Gasgoo Munich- Here are the top stories in embodied intelligence and driver assistance this week:
Chinese humanoid robot models account for more than half of global total
Data released by the Ministry of Industry and Information Technology on July 20 shows that in the first half of the year, China’s competitive industries accelerated their expansion. In intelligent robotics, quadruped robots developed in China now claim nearly 70% of global sales, while the number of humanoid robot models has surpassed 400 — accounting for more than half the world’s total.
Driving this growth is not only China’s wealth of deployment scenarios but also a fully developed intelligent robotics supply chain.
Image credit: Gasgoo Auto
On the demand side, sectors ranging from automotive manufacturing, electronics assembly, and logistics to commercial and home services — even specialized fields like underground exploration and earthquake rescue — show robust demand. These applications provide a natural testing ground for iterating core technologies.
On the supply side, a complete ecosystem has taken shape, spanning core component R&D to final assembly and debugging. Crucially, the domestic production rate for key parts — servo motors, reducers, and controllers — continues to climb. Strengthening this supporting capacity effectively backs both innovation and mass production, laying a solid foundation for the industry’s rapid ascent.
Xiao Zhi Take: Over 400 models capturing half the global market isn’t just about numbers piling up. It is the inevitable outcome of a dual engine: China’s complete supply chain and its vast application scenarios.
China’s first quality evaluation standard for humanoid robot datasets launched
The national standard plan, "Humanoid Robot Dataset Part 6: Quality Evaluation," was recently officially approved. The project falls under the National Robot Standardization Technical Committee (TC591) and is executed by its Humanoid Robot Branch (TC591SC5). The Beijing Research Institute of Automation for Machinery Industry Co., Ltd. — the TC591 secretariat — is leading the drafting alongside UBTECH.
Addressing the specific challenges of humanoid robot datasets, the standard aims to establish unified rules covering evaluation objects, principles, metrics, processes, methods, and result reporting. This will make data quality definable, measurable, and verifiable, with a sharp focus on five key areas: data reliability, multimodal alignment, task coverage, annotation verifiability, and evaluation comparability.
For the industry, this standard will span the entire data lifecycle — from collection, processing, and annotation to storage, release, and application. It provides a unified benchmark for dataset selection, quality grading, third-party assessment, and acceptance testing. By helping R&D teams cut down on invalid data and rework, the standard shifts the focus from sheer volume to quality, lowering communication and transaction costs across the supply chain.
Xiao Zhi Take: Shifting the focus from quantity to quality marks a watershed moment as the industry moves from wild expansion to intensive cultivation.
First batch of "Henan-made" humanoid robots officially rolls off the line
On July 24, the first batch of "Henan-made" Zhongqing T800 humanoid robots rolled off the line at the company’s Zhengzhou base. The milestone marks another step forward in Zhongqing Robot’s push for mass production.
Image credit: Zhongqing Robot
Moving forward, these "Henan-made" T800 units will ship from Zhengzhou to destinations worldwide, including China, North America, Europe, and Southeast Asia, delivering intelligent solutions to a broader client base.
Prior to this, Zhongqing’s Shenzhen (Honghualing) manufacturing base began operations in May, producing the initial batch of T800s. That facility spans 12,000 square meters and operates a closed-loop manufacturing process — from incoming quality checks and component testing to final assembly and post-sales maintenance. It is capable of rolling out one humanoid robot every 15 minutes.
Zhongqing Robot focuses on developing and commercializing general-purpose intelligent robots and industry-specific solutions. Its portfolio includes the heavy-duty, full-size T800 humanoid; the lightweight companion SA02; the PM01 humanoid; and the JS01 quadruped robot. This diverse product matrix spans various forms and sizes, gradually penetrating typical applications like inspection and industrial operations.
Xiao Zhi Take: The cost and delivery advantages of Chinese manufacturing are now replicating the global expansion path once seen in consumer electronics — this time on the humanoid robot track.
NVIDIA unveils next-generation Jetson Thor computing platform
NVIDIA recently announced the T3000 and T2000, new modules built on the NVIDIA Jetson AGX Thor architecture designed to support mass-market robotics and edge AI applications. Both modules are slated for release in the first quarter of 2027.
The T3000 module delivers 865 FP4 TFLOPS of AI computing performance in a compact form factor roughly half the size and power consumption of the T5000. Meanwhile, the Jetson T2000 offers 400 FP4 TFLOPS of AI compute power and 16GB of memory, positioning it as an ideal entry-level solution for developers building visual AI agents, autonomous mobile robots, industrial arms, and other intelligent machines.
With this, NVIDIA’s scalable edge AI platform now spans a performance range from 70 TOPS to 2,000 TFLOPS, enabling developers to tackle nearly any edge AI workload.
Jetson AGX Thor is currently powering a new generation of humanoid robots and diverse robotic systems, accelerating adoption across multiple industries. Leading companies — including 1X, Agile Robots, Amazon Robotics, Boston Dynamics, FANUC, Hitachi, and Techman Robot — are already developing products based on this platform.
Xiao Zhi Take: By locking down the computing entry point for the robotics era with a unified platform, NVIDIA is essentially replaying its consumer GPU ecosystem success story.
Mobileye founder announces resignation as CEO, shifts focus to humanoid robots
Mobileye recently announced that founder and CEO Amnon Shashua plans to step down after nearly 30 years at the helm.
Shashua’s departure marks the biggest management shake-up in Mobileye’s history. He steered the company through Intel’s $15.3 billion acquisition in 2017 and its return to the public markets in 2022. However, since its relisting, demand has fluctuated, and Mobileye’s stock has tumbled 60%.
Image credit: Mobileye
Shashua will officially step down once a successor is named. During an earnings call, he stated that he intends to pivot toward long-term technology R&D, with a specific focus on humanoid robots — a sector he views as a critical frontier for innovation.
Mobileye is currently pushing ahead with its robotaxi business and robotics initiatives. Months ago, the company spent roughly $900 million to acquire Mentee Robotics, a humanoid robot startup co-founded by Shashua, where he previously served as co-CEO.
Xiao Zhi Take: Shashua is betting both his capital and his career on this. It is a strong signal that the humanoid robot sector is winning validation from top-tier experts.
Yuejiang Technology responds to whistleblower complaint ahead of H-share to A-share listing
On July 22, Shenzhen Yuejiang Technology Co., Ltd. faced an IPO review by the Shenzhen Stock Exchange. As the first "H-share to A-share" case in the Greater Bay Area following Shenzhen’s comprehensive reform pilot, Yuejiang moved from acceptance to review in just 86 days.
On the same day, Yuejiang released a voluntary clarification announcement to officially address an equity incentive dispute involving a former employee, identified only by the surname Song.
The announcement clarified three key points: Song’s identity and the origin of the equity, the history of the dispute, and the current status of legal proceedings.
Yuejiang stated that Song has not sought legal resolution over the past five years since leaving, but instead has filed complaints at critical junctures in the listing process, attempting to derail the IPO. The company emphasized that this is an internal matter within the employee stock ownership platform. It will not affect the stability of the equity structure or control rights, nor will it have a material adverse impact on operations, business development, or financial condition.
Yuejiang specializes in the R&D, production, and sales of collaborative robots, with products widely used in automotive, 3C electronics, new energy, and industrial manufacturing. Building on this, the company is rapidly entering the embodied intelligence arena. It has developed its own embodied large models and built a comprehensive product matrix covering humanoid, multi-legged, and dual-arm robots. Yuejiang is among the first domestic companies to enter mass production of humanoid robots and is one of the few global pioneers achieving large-scale deployment across industrial, commercial, and educational settings.
Xiao Zhi Take: Equity disputes surfacing at critical listing moments serve, in some ways, as a stress test for corporate governance transparency and crisis management capabilities.
Axera establishes wholly-owned subsidiary "Aixin Computing"
On July 24, Axera announced the launch of a wholly-owned subsidiary, Aixin Computing. The move marks a comprehensive upgrade of Axera’s strategic layout in AI computing, aiming to sharpen its focus on core AI computing capabilities and accelerate innovation in AI inference technology.
Image credit: Axera
Aixin Computing will focus on delivering leading AI inference solutions. By providing complete, standardized reference designs — such as compute cards — it aims to empower diverse "physical AI" scenarios, including embodied intelligence, smart offices, retail, education, and industry, helping clients rapidly achieve commercial viability.
To achieve this, Aixin Computing will fully inherit and deepen the parent company’s technical expertise in its self-developed "Aixin Tongyuan" mixed-precision NPU. Leveraging ultra-high computing efficiency and minimal power consumption, it aims to establish a technological lead in on-device and edge inference.
At the recently concluded WAIC 2026, Axera unveiled its Yuanxi series AI inference cards and inference servers. These were showcased alongside a comprehensive embodied intelligence solution and software adaptation ecosystem, demonstrating the company’s full-stack capabilities — from proprietary chip hardware to delivery of industry-specific scenarios.
Xiao Zhi Take: Beyond the high-end computing arena dominated by NVIDIA, local AI chip companies are carving out a second front in the edge inference market.
HiDream.ai raises 1.5 billion yuan, marking 2.1 billion raised in three months
HiDream.ai, a large model technology innovator, recently closed a 1.5 billion yuan Series C financing round. The round was jointly led by the Sichuan Revitalization Science and Technology Innovation Fund of the National Social Security Fund, ICBC Capital, Hongyi Asset Management, and Dunhong Capital. Participants included Xiamen ITG Capital, the Shanghai Film New Vision Fund, the Hubei Yangtze River Industrial Investment Group, and Huace Film & TV, with existing shareholders like Hefei Industrial Investment and Oriental Fortune Capital increasing their stakes.
With this, HiDream.ai has completed three funding rounds in the past three months, raising a total of over 2.1 billion yuan.
As one of the earliest Chinese firms to deploy multimodal generative AI, HiDream uses its proprietary HiDream series large models as a technological foundation. This base supports a diversified product matrix above and extends into a global commercial network, creating a closed-loop ecosystem where technology, product, and business drive one another.
Specifically, built on the HiDream-O1 series model, HiDream.ai has established a "1+1+3" strategy. This involves building one Token Hub platform to deliver standardized model capabilities, and atop that, driving AIGC product innovation across three key scenarios: commercial marketing, film production, and content creation.
Currently, HiDream.ai’s products serve over 50 million professional users and more than 40,000 corporate clients across over 100 countries and regions.
Xiao Zhi Take: The entry of media capital like Shanghai Film and Huace signals more than just financial investment — it is an injection of critical scenario resources.
Futuring Robot closes nearly 1 billion yuan in funding
On July 22, Futuring Robot announced the completion of a Pre-A funding round. Investors included internet giants, industrial leaders, major consumer brands, prominent funds, and existing shareholders, totaling nearly 1 billion yuan.
This marks Futuring Robot’s third announced funding round this year. In early January, the company closed a 200 million yuan angel round led by ZhenFund, with participation from Lianxin Capital, Yuanlai Capital, and existing shareholders. In early March, it raised several hundred million yuan in a new round led by Boyu Capital, with continued support from Lianxin Capital.
Image credit: Futuring Robot
Founded in 2022, Futuring Robot is a key player in China’s consumer-grade home robotics sector. Committed to a full-stack in-house development path, its core team comprises talent from top global universities and industry veterans, focused on refining embodied intelligence technology for home environments.
The company has launched two products, the F1 and F2. The F2 was officially released in April of this year. In May 2026, Futuring Robot unveiled its new "Self-Evolving WAM" robot brain and simultaneously deployed it within the F2.
Regarding deployment, Futuring Robot has entered over 500 homes on a paid basis, accumulating more than 50,000 hours of service with a 97% satisfaction rate. This long-term feedback from real households is driving continuous iteration of the F2 in interaction, task execution, home adaptability, and service stability.
Xiao Zhi Take: The collective entry of internet giants, industrial leaders, and consumer brands demonstrates that the value of home robotics is gaining recognition across diverse industries.
