Breaking

China's AI Models Surge to Near-Parity in Global Use as Xi Prepares WAIC Debut in Shanghai

July 16, 2026 at 22:41 EDT

Foundation Models
Policy & Regulation
Open Source

China's rapidly advancing artificial-intelligence models are giving President Xi Jinping fresh leverage to shape the technology's global rules, even as their expanding reach raises security alarms in both Washington and Beijing. The dynamic comes into sharp focus this week as Xi prepares to make his first-ever appearance at the World Artificial Intelligence Conference in Shanghai, held July 17-20, where he is expected to deliver a keynote address alongside a High-Level Meeting on Global AI Governance.

July 17–20 · World AI Conference, Shanghai

China's AI Surge Hands Xi a Global Stage — and Sets Off Security Alarms

As Xi Jinping makes his first-ever appearance at WAIC, Chinese-built models have vaulted toward record usage share worldwide — giving Beijing fresh leverage to shape AI's rules, even as trust concerns follow the models everywhere they spread.

61%

of top-ten model usage was Chinese-built in late February

5.3T

tokens routed to Chinese models, of 8.7T total that week

300K

attendees drawn by past editions of the conference

Chinese vs. Western share of top-ten model usage

Tokens consumed via a real-world routing platform, late February

61%

Chinese
models

39%

Western
models

Even by April, Chinese models still held above 45% of that consumption.

Leading Chinese models by individual usage share

Xiaomi MiMo-V2-Pro

21%+

Alibaba Qwen

~14%

MiniMax · DeepSeek · Zhipu

How the ascent was built

Open weights

Published weights let developers fine-tune and run models locally.

Aggressive pricing

Undercutting Western systems on cost while staying production-capable.

Efficiency under sanctions

U.S. chip curbs pushed labs toward domestic substitution and lean design.

The leverage

Usage-share leadership positions China to author global AI norms
Strong benchmark scores in coding, math and multilingual output
Xi's keynote frames openness against a tech "iron curtain"

The trust deficit

Data-privacy and censorship concerns around Chinese-built models
Washington frets over security exposure from foreign models
Weaknesses remain in ambiguous reasoning tasks

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.