StockDuty ← Dashboard Sep 23, 2026 10:14 PM ET
CatalystChina AI Digest

Wed, Sept 23, 2026: Huawei’s large-model twin stars raised hundreds of millions to build a physical foundation model. DeepSeek published a new agent-training paper with Liang Wenfeng credited as autho

Alibaba’s Qwen platform fully upgraded model services, agent services, and AI apps. Huizhi Intelligence launched Hellome, China’s first FDE direct-connect agent service platform, to compress AI delivery. Huawei is the chip-to-model play. It’s using its large-model teams to build a physical foundation model. China’s embodied-intelligence top team says the sector is entering large-scale implementation. Robotics is moving from demo to deployment. DeepSeek wants large-scale agent training. Huawei wants physical foundation models. Qwen wants agents inside every app. China is racing to own the agent layer before the interface war ends.
Alibaba’s Qwen platform fully upgraded model services, agent services, and AI apps. Huizhi Intelligence launched Hellome, China’s first FDE direct-connect agent service platform, to compress AI delive
Sources
digest