XPeng puts its humanoid robot AI brain into a 75k Euro electric SUV
XPeng just launched the G9L SUV in Paris. It uses the exact same Turing AI chip and foundation model as their humanoid robot.
Business · Source: Morningstar (PR Newswire)
What happened
XPeng just revealed the G9L electric SUV at the Paris Motor Show. The starting price in France is 74,990 Euros. But this is not just another car launch. The Chinese automaker is officially branding itself as a Physical AI company. The G9L runs on the exact same Turing AI chip that powers their IRON general-purpose humanoid robot. CEO He Xiaopeng says they believe in the concept of the car as a robot. They built an in-house AI foundation that powers both machines.
The vehicle uses a driving system called Next Generation Pilot. This system is powered by the VLA 2.0 model. VLA stands for vision, language, and action. The model retains up to 30 seconds of contextual information from the road. It uses that data to predict how a traffic scenario will unfold up to six seconds into the future. XPeng is currently offering public test rides in Paris. They are one of only two automakers in the official Autonomous Lab alongside Tesla.
The hardware specifications are built for aggressive global expansion. The SUV delivers up to 591 kilometers of range. Its 800V architecture and 5C charging technology allow the battery to go from 10 to 80 percent in exactly 11.7 minutes. To avoid European import tariffs on Chinese vehicles, XPeng is manufacturing the G9L through a contract with Magna in Graz, Austria. The company plans to sell the vehicle in 64 global markets and is actively pursuing European regulatory certification for its autonomous software.
Key facts
- 74,990 — Starting price in Euros for the G9L SUV in France
- 11.7 — Minutes it takes to charge the battery from 10 to 80 percent
- 30 — Seconds of contextual information retained by the VLA 2.0 model
- 6 — Seconds into the future the AI predicts traffic scenarios
- 64 — Number of global markets where the G9L will be available
Why it matters
The line between automotive engineering and robotics is completely gone. XPeng built a single AI foundation for both cars and humanoid robots. If you are building AI hardware today, you no longer need separate compute architectures for different form factors. You build one brain and put it in a car, a robot, or a flying vehicle. This unified approach reduces development costs and speeds up deployment. The software learns from the physical world regardless of the chassis it sits inside.
This forces a massive shift in how companies scale artificial intelligence. By using the same Turing chip across multiple product lines, XPeng accelerates its data collection and model training. Every single mile driven by the G9L SUV improves the IRON humanoid robot. European legacy automakers now have to compete against a unified AI ecosystem, not just a cheap battery on wheels. Companies that still treat cars as mechanical products with bolted-on software will not survive this transition.
For builders
Unified AI architectures reduce hardware costs
XPeng uses the same Turing AI chip for cars and robots. Hardware startups can save millions by standardizing their compute layer across different physical products. You pay for one architecture and deploy it everywhere.
Context windows matter for physical AI
The VLA 2.0 model holds 30 seconds of real-world context to predict events six seconds ahead. Engineers building autonomous agents must prioritize temporal memory over simple frame-by-frame reactions. The AI must understand the flow of time to navigate physical spaces.
Local manufacturing bypasses trade barriers
XPeng is building cars in Austria with Magna to dodge European import tariffs. Hardware founders targeting global markets must secure local assembly partners early. If you rely entirely on cross-border shipping, you risk getting priced out by sudden regulatory changes.
My take
XPeng gets it. A car is just a robot you can sit inside. While legacy automakers are still treating software as an afterthought, XPeng is dropping humanoid robot brains into premium SUVs. If your hardware company is not building a unified AI foundation right now, you are already obsolete. The future belongs to physical AI.
Original reporting: Morningstar (PR Newswire). This is my rewrite and opinion.