PHYSIX FRONTIER EASTVOICE
2026.08.22 · Saturday
Issue 016
FIVE MINUTES A DAY, THE GLOBAL AI SHIFT
629%
Unitree intraday surge on listing day
¥444.9B
Unitree peak market cap on listing day
30,000 GitHub stars
DeepSeek Harness first day count
4.2 tons
Jushi Z20 autonomous light truck payload
Top Stories 3 · dual commentary ｜ Section Picks 22 · one-line takes
Unitree listing frenzy, SenseTime and Ant open-source releases, JPMorgan Alibaba Cloud valuation
SenseTime released the stable version of SenseNova U1.5 Lite, an open-source 8B natively unified multimodal model. It handles 3-4K context, juggles subject, quantity, spatial-relation, typography and layout constraints at once, outputs native 4K images, and was retrained around real visual-production tasks.
Editor's take｜Open-sourcing an 8B model that does production-grade 4K generation is a quiet shot at the mid-tier closed models. Small, cheap and controllable is where the market is heading.
Ant Group released the Ling-3.0-tiny-base and Ling-3.0-flash-base checkpoints and, unusually, is also publishing key training-process milestones so outsiders can trace how the models were built.
Editor's take｜Opening up intermediate checkpoints is a transparency move most labs won't touch. Ant is buying developer trust with receipts.
Alibaba Cloud plugged GLM-5.3 and DeepSeek-V4-Pro stable versions into its Qianwen AI platform, with APIs live for developers and both pitched at coding and agent workloads.
Editor's take｜The country's biggest model marketplace now openly resells its rivals' flagships. Aggregation is becoming Alibaba's moat, not any single model.
22 AI stories worth your time, each with a one-line take.
DeepSeek shipped rc.8 of its open-source agent workbench Harness just two days after rc.7. The August 13 debut passed 30,000 GitHub stars in its first day; the new version lets Harness orchestrate Claude Code and Codex as subordinate agents.
One-line take｜DeepSeek's endgame may not be another model, it could be the harness everyone's agents run inside. Control the orchestration layer and you tax the whole stack.
JPMorgan analysts argue Alibaba Cloud's current 12% operating margin understates maturity, with steady-state ROIC closer to 20% once recently deployed GPUs and data centers finish ramping from around 60% utilization.
One-line take｜The vintage-stacking math says Alibaba's CapEx binge converts into profit if utilization keeps climbing. That is the whole China AI capex trade in one chart.
UBTech Robotics' Walker embodied-intelligence model passed the CAC review and completed China's generative-AI service filing, among the first embodied models to clear the process.
What it means for you｜UBTech just got the paperwork that unlocks procurement.
One-line take｜A compliance stamp matters more than a benchmark score when selling robots to Chinese enterprises. UBTech just got the paperwork that unlocks procurement.
USTC robotics PhD Duan Yifan's Shanghai startup Lingxi Zhiyong argues pretrained models alone fail on real production lines, where deployment success runs under 60%, and sells a model-plus-harness architecture instead.
One-line take｜The bottleneck in industrial embodied AI is wiring, not weights. Anyone who treats the factory as a pure model problem will keep losing the last mile.
Chinese media report banks in Guangzhou now factor companies' AI compute consumption into lending decisions, while Wall Street packages GPU assets for pension funds. Compute is being securitized as an asset class on both shores.
One-line take｜When banks underwrite GPU rental contracts, compute has officially become real estate. The question shifts from "is it overpriced" to "who's the landlord".
MiningLamp Technology (2718.HK) and Hikrobot co-exhibited at WRC 2026, pairing AI agents with embodied machines for commercial service deployments.
One-line take｜The smart move at robot shows this year is software-plus-hardware pairing. Agents that never move and robots that never reason both lose deals.
Yangtze Memory, China's NAND champion, had its STAR Market IPO application accepted, aiming to raise ¥33 billion, with CITIC Securities and CSC as sponsors.
One-line take｜The AI trade needs memory as much as compute. A ¥33 billion raise for YMTC is the market pricing storage as strategic infrastructure.
TMTPost reports on the AI-for-science wave, where domestic accelerators increasingly power ocean forecasting and research workloads, exemplified by the full-stack domestic "Pangu" marine forecasting model.
One-line take｜Science is the least forgiving benchmark for domestic chips, and it's where they're quietly gaining share. That is a more credible story than any spec sheet.
Unitree listed on the STAR Market: IPO price ¥150.80, opened at ¥1,100 (up 629%), intraday market cap peaked around ¥444.9B, and the stock closed up roughly 460%. Founder Wang Xingxing looked visibly uneasy at the bell-ringing ceremony.
One-line take｜A valuation of that size prices in years of flawless execution. The market is buying the category's future, and the founder's expression says he knows the scrutiny that comes with it.
At WRC 2026, Chaowei Power demonstrated what it calls the world's first complete autonomous ping-pong match between humanoid robots, alongside its SMASH 2.0 system, the KAI world model, and a KAIBot body with 117 degrees of freedom.
One-line take｜Table tennis is the new backflip, a demo that forces real perception latency and body control. Fun to watch, harder than it looks, and every lab knows it.
UBTech displayed humanoid lines for factories, commercial services and home consumption at WRC 2026, promoting full-chain deployment readiness across the three scenarios.
One-line take｜After clearing its embodied-model filing with the CAC, UBTech is playing the compliance-plus-lineup game. Certifications and catalogs, not demo stunts.
QbitAI digs into Qianxun Intelligent's full-stack play at WRC, arguing the quiet, repetitive demos reveal more about real embodied capability than crowd-pleasing stunts.
One-line take｜At this stage of the industry, boring reliability beats spectacular novelty. Ask how many hours the demo ran, not how high it jumped.
Zhongjian Technology's subsidiary Huazhijian unveiled ZERO, a roughly two-meter-tall concept humanoid at WRC 2026 that drew the biggest crowd on the floor.
One-line take｜A giant concept robot is a branding statement, not a product. Fine for buzz, but the gap between concept scale and deployment reality is exactly where the hype hides.
China's SAMR announced phase-two "quality strong chain" projects covering embodied-intelligence robots and solid-state batteries as strategic quality-improvement targets.
What it means for you｜Embodied robots just got put on the industrial-policy map.
One-line take｜When the quality regulator names a category, procurement standards and subsidies usually follow. Embodied robots just got put on the industrial-policy map.
TMTPost profiles the new livestream trade of robot maintenance, where technicians who once fixed phones and computers are retraining to service robots, with "40 days to learn" courses popping up.
One-line take｜When repair classes for robots start selling, the installed base has crossed into real-world scale. After-sales is the sector's quietest leading indicator.
Unmanned-delivery leader Jushi Intelligent and commercial-vehicle maker Yutong unveiled the Jushi Z20 autonomous light truck with a 4.2-ton payload and 19.32 m³ cargo volume.
One-line take｜Last-mile delivery moving up to 4.2-ton trucks is the tell. Autonomous freight is skipping the passenger-car detour and going straight where the margins are.
The Qijing GX7 made its real-car debut at the Chengdu Auto Show with dual 15.6-inch screens, one-key-to-bed rear seats and L3-ready architecture with eight-lane redundancy.
One-line take｜"L3 pre-wired" is becoming the standard spec for Chinese family EVs, a hedge that pays off the day regulations open the lane.
XPeng says the redesigned GX topped 15,000 deliveries within two months of launch, with 2,000+ orders from the UAE region, riding its second-generation VLA platform.
One-line take｜Overseas order books validate the VLA-first product bet. Selling AI driving to the Gulf is the fastest revenue path China's EV makers have found.
The Changan Mazda EZ-60 LiDAR edition entered pre-sale with 1 lidar, 3 mmWave and 12 ultrasonic sensors, plus highway-and-city navigation assistance.
One-line take｜A joint-venture brand shipping a native lidar trim at pre-sale is the price war reaching the ADAS bin. Mid-tier autonomy is now a spec sheet, not a premium.
At WRC 2026, Xingdong Jiyuan founder Chen Jianyu argued VLA models are transitional and embodied intelligence is moving to world-model-driven architectures.
One-line take｜Every VLA lab is now saying the same thing: actions alone don't generalize, world prediction does. The narrative shift is real, the architecture race is wide open.
Xinghaitu CEO Gao Jiyang says the ultimate business model for embodied intelligence is selling physical-world tokens, metering real-world interaction as a consumable.
One-line take｜Framing robot data as tokens is a pricing model dressed as philosophy. Whatever you call it, metered physical intelligence is how this sector eventually makes money.
In a Leiphone interview at IJCAI 2026, ASU professor Wei Hua warns foundation-model agents risk repeating reinforcement learning's history, failing to bridge the simulation-to-reality gap.
One-line take｜The sim-to-real gap was RL's graveyard, and today's agent benchmarks have the same smell. Worth listening when a researcher says the field keeps re-buying the same lesson.
📖 The full EastVoice issue on the Chinese forum
📈 Physix Frontier · Alpha — the investor's read
Headline teardowns · curated signals · market moves. Five minutes to find the alpha of the AI era →