China’s EV and smart-driving sector closed August with three telling signals: GAC Aion shocked the domestic ADAS leaderboard with a low-cost Aion N60 running WeRide’s WRD 3.0 software, Huawei-backed Shangjie validated high-speed chassis-and-perception safety in extreme wet-road testing, and embodied robotics startup Xuanchuang Robotics raised fresh funding to automate hazardous industrial sites. Taken together, the news shows how China’s next phase of competition is no longer just about battery range or brute-force compute—it is increasingly about algorithm efficiency, system integration, and safety in real-world edge cases.
Aion N60’s Breakthrough Shows Software Can Still Beat Hardware
The biggest surprise came from D1EV’s latest intelligent-driving competition update, where the GAC Aion N60 emerged from relative obscurity to win the Guiyang round.
Before the event, the N60 had rarely been seen as a favorite. It had consistently trailed its in-house rival, the Exeed Sterra ET, which also uses a WeRide-based solution but entered mass production more than a year earlier. Yet just 10 days before the Guiyang event, GAC Aion quietly pushed a new OTA update to users: AEP 2.3.0.0, corresponding to WeRide WRD 3.0.
That update specifically improved performance in scenarios that mattered most in Guiyang’s mountainous urban environment:
- Steep uphill roads
- Consecutive tight bends
- Narrow-road passing
- Tunnel light-to-dark transitions
- Complex intersections
- Cut-in and merge conflicts
The result was dramatic. The Aion N60 posted:
- 117.81/120 in qualifying, the highest score of the event
- Zero human takeovers across the full route
- 91.96 points in the final, the only car to score above 90
- A margin of roughly 15 points over the third-place Luxeed R7
What makes the win especially notable is the car’s modest hardware profile. According to the report, the top-trim Aion N60 costs about RMB 120,000 and runs on a Qualcomm 8650P with roughly 200 TOPS sparse compute, paired with one 192-line LiDAR.
That has prompted a wider debate in China’s EV community: is smart driving entering a phase where software efficiency matters more than headline compute numbers?
Why Guiyang Matters More Than a Single Trophy
Guiyang is not just another city-course result. Mountainous roads, tunnels, narrow lanes, and aggressive local traffic patterns create a far tougher testbed than standard urban NOA demonstrations.
In that context, Aion’s performance suggests that:
- OTA cadence can be a decisive competitive weapon
- Algorithm tuning for local scenarios remains critical
- Mid-priced EVs can deliver premium smart-driving experiences if the software stack is mature
D1EV even quoted a judge calling the result China smart driving’s “DeepSeek moment,” comparing it to the AI industry’s shift toward more efficient models rather than pure hardware scaling. The analogy may be imperfect, but the underlying point is valid: software optimization is becoming a more visible differentiator in Chinese autonomous-driving competition.
Huawei ADS Regains Momentum—But Still Has Work To Do
Huawei’s driver-assistance stack also had a meaningful August. In the Guiyang competition, the Luxeed R7 returned Huawei ADS to the top three for the first time in roughly a year.
The story was unusually dramatic. During qualifying, the Luxeed R7 was still running ADS 4.1.5 and advanced ahead of a newly updated Aito M9 equipped with ADS 5.0. On the morning of the final, the R7 reportedly received an ADS 5.0 OTA push, updated with the organizers’ approval, and then went on to secure Huawei’s best placing in the competition over the past 12 months.
According to D1EV, the upgrade delivered measurable improvements:
- All-scenario penalty points fell from 30 at a previous Jinhua round to 15 in Guiyang
- Decision-making and lane-change behavior improved
- But efficiency still remained a weakness, with 19 points deducted
That mirrors a broader pattern seen across Huawei’s ADS-equipped brands, including Aito, Luxeed, Avatr, and Voyah. On normal roads, Huawei remains highly competitive. In more extreme scenarios—especially industrial parks, complex roundabouts, narrow road U-turns, and mountainous routes—its system can still trigger downgrades or fail to fully generalize.
Shangjie Z7 and Z7T Highlight Huawei’s Cross-Domain Safety Strategy
If the competition result showed Huawei making progress in intelligent driving, a separate validation for the Shangjie Z7 and Z7T underlined where the company may still hold an edge: integrated active safety.
According to a China Automotive Engineering Research Institute certification cited by D1EV, the Shangjie twins passed two high-difficulty wet-road chassis safety tests.
Wet-road validation results
-
Emergency obstacle avoidance after a disappearing lead vehicle
- Simulated rainy/foggy expressway scenario
- Assisted driving activated
- Vehicle completed emergency avoidance at 121 km/h
-
High-speed tire blowout in a wet curve, followed by a moose test
- Wet asphalt friction coefficient: about 0.7
- Curve radius: 125 meters
- Lane width: 3.75 meters
- Blowout occurred at speeds above 90 km/h
- Vehicle remained stable and completed the evasive maneuver
Huawei attributes the result to coordination between its Tuling platform and ADS fused perception system, effectively linking sensing, decision-making, and chassis control.
The hardware specification is also substantial. The Shangjie Z7/Z7T comes standard with:
- Huawei ADS 4.1
- 896-line dual-path imaging LiDAR
- 1 rear high-precision solid-state LiDAR
- 5 4D millimeter-wave radars
- Multiple HD cameras
- 27 sensors in total
On the product side, the vehicles are positioned as midsize-to-large EVs:
- Z7 starting price: RMB 219,800
- Z7T starting price: RMB 229,800
- Wheelbase: 3,000 mm
- Maximum CLTC range: 905 km
- 0-100 km/h acceleration: 3.44 seconds for AWD versions
- June deliveries: over 10,000 units
This matters because Huawei’s competitive case is increasingly broader than just NOA. The company is pushing a full-stack proposition that combines perception, control, cockpit software, and chassis dynamics.
Competition Snapshot: Who’s Leading China’s Smart-Driving Rankings?
D1EV’s latest six-month final-round ranking offers a useful snapshot of where major Chinese smart-driving stacks stand today.
| Rank | Brand / Solution | Final Appearances | Average Score / Key Note |
|---|---|---|---|
| 1 | Exeed (Bosch + WeRide) | 6 | Only solution averaging above 100; lead shrinking |
| 2 | Exeed (Horizon) | 3 | Just 0.35 points behind No. 1, but with limited sample size |
| 3 | Zeekr (Qianli Haohan) | 6 | Average 80.64; consistency and safety remain concerns |
| 4 | Avatr (Huawei ADS) | 3 | Average 80.63; strongest Huawei-affiliated ranking |
| — | GAC Aion (WeRide) | 1 | 91.96 in Guiyang final; not yet officially ranked |
| — | Aito (Huawei) | 1 | 83.82 in one final; not yet officially ranked |
Two trends stand out.
First, WeRide’s stack looks increasingly dangerous to incumbents, especially when OEM partners update quickly. Second, Huawei remains competitive but still needs stronger scenario generalization in edge cases.
A Reminder That Smart Driving Can Still Fail in Strange Ways
The same D1EV report included a cautionary anecdote that says a lot about the current state of advanced driver assistance.
The Chery Fulwin T9L, which had won the Kunming round with 100.87 points, collapsed to 57.34 points in Guiyang—last place. According to the report, one contributing factor was that protective covers had reportedly been added to front and rear cameras by the owner, affecting system performance.
The score drop was severe:
- Repeated “stuck and failed to reroute” penalties
- A side-scrape incident under NOA leading to a 10-point accident deduction
- Additional deductions for violations such as wrong-way driving
Whether this was the decisive factor or not, the episode underlines a critical truth: sensor cleanliness, calibration, and hardware condition remain fundamental. In real-world use, advanced ADAS is not just software-defined—it is still sensor-dependent.
Robotics Funding Shows Where China’s EV-Tech Ecosystem Is Heading Next
Away from passenger cars, another D1EV item highlighted a fast-growing adjacent sector: embodied intelligence for industrial automation.
Xuanchuang Robotics, a Chinese startup focused on hazardous industrial environments, announced an A1 funding round worth tens of millions of yuan. Investors include:
- Qianhai Ark
- Xihu Sci-Tech Venture Capital
- Guangyang Co.
The company said the capital will go toward:
- Iterating its XBOT-AEGIS embodied intelligence system
- Developing multiple specialized robot form factors
- Expanding into domestic and overseas industrial scenarios
Its target sectors include:
- Oil and gas
- Chemicals
- Nuclear power
The startup’s architecture is notable because it echoes the same engineering priorities now shaping smart EVs: autonomy paired with hard safety controls. XBOT-AEGIS uses a three-layer decoupled structure:
- Upper layer: task understanding and decision-making
- Middle layer: motion planning and execution reasoning
- Bottom layer: safety guardrail that can intercept unsafe commands within milliseconds
Founded in December 2022, Xuanchuang says it has already delivered products to customers including PetroChina and Wanhua Chemical, while also winning overseas orders in South Sudan and Malaysia.
For EV watchers, the significance is broader than robotics alone. China’s leading mobility technologies—perception, control stacks, safety middleware, and autonomous decision systems—are increasingly spilling over into industrial automation. The boundary between “smart EV tech” and “physical AI” is getting thinner.
Why This Matters Globally
For international readers, these stories collectively show how China’s EV race is evolving in three important ways.
1. OTA speed is becoming a strategic moat
Aion’s result in Guiyang was heavily tied to a software update delivered just days before the event. Fast iteration is now a competitive weapon, not just a customer-convenience feature.
2. Safety integration is the new premium benchmark
The Shangjie Z7/Z7T tests reinforce an industry shift from standalone ADAS features to tightly integrated perception-plus-chassis control. This is where next-generation safety leadership may be decided.
3. China’s autonomy stack is expanding beyond cars
From EVs to industrial robots, Chinese companies are building reusable autonomy capabilities across sectors. That could strengthen supply chains, reduce development costs, and accelerate commercialization.
What To Watch Next
Several near-term questions now matter for China’s smart EV market:
- Can GAC Aion repeat its Guiyang result and prove the N60 was more than a one-off breakthrough?
- Will Huawei ADS 5.0 close the remaining gap in difficult park, roundabout, and mountain-road scenarios?
- Can Exeed defend its top ranking if its update cycle remains slower than rivals’?
- Will players like Zeekr improve consistency and reduce safety-related deductions?
- And beyond vehicles, how quickly will embodied AI firms like Xuanchuang Robotics turn pilot deployments into scale businesses?
The bigger picture is clear. China’s EV contest is entering a more mature phase where winners will be defined less by spec-sheet bravado and more by the quality of their software iteration, sensor fusion, and safety engineering under pressure. August’s headlines did not just produce a few isolated surprises—they offered a glimpse of the next competitive map for intelligent mobility.
