CCID reports that AI inference data volume in China has surpassed training data volume, reaching 175 trillion daily token calls

The topic of recursive self-improvement—the ability of AI to assist in creating the next AI—has become one of the most prominent among model developers in China this autumn. According to a survey by Leiphone, there is a general trend: developers are preventing AI from rewriting model weights. Instead, labs direct AI to work with training pipelines, inference systems, GPU kernels, sandboxes, and agentic mechanisms.

In a September study prepared by 35 researchers from Shanghai Jiao Tong University, Tsinghua, ByteDance, SenseTime, and others, five levels of self-improvement were proposed, with most advanced systems situated between the first and second levels.

Examples of Self-Improvement Implementation

Zhipu demonstrated the most obvious production example. An infrastructure agent managed by GLM-5.3 created service infrastructure for GLM-5.3-Flash on a cluster comprising approximately 100,000 Chinese-made accelerators within two weeks, preserving fixes as reusable templates. Subsequently, in its HarnessPro work with Tsinghua, agents were able to rewrite their own management mechanism.

DeepSeek began work with a training environment. Its DSec cluster, developed with Tsinghua's involvement, generates about 3 million sandboxes daily, allowing agents to create environments for training stronger agents. The DeepSeek Harness system treats tools and even its core runtime as replaceable plugins without requiring a restart or rollback.

ModelBest's Forge Engineering project uses AI to write the software required for creating AI. It reported that the AI-generated training structure, ForgeTrain, matched Megatron-LM in eight hours, and the ForgeKernel operators surpassed the official leaderboards of Huawei and NVIDIA.

Moonshot AI relies on large-scale reinforcement learning. Training Kimi K3 required over 51 million sandboxes on Firecracker microvirtual machines, as well as an in-cluster agent that recovers failed runs across 26 scenarios, and an agent reward model that evaluates training recipes. Risks emerged in October when sub-agent calls in K3.1 Pro spiraled out of control, causing Kimi to drop the monthly quota for 25 percent of participants.

Alibaba reported that Qwen3.8-Max independently ran its training pipeline for over a month, completing 33 iterations, and Qwen serves as the foundation for external initiatives such as Apex-35B-V1.0. ByteDance's Seed team proposes a self-evolving model identifier, arguing that most systems are merely semi-closed loops because evaluation rules are still set by humans. Xiaomi broadcast six days of reinforcement learning on MiMo-V2.6.

Despite the existence of engineering cycles, a complex challenge remains, noted by Zhipu Chief Scientist Tan Ze: achieving reliable determination by the model of when to stop, when to correct itself, and how to evaluate its own learning.

This year, the number of ultra-wealthy individuals in China reached an all-time high as tech magnates, including the founder of DeepSeek, joined the closely tracked list amid the country's expanding presence in artificial intelligence.

According to the annual Hurun China Rich List published on Tuesday, the number of people whose net worth exceeds 5 billion yuan ($750 million USD) grew to a record 1,529, significantly surpassing the previous high of 1,460 recorded in 2021.

Lian Wenfeng, founder of the AI startup DeepSeek, debuted at seventh place with an estimated net worth of $41.7 billion—one of 263 newcomers, most of whom come from sectors such as semiconductors, AI, and robotics.

His wealth increased thanks to DeepSeek's R1 reasoning model, which caused a stir in the industry last year and triggered a sell-off in American tech stocks.

This year's list reflected China's concept of a 'dual-speed economy,' where technological fortunes grew against a backdrop of declining traditional wealth, noted Rupert Huggeverf, a senior researcher at the Hurun Research Institute that publishes the list.

Of the top 100 individuals, 60 percent amassed their fortunes in sectors designated by China under a plan launched in 2015 to achieve technological leadership within ten years.

Huggeverf stated in a press release: 'It was a year when 'Made in China 2025' became a reality.'

Zhang Yiming, founder of ByteDance and China's richest person in 2024, reclaimed the top spot with an estimated net worth of $99.7 billion—a record for this list. He overtook John Shanshan, founder of Nongfu Spring, known as the 'King of Bottled Water.'

ByteDance, which includes TikTok and Douyin, has faced increasing regulatory pressure globally; however, its AI chatbot Doubao boasts nearly 420 million monthly active users, significantly outpacing local competitors.

Robin Zeng from CATL, Pony Ma from Tencent, William Ding from NetEase, and Chen Tianshi from Cambricon also made the top 10.

It is reported that DeepSeek is close to completing a new funding round of 100 billion yuan—a record for the Chinese AI industry—which could value the company at approximately 500 billion yuan.

Hurun's list of China's richest women, published on Wednesday, showed that Zhou Cunfan, founder of sensor screen manufacturer Lens Technology, maintained her first-place position with a net worth of $18.9 billion.