🔥Stalecollected in 3m

Tesla FSD seeks EU regulatory approval amid sales surge

Tesla FSD seeks EU regulatory approval amid sales surge
PostLinkedIn
🔥Read original on 36氪

💡Tesla's FSD expansion into the EU could trigger a massive influx of real-world driving data for model training.

⚡ 30-Second TL;DR

What Changed

Tesla European registrations rose 46.5% YoY to 10,654 units.

Why It Matters

Successful FSD deployment in Europe would significantly expand Tesla's addressable market for autonomous driving software and data collection.

What To Do Next

Monitor the EU regulatory portal for updates on Tesla's FSD safety certification progress.

Who should care:Developers & AI Engineers

Key Points

  • Tesla European registrations rose 46.5% YoY to 10,654 units.
  • EU-specific sales saw a significant growth of over 67%.
  • Tesla is currently applying for FSD regulatory entry in the EU.

🧠 Deep Insight

Web-grounded analysis with 28 cited sources.

🔑 Enhanced Key Takeaways

  • Tesla FSD (Supervised) received its initial European type approval from the Dutch RDW on April 10, 2026, following 18 months of rigorous testing and over 1.6 million kilometers driven on EU roads.
  • Despite the Dutch approval, regulators in Sweden, Finland, Denmark, and Norway have voiced significant skepticism regarding FSD's safety, raising concerns about its tendency to speed, performance on icy roads, and potential for driver distraction.
  • The European iteration of Tesla's FSD software is distinct from its U.S. counterpart, with functionalities and software versions explicitly stated as 'not comparable one-to-one' due to necessary adaptations for Europe's unique infrastructure, traffic patterns (such as prevalent roundabouts), and multilingual signage.
  • Tesla's advocacy methods, including encouraging vehicle owners to exert pressure on regulatory bodies, have reportedly caused frustration among European officials who prioritize decisions based on technical merit rather than public campaigns.
  • For FSD to achieve EU-wide approval, a vote from the Technical Committee on Motor Vehicles is required, necessitating approval from 55% of member states representing 65% of the EU's population, with a decision potentially anticipated in July or October 2026.
📊 Competitor Analysis▸ Show
Feature/AspectTesla FSD (Supervised)Mercedes-Benz DRIVE PILOT (Level 3)BMW Personal Pilot L3 (Level 3)
Autonomy LevelSAE Level 2 (Driver Supervised)SAE Level 3 (Conditionally Automated)SAE Level 3 (Conditionally Automated)
Sensor SuiteVision-only (cameras)Cameras, radar, ultrasonic, LiDARCameras, ultrasonic, radar, 3D LiDAR, HD Maps
Operational DomainMost driving situations, requiring supervisionHighways, heavy traffic/congestion, up to 95 km/h in GermanyMotorways with separated carriageways, up to 60 km/h in Germany
Driver ResponsibilityDriver must continuously supervise and be ready to interveneSystem takes responsibility under specific conditions; driver can perform ancillary tasksSystem takes responsibility under specific conditions; driver can perform ancillary tasks
EU Approval StatusFirst European type approval in Netherlands (April 2026); EU-wide pendingApproved in Germany (December 2021, updated December 2024)Available in Germany (March 2024)
Pricing (Germany)Subscription/Purchase cost for FSD (specific EU pricing not detailed in sources)Starts at €5,950 (incl. VAT) for S-Class/EQSAround €6,000 for 7 Series

🛠️ Technical Deep Dive

  • Vision-Only Philosophy: Tesla FSD operates primarily on a camera-only perception system, utilizing neural networks to interpret raw video data, a distinct approach compared to competitors that often integrate LiDAR and radar.
  • Neural Network Stack: The FSD system functions as a pipeline of interconnected neural networks, processing multi-camera inputs, extracting features through video-based neural networks that analyze temporal continuity, and converting perception into a vector space representation where elements like lanes become splines and vehicles are dynamic objects.
  • Occupancy Networks: A significant advancement in FSD is the implementation of occupancy networks, which predict occupied versus free 3D space. This enhances the system's ability to handle occlusion and improve depth estimation beyond basic object detection.
  • End-to-End Learning: With FSD v12, Tesla has transitioned towards a fully end-to-end deep learning system. In this architecture, raw camera inputs directly generate control commands for steering, braking, and acceleration, effectively replacing extensive hard-coded C++ logic.
  • Training Infrastructure: The development and refinement of FSD necessitate immense computational power, historically relying on NVIDIA GPUs, and increasingly leveraging Tesla's custom-built supercomputer, Dojo, which is optimized for video training workloads.
  • Hardware Evolution: Tesla's Autopilot hardware has undergone several iterations, from Hardware 1 (Mobileye EyeQ3) to Hardware 2/2.5 (Nvidia Drive PX2), then to Hardware 3 (custom "FSD Chip"), and most recently to Hardware 4, which began shipping in January 2023. Hardware 4 boasts 3 to 8 times more computational capability than HW3 and features higher-resolution cameras. FSD v13 is the first software version designed to utilize the native resolution of all HW4 cameras.
  • Data-Driven Improvement: The system continuously improves its performance and ability to handle complex edge cases by learning from billions of miles of real-world driving data, including human driver reactions captured in "shadow mode."

🔮 Future ImplicationsAI analysis grounded in cited sources

Broader EU FSD approval will accelerate the adoption of advanced ADAS across the continent.
A harmonized regulatory framework and successful deployment by a major player like Tesla will likely push other manufacturers to fast-track their own advanced systems to remain competitive.
Tesla's vision-only approach will face continued scrutiny and potential adaptation requirements in Europe.
European regulators have raised specific concerns (e.g., icy roads) that might necessitate sensor redundancy or further validation beyond a purely camera-based system, especially given competitors' use of LiDAR.
The EU will likely establish more unified and stringent regulations for Level 3 and higher autonomous driving systems by 2026-2027.
The current regulatory landscape is fragmented, and the EU has expressed intentions to create a more harmonized framework and regulatory sandboxes to facilitate deployment and address safety concerns.

Timeline

2014-10
Tesla introduces Autopilot with Hardware 1 (Mobileye EyeQ3, camera, radar, ultrasonic sensors).
2016
Tesla ends partnership with Mobileye, begins shipping cars with Hardware 2 (Nvidia Drive PX2, more cameras).
2019-04
Tesla shifts to Hardware 3 ("FSD Chip") and begins rolling it out to all vehicles.
2020-06
UN Regulation No. 157 (ALKS) adopted, enabling Level 3 driving features up to 60 km/h.
2021
Tesla stops installing radar sensors in new vehicles, moving towards a vision-only approach.
2022-06
UN Regulation No. 157 (ALKS) speed limit for Level 3 systems extended to 130 km/h.
2023-01
Tesla begins shipping cars with Hardware 4.
2023-08
Elon Musk live-streams the release of Tesla's FSD V12.0, an end-to-end AI autonomous driving system.
2025-Late
Tesla begins offering FSD ride-alongs in key European cities (Germany, France, Italy).
2026-04-10
Dutch vehicle authority RDW grants Tesla the first European type approval for FSD (Supervised).
📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪