Xi Jinping to Deliver First-Ever Keynote at WAIC 2026, Signaling China's Bid for Global AI Leadership

Chinese President Xi Jinping will deliver the opening keynote at the 2026 World Artificial Intelligence Conference (WAIC) in Shanghai on July 17, making his first-ever appearance at the annual summit and marking a dramatic escalation in Beijing’s ambition to shape how artificial intelligence is governed worldwide.

The announcement, made on July 13 by Chinese state authorities, ends years of speculation about when China’s top leader would personally engage with the event that has served as the country’s showcase for domestic AI capability since 2018. That Xi is not merely sending a delegate but appearing himself signals that the Chinese government now views AI governance as an arena of strategic geopolitical competition — one where it intends to claim the high ground.

A Summit Redefined

WAIC 2026 runs from July 17 to July 20 in Shanghai and carries the theme “Intelligent Partners, Co-create the Future.” The scale alone tells a story: for the first time, the exhibition floor will exceed 100,000 square meters, with more than 1,100 enterprises participating and over 3,000 exhibits on display. More than 300 products will make their global debut at this single event — a figure that rivals the launch cadence of an entire consumer electronics trade show season compressed into four days.

Nine Nobel and Turing award laureates are scheduled to participate, including deep learning pioneers Geoffrey Hinton and Yoshua Bengio, reinforcement learning pioneer Richard Sutton, and computer scientist Andrew Chi-chih Yao. Their presence alongside Xi and Chinese government officials creates an unprecedented convergence of global AI scientific leadership with Chinese state authority — precisely the image Beijing is cultivating.

Running in parallel with the main conference is a High-Level Meeting on Global AI Governance, a diplomatic track designed to advance China’s proposals for international AI coordination. The most consequential of these is the proposed World AI Cooperation Organization (WAICO), which Beijing wants headquartered in Shanghai.

China’s Governance Gambit

Xi’s speech, according to official statements from Beijing, will “systematically elaborate on China’s policies, position, visions and propositions on AI development and governance.” Analysts who follow Chinese technology policy closely read this as a formal declaration of intent: China aims to be a rule-maker in AI, not merely a rule-taker.

The timing is pointed. The European Union’s AI Act entered enforcement earlier this year. The United States has assembled voluntary frameworks and export controls but has yet to pass comprehensive AI legislation. Into this gap, Beijing is stepping with a rival vision — one that emphasizes state sovereignty, developing-world inclusion, and what Chinese officials call “AI for good” rather than the risk-mitigation framing dominant in Western capitals.

The WAICO proposal, if realized, would give China something it has long sought in global technology governance: an international institution it helped create, operating under principles sympathetic to its interests. Critics in Washington and Brussels note that proposed membership criteria could effectively exclude Western democracies while giving China outsized influence over AI standards in the Global South.

“Regulators picked a word for AI this week, and the word is systemic,” one technology policy observer noted on the eve of the conference. “Xi delivering the keynote at WAIC is the clearest possible signal that China intends to compete on the governance layer, not just the model layer.”

Hardware Takes Center Stage

Beneath the geopolitical theater, WAIC 2026 is also a product showcase for China’s homegrown AI ecosystem, and the hardware announcements are striking.

Huawei is scheduled to unveil its Atlas 950 computing cluster — a system built around 8,192 of the company’s Ascend neural processing unit cards — which has been optimized specifically for large-scale model training and inference. The Atlas 950 represents Huawei’s answer to NVIDIA’s H100 and B200 systems, developed entirely without American components following U.S. export restrictions that cut off Huawei’s access to leading-edge chips in 2020. While independent benchmarks of the Atlas 950 are not yet publicly available, the system’s 8,192-NPU configuration positions it as Huawei’s most serious bid yet for hyperscale AI training.

Also making its world debut is what organizers are billing as the first AI Agent smartphone — a handset designed from the ground up around persistent, autonomous AI agents rather than traditional app-based computing. ZTE is reported to be the manufacturer. Unlike existing AI smartphones that run agent capabilities as software layers atop conventional app architectures, the device integrates on-device AI agent orchestration at the hardware level, enabling persistent background agents capable of monitoring and taking actions across the device without explicit per-task user invocation.

MiniMax, one of China’s better-funded model startups with reportedly strong revenue and a multi-billion-dollar valuation, will launch its M3 multimodal large model at the conference, offering extended reasoning capabilities across text, code, image, and video inputs. The Jieyue Agent Operating System and several humanoid robotics platforms are also among the 300-plus debut products.

Reading the Moment

The decision to have Xi personally attend is not incidental. China’s AI sector has undergone a remarkable transformation over the past 18 months. Models from DeepSeek, Alibaba’s Qwen family, and ByteDance’s Doubao have achieved competitive benchmark scores against frontier models from OpenAI and Anthropic. DeepSeek’s V3 and R1 releases in late 2025 demonstrated that Chinese labs could reach the performance frontier at a fraction of the compute cost — and shook Western assumptions about the pace of the race.

But hardware remains a constraint. U.S. export controls on advanced GPUs have pushed Chinese companies toward domestic alternatives like Huawei’s Ascend and Biren’s chips, which still lag behind NVIDIA’s latest offerings in raw throughput. The gap is closing, but not yet closed. Against this backdrop, WAIC becomes part product fair, part policy declaration, and part diplomatic push — a coordinated signal that China intends to compete not just on model quality but on the rules of the game itself.

For global policymakers attending the parallel governance track, the challenge is clear: how do you engage constructively with a framework-setting process when the frameworks themselves may encode fundamentally different values about state control, data privacy, and accountability? China’s preferred governance architecture — one in which AI development is a matter of national sovereignty first and international harmonization second — clashes directly with the multilateral approach favored by the EU, the OECD, and much of the G7.

The answers — or the failure to find them — will define international AI governance for years to come. They start being negotiated, formally, in Shanghai on July 17.