PHYSIX FRONTIER EASTVOICE
2026.10.29 · Thursday
Issue 069
FIVE MINUTES A DAY, THE GLOBAL AI SHIFT
¥1.817 billion
Unitree nine-month revenue, up 55.71% year on year
2027
Volkswagen Group China target date for L3 functions
60%
Intelligence Index v5 weight on hidden test sets
88
Infinigence first places on NVIDIA SOL-ExecBench categories
Top Stories 3 · dual commentary ｜ Section Picks 24 · one-line takes
Unitree's first post-IPO quarter, Volkswagen's 2027 L3 date and chip, and Artificial Analysis rebuilding its index with hidden answers.
Unitree reported nine-month revenue of ¥1.817 billion, up 55.71%, and net profit of ¥387 million, up 270.54%. The third quarter is where the line bends. Revenue grew nearly 70% year on year while quarterly net profit fell, and R&D spending rose roughly 2.5x. The nine-month figures look strong because the first half carried them.
What it means for you｜Unitree's quarterly profit fell while R&D rose roughly 2.5x, so investors will forgive that for two more quarters before R&D must show up as units shipped.
Editor's take｜A hardware company that doubles R&D while quarterly profit slips is still buying its way into a product cycle. Investors will forgive that for two more quarters. After that, the R&D has to show up as units shipped.
Group vice president Gu Jie, speaking at the SAECCE conference, said the solid-state battery for XPeng's IRON humanoid has entered small-volume production and the robot is now in the stage before SOP. Solid-state buys runtime per charge and frees torso volume that a conventional pack eats.
What it means for you｜XPeng's solid-state cell line is run by a carmaker, so yield data will be audited to automotive standards rather than to a demo schedule.
Editor's take｜The notable part is not the chemistry. It is that a carmaker is running the cell line, which means the yield data will be audited to automotive standards rather than to a demo schedule.
The two announced the "Shanyu Venture Program", aimed at seed and early rounds across embodied AI, robot applications and upstream components. Unitree frames its role as a co-founder rather than a passive cheque.
What it means for you｜Unitree funding its own supply chain means its balance sheet now has a stake in deciding which competitors survive.
Editor's take｜When a hardware maker starts funding its own supply chain, it is buying priority on parts before they get scarce. It also means Unitree's balance sheet now has a stake in deciding which competitors survive.
24 AI stories worth your time, each with a one-line take.
One of Sichuan's largest community convenience chains put humanoid robots behind the counter in early 2025 to sell drinks and snacks. QbitAI's account of the wind-down is a list of ordinary failures: slow grasping, aisles blocked by a machine that would not yield, shelves no demo had prepared it for.
What it means for you｜The Sichuan convenience chain's robot clerks failed on slow grasping and blocked aisles, and the graveyard in retail is now longer than the pilot list.
One-line take｜Retail was billed as the easy on-ramp, with fixed shelves, repeatable SKUs and hours you can measure. The graveyard in that segment is now longer than the pilot list.
Unitree rose 16% in a session and its market cap returned to ¥200 billion. In the same stretch, Shenlan Robotics was in bankruptcy proceedings with wages unpaid. TMTPost's reading: the companies booking actual embodied-AI revenue are largely automakers and their suppliers, while humanoid makers book valuations.
What it means for you｜Unitree's market cap returned to ¥200 billion while Shenlan Robotics was in bankruptcy with wages unpaid, which is what a hardware shakeout looks like before volume arrives.
One-line take｜One company's market cap and another's bankruptcy filing, in the same week, in the same sector. That is what a hardware shakeout looks like before volume arrives.
At its first China technology day, Volkswagen Group China said it plans to have L3 functions ready by 2027 and confirmed it is developing C7H, a system-on-chip built to support up to L4. The chip work sits alongside the group's VEGA platform and electronics architecture.
What it means for you｜Volkswagen's 2027 L3 date is a procurement schedule, and whoever owns the SoC owns the update cadence for the next decade of that car line.
One-line take｜A 2027 L3 date from a legacy OEM is a procurement schedule, not a product promise. The chip matters more. Whoever owns the SoC owns the update cadence for the next decade of that car line.
Dongchedi's test video, published October 8, showed three new Zunjie V800s in 100 km/h emergency stops. The pedal bracket failed on the second, third and fourth full presses, at the same point, on cars from different production batches. The case is being read as a test of Huawei's car BU, which is trying to become the Bosch of Chinese auto.
What it means for you｜Three Zunjie V800s failing identically puts the supplier qualification process under review, not the driver's foot.
One-line take｜A pedal bracket is a cheap stamped part with a ten-year liability tail. Three batches failing identically puts the supplier qualification process under review, not the driver's foot.
Video of a Yijing X9 burning in Taizhou, Zhejiang circulated on October 28. Brand general manager Zeng Qinglin said the car had already braked to a stop when a loaded dump truck hit it from behind, and that none of the three people inside were injured.
What it means for you｜The Yijing X9 fire is a crash-structure story, and answering within hours is the part other OEMs should copy.
One-line take｜EV fire footage gets read as a battery story. This one is a crash-structure story, and answering within hours is the part other OEMs should copy.
Reuters, citing Manager Magazin, reported that VW is weighing whether to put the E/E architecture currently used only in China into European mass-market models. The stack was developed with XPeng and CARIAD China. The argument for moving it west is cost.
What it means for you｜The cheap, shipped software stack is now Chinese and Europe is the follower market, reversing twenty years of technology transfer from Wolfsburg.
One-line take｜The direction of technology transfer has flipped. For twenty years the platform came from Wolfsburg and was localized in China. Now the cheap, shipped software stack is Chinese, and Europe is the follower market.
Cailianshe reported that Baidu plans to disclose progress on a world model in the coming days, with autonomous driving as the first application. A driving world model predicts how a scene will evolve instead of only labelling what is in it, which is what a planner needs at an unprotected intersection.
What it means for you｜Baidu's edge is operational data from Apollo's live zones, which a driving world model needs and cannot be bought.
One-line take｜Baidu's edge here is operational data from Apollo's live zones, which is precisely what a driving world model needs and precisely what cannot be bought.
Leiphone walks through the technical report behind FluxVLA Engine, an open engineering stack for vision-language-action models. The pitch is that teams should reuse the data pipeline, training harness and deployment tooling, and spend their engineers on the model instead.
What it means for you｜Shared infrastructure lets embodied teams stop rewriting the same data loader and evaluation rig, the least glamorous fix that pays back fastest.
One-line take｜Every embodied startup rewrites the same data loader and the same evaluation rig, then wonders why it ships slower than a lab with a platform group. Shared infrastructure is the least glamorous fix and the one that pays back fastest.
LiberAI, founded in December 2025 and working on physical-world models, took a Sequoia China angel round in March 2026 and four follow-on rounds within months. The company came out of a Tsinghua cohort. The article's frame is that the entry ticket for this category has changed.
What it means for you｜Money is chasing the few dozen people in China who have trained a world model at scale, not the idea.
One-line take｜Stacking four rounds into a seed-stage company is a supply-constrained bet. There are perhaps a few dozen people in China who have trained a world model at scale, and the money is chasing them, not the idea.
WIKA Mobile Control spent the second half of 2026 at Huawei Connect and Alibaba's Yunqi conference pitching datasets that connect construction machinery to embodied models. The target is machine autonomy in an industry where the operator is the bottleneck and the site is already fenced.
What it means for you｜On construction sites the tasks repeat and a machine running a full shift without a driver pays for itself in one season.
One-line take｜Construction is a better first market than the home. The site is controlled, the tasks repeat, and a machine that runs a full shift without a driver pays for itself in one season.
Intelligence Index v5 is the largest rebuild since v4 landed in January 2026. 60% of the evaluation weight now rests on private or answer-hidden test sets, a direct response to benchmark contamination. On the new FrontierCode 2.0, DeepSeek's V4.1 Flash 1017 is the highest-scoring open-weight model.
What it means for you｜Hiding the answers moves the advantage to whoever can afford fresh evaluation data, rather than ending the arms race.
One-line take｜Contamination was the open secret behind half the launches this year. Hiding the answers does not end the arms race, it moves the advantage to whoever can afford fresh evaluation data.
Developers found three previously unpublished model strings, deepseek-v5, deepseek-v5-open and deepseek-v5-next, in the provider spec test file of the official DeepSeek Harness repo, around line 154.
What it means for you｜DeepSeek now ships a harness, an agent runtime alongside the weights, which is where the lock-in actually lives.
One-line take｜A test fixture is a weak signal and cheap to plant. The useful read is that DeepSeek now ships a harness, meaning an agent runtime alongside the weights, which is where the lock-in actually lives.
Moonshot released K3.1 Pro on October 24 with flagship-tier claims and API prices roughly one third below the previous generation. Users then found that token consumption per task ate the discount, and the price-per-useful-answer argument fell apart on social feeds.
What it means for you｜Cutting price per token while raising tokens per task is a margin transfer, and buyers have started pricing the outcome.
One-line take｜Cutting the price per token while raising tokens per task is a margin transfer, not a price cut. Buyers have started pricing the outcome, and that is bad news for anyone selling on sticker price.
Ant Group's Ling team released Ling-3.1-flash-1019 as open weights on October 28, positioned for personal-agent workloads rather than chat. The pitch is tool use and multi-step tasks at a cost that lets an assistant run all day without the bill becoming the product decision.
What it means for you｜A cheap open flash model aimed at personal agents bets the assistant layer, not the frontier, is where Chinese labs can win.
One-line take｜Personal agents live or die on cost per hour of attention. A cheap open flash model aimed at that job is a bet that the assistant layer, not the frontier, is where Chinese labs can win.
Moore Threads released Megatron-MUSA-Titan, a training infrastructure project built on its own full-function GPUs, targeting trillion-parameter MoE training on clusters of 10,000 cards and covering both pretraining and reinforcement learning.
What it means for you｜Open-sourcing the trainer gets other people to debug your interconnect for free, the fastest route to being taken seriously by model teams.
One-line take｜Software is where Chinese GPU vendors kept losing. Open-sourcing the trainer gets other people to debug your interconnect for free, which is the fastest route to being taken seriously by model teams.
On SOL-ExecBench, an NVIDIA Research benchmark for GPU operator performance, Infinigence placed first in 88 categories. Part of the work involved agents participating in low-level kernel optimization rather than humans tuning by hand.
What it means for you｜A team using agents to search the optimization space suggests at least part of the CUDA moat is software-addressable now.
One-line take｜Kernel work was supposed to be the moat around CUDA. A team using agents to search the optimization space suggests that at least part of that moat is software-addressable now.
Tencent released Hunyuan Hy-SimulMT, an open-source simultaneous machine translation model that the company says holds accuracy while keeping average response time to about 3 seconds.
What it means for you｜Three seconds is roughly the width of a spoken clause; below that simultaneous translation feels like subtitles, above it listeners lose the thread.
One-line take｜Three seconds is roughly the width of a spoken clause. Below that, simultaneous translation feels like subtitles. Above it, listeners lose the thread of the meeting.
ByteDance's Doubao Work upgraded its PC-control feature with better click precision and speed, and added the ability to record a demonstration of a desktop task and save it as a reusable skill. Users can narrate while they demonstrate.
What it means for you｜Recording a workflow is the cheapest form of programming handed to an office worker, but the skill breaks the moment an app moves a button.
One-line take｜Recording a workflow is the cheapest form of programming ever handed to an office worker. The failure mode is the one macros had in 1998: the skill breaks the moment an app moves a button.
WorkBuddy added remote control, built with support from Tencent's Edge One team. The phone shows a live view of the desktop and accepts direct input, so a user can watch a running task instead of reading its output after the fact.
What it means for you｜People do not trust what they cannot watch, so a live screen is a trust feature wearing the costume of a convenience feature.
One-line take｜Agent products keep rediscovering one thing: people do not trust what they cannot watch. A live screen is a trust feature wearing the costume of a convenience feature.
Today AI, part of Huihe Intelligence in Shanghai's Xuhui district, closed an angel round above $200 million for a software-only personal agent. Flowtica AI raised tens of millions of yuan for a magnetic clip-on device in the same category.
What it means for you｜A $200 million angel round for an assistant with no hardware bets on distribution, and the real contest is which app gets permission to act on your behalf.
One-line take｜A $200 million angel round for an assistant with no hardware is a bet on distribution, not technology. The real contest is which app gets permission to act on your behalf, and that is a platform fight.
The State Council General Office released a document on adapting to AI's effect on jobs, with 18 tasks including softening displacement, building employment-risk assessment, and adjusting income distribution and social security.
What it means for you｜The risk-assessment mechanism makes AI deployment inside Chinese firms reviewable before it happens, which no other major economy has put in writing.
One-line take｜The risk-assessment mechanism is the item to watch. It makes AI deployment inside Chinese firms reviewable before it happens, which no other major economy has put in writing.
A September 29 provincial newspaper piece quoted a Lao Gan Ma executive saying its AI visual inspection system was fully live, raw materials 100% traceable, and the defect rate down to 0.02%. Readers turned the numbers into a question about what the company had been shipping before.
What it means for you｜Announcing a 0.02% defect rate invites buyers to ask what the company had been shipping before.
One-line take｜Consumer brands keep learning that an efficiency metric doubles as a confession. Announcing a 0.02% defect rate invites buyers to ask what it used to be.
A software firm with fewer than a hundred staff in a fifth-tier Shaanxi city spent about ¥700,000 from August on a system pitched as lead generation, customer service, recruiting and decision support. What arrived, per the account, was a set of demos that never touched the actual workflow.
What it means for you｜Vendors sold an outcome, shipped a dashboard, and left the integration to a company with nobody to do it.
One-line take｜This is the forward-deployed-engineer model without the engineer. Vendors sold an outcome, shipped a dashboard, and left the integration to a company with nobody to do it.
Kingdee said its Lingji agent platform has 3,706 paying customers, with 26 signed at launch and more than 40 agents live; AI-native product revenue grew 189% year on year. The positioning is ERP as the context base an agent reads before it acts.
What it means for you｜ERP vendors hold twenty years of how a specific company actually runs, which agent startups cannot buy; whether they can expose it without breaking their permission model is open.
One-line take｜ERP vendors hold the one asset agent startups cannot buy, which is twenty years of how a specific company actually runs. Whether they can expose it without breaking their own permission model is the open question.
📖 The full EastVoice issue on the Chinese forum