Xiaomi Unveils Xuanjie O100, an AI Accelerator Chip Built for Large Models
On August 24th, at the Xiaomi Xuanjie Chip Technology Communication Conference, Xiaomi officially launched the Xuanjie O100, an AI accelerator chip dedicated to large models. It features a 6nm 3D wafer-level stacking solution, achieving a bandwidth of 1.22TB/s. Architecturally, the Xuanjie O100 employs a near-memory AI computing architecture, paired with the Xuanjie high-bandwidth matrix bus, supporting multi-core parallel computing. Running the Xiaomi MiMo 3B model, it can hit an inference speed of 330 Tokens/s.
Xpeng Robot Unit’s First Funding Round Exceeds $900 Million, Setting a New Record for Single-Round Private Equity in China’s Embodied Intelligence Sector
On August 24th, Xpeng Group announced that its humanoid robot business has signed equity financing agreements with multiple investors. This round marks the first funding for Xpeng’s robot business, with total financing exceeding $900 million and a post-investment valuation surpassing $6.3 billion. This sets a new record for single-round private equity financing in China’s embodied intelligence industry. The round was led by IDG Capital with participation from GSR Ventures, and also secured support from strategic investors Tencent and Alibaba. Xpeng Group continues to hold a controlling stake in the robot business. According to NUPIAO, the funds from this round will accelerate the mass production of the Xpeng IRON and the R&D iteration of Xpeng’s physical AI models, driving the commercial application of advanced general-purpose humanoid robots.
Zhiyuan Lingxi X2 Wins Gold in 100-Meter Obstacle Race at Robot Games
On the evening of August 24th, the gold medal for the 100-meter obstacle race was decided at the 2nd World Humanoid Robot Games, with the Zhiyuan Lingxi X2 taking the crown. It’s reported that the Zhiyuan Lingxi X2 competed in its mass-production version without any body modifications. According to the Zhiyuan team members, the 100-meter obstacle race is a high-intensity test of a humanoid robot’s comprehensive athletic abilities. The event focuses on assessing five core capabilities: high-speed mobility, dynamic balance, real-time perception and navigation, traversal of complex terrain, and whole-body posture adaptation.
Leapmotor Confirms Robot Business Plans, Says Formal Announcement Coming Soon
On August 24th, during Leapmotor’s (09863.HK) 2026 interim earnings call, company Vice President Li Tengfei responded to rumors about a robot business. He stated that new energy vehicle companies, especially those with full-stack in-house R&D capabilities, are among the best-positioned enterprises to build robots. Leapmotor indeed has its own plans in this area, and specific information will be released soon through a formal announcement.
Previously, business registration information showed that Leapmotor’s subsidiary, Zhejiang Lingsheng Power Technology Co., Ltd., wholly established Huzhou Lingsheng Precision Manufacturing Co., Ltd. in July 2026, with a registered capital of 210 million yuan. Its business scope includes industrial robot manufacturing and intelligent robot R&D.
TRAE and Coze Merged into Doubao; ByteDance to Launch Unified Office Brand “Doubao Work”
ByteDance has completed a team integration for its office AI products. The TRAE and Coze teams will be fully integrated into the Doubao ecosystem. Specifically, TRAE Work and Coze will be merged with Doubao’s product capabilities for work scenarios, while TRAE IDE and CLI will continue to develop as a programming product line under the Doubao brand. Following this adjustment, ByteDance has further clarified Doubao’s role as its core AI business, concentrating its AI office product efforts on Doubao. Additionally, Doubao is expected to launch a standalone AI office product called “Doubao Work” within this week, serving as ByteDance’s unified product and brand for AI office scenarios. In response to this news, ByteDance stated that the adjustment aims to better coordinate product and technical resources to provide users with a superior AI work experience, and that existing user benefits will not be affected.
