H1 dives deep into the core segments of the full AI industry chain, with a comprehensive business layout covering general and multimodal large models, vertical domain model applications, industrial large models and industrial platforms, AI agents, underlying AI infrastructure, fintech, AI for Science, data services, and various AI terminal implementation cases. It connects the complete industry chain from underlying computing power, foundational models, industry empowerment to physical deployment, fully demonstrating the implementation outcomes and technological iterations of the artificial intelligence sector.

The exhibition area brings together leading enterprises across all niche AI tracks, with a strong lineup and abundant highlights. The participating brands include Alibaba Group, Baidu, Baichuan Intelligence, Kingsoft Office, Luminance Technology, Beijing Digital Intelligence, Juyun Technology, Yili, Luckin Coffee, Guotai Haitong, Zhongke WenGe, Huawei, JD.com, Kaos, PipeChina, iFLYTEK, Lenovo, Deep Principles, Ant Group, CCII, SenseTime, Huifu Payment, Inspur Group, Coopas, Luxshare Precision, Sugon, PaiCloud, Infinicore, Vertiv, Baosight Software, China Mobile, Siemens, Tencent, China Tower, MiniMax, Schneider Electric, Stone Technology, Infinite Light, Transwarp, PatSnap, TTC, and many more.

                                              Selected Exhibit Showcase

🆕 Baidu: DuMate General Agent

A newly launched general AI companion agent that uses natural language interaction to break the limits of traditional conversations, completing complex cross-application and cross-document tasks to form a full closed loop from understanding to execution. Leveraging long-term memory to continuously learn user preferences, its interaction experience keeps improving, balancing personal productivity enhancement and enterprise process standardization. Users can initiate tasks anytime, anywhere, creating a 7×24-hour online, business-aware, self-evolving execution-focused AI assistant.

🆕 MiniMax: MiniMax M3 Large Model

Built with the self-developed MSA sparse attention architecture, it supports ultra-long contexts of millions of words. Its programming capabilities have been greatly upgraded, with code debugging, full-stack development, and performance optimization levels matching top overseas closed-source models. It delivers excellent performance in office search and document processing agents, making it well-adapted to financial scenarios. New multimodal features enable it to recognize images, text, and videos, as well as control computers, expanding the perception boundaries of intelligent agents.

Infinicore: InfiniClawBox Secure Anonymization Agent

Adopting an edge-cloud integrated architecture and an innovative three-stage anonymized inference solution, it balances cloud computing power and local data privacy, solving the long-standing pain point of balancing cost, performance, and privacy for local agents. It prevents sensitive data from being uploaded to the cloud, making it suitable for government and enterprise compliance scenarios. It has already been deployed in use cases such as contract review and document analysis for law firms, with open APIs to facilitate industry integration.

Vertiv: Vertiv SmartRun Prefabricated Overhead Infrastructure

It integrates busbars, walkways, and piping wiring systems into a single unit, installed via integral hoisting to revolutionize traditional overhead construction methods for data centers. With excellent deployment speed and scalability, it cuts on-site construction time by up to 85% and overall deployment costs by up to 40%, effectively addressing the schedule and cost pain points of data center projects.

PPIO: Agent Platform

Two complementary products build a full closed loop for AI agent development and deployment. The Agent Sandbox provides an isolated, secure cloud runtime environment that supports code execution, data operations, and web interactions, with full risk isolation, instant startup, and pay-as-you-go billing. Agent Hosting enables persistent cloud deployment, with instant on/off, state retention for second-level restarts, no server maintenance required, and seamless collaboration directly through Feishu.

Moonshot AI: Kimi K2.7 Code Model

Compared to the previous K2.6 version, it optimizes long-context programming instruction execution, reduces redundant inference, and cuts average Token consumption by 30%. Scores on multiple code benchmark tests have been significantly improved, while the autonomous execution capabilities of agents are enhanced, boosting agent-related benchmark performance by approximately 10%.

01.AI: Wance Decision Hub Platform

As a foundational AI base for top enterprise decision-makers, it is equipped with the Ontology 2.0 dynamic context engine and multi-agent planning and decision-making engine. It can convert raw enterprise data into actionable decision facts, realizing a controllable closed loop of collaborative agent execution. It integrates enterprise data, systems, and accumulated experience to build continuously iterable AI decision-making capabilities for organizations.

Siemens: Eigen Engineering Agent

The world's first dedicated AI agent for industrial automation engineering, built on Siemens' deep industry expertise, balancing precise engineering guidance and data security. It supports PLC code generation, HMI development, drive optimization, and other functions, with operational efficiency 2-5 times that of manual work, boosting engineering efficiency by 50% and solution quality by 80%, enabling industrial AI to evolve from assisting roles to fully autonomous execution.

Deep Principles: AI Scientist Product Matrix for Industrial Materials R&D

Making its debut exhibition, the lineup includes a material prediction large model, Agent Mira 2.0, and an AI glasses-integrated automated lab solution. It covers the full R&D chain from requirement parsing, molecular generation, property prediction, automated experimentation to data review, suitable for new materials, new energy, daily chemicals, electronics, and other sectors, effectively reducing R&D trial-and-error costs and accumulating iterative R&D knowledge.

Tianwu: MatwingsVenus Protein R&D Platform

An all-in-one protein R&D platform centered on intelligent agents, which allows users to complete enzyme mining, targeted protein evolution, de novo design, and other tasks through natural language conversations. It can connect to its self-owned automated laboratory, where robots perform sample preparation, purification, and testing, with experimental data fed back to the AI to form a conversational closed loop integrating dry (computational) and wet (lab) R&D processes.

Coopas: Full-Link AI Data Service System

Coopas has built an integrated AI data service system, leveraging Corpus Magic 2.0 to realize automated data pipelines and systematic iterative upgrades. It has established a physical AI data training ground covering multi-dimensional scenarios such as world models, autonomous driving, low-altitude equipment, and embodied intelligence, innovating data production and governance paradigms with a complete set of supporting evaluation and alignment tools. Through the FDE engineering platform, non-standard data requirements are standardized and broken down into the full workflow of collection, cleaning, annotation, testing, and application, with automated scheduling and delivery.

🆕 Kingsoft Office: WPS Comate

A brand-new enterprise-grade AI office agent that empowers WPS 365 to evolve into an enterprise intelligent brain. Covering the full-scenario office ecosystem, it can autonomously understand organizational knowledge, connect business workflows, execute complex office tasks, and retain traceable outcomes, precisely solving enterprise AI implementation pain points in data utilization, process execution, process control, and result traceability, helping organizations convert their data and workflows into manageable, deployable intelligent office productivity.

Ant Group: Ant A-Fu