-+ 0.00%
-+ 0.00%
-+ 0.00%

FSD expands European footprint to six countries! Slovenia approves Tesla (TSLA.US) physical AI ambition to accelerate implementation of autonomous driving system

Zhitongcaijing·09/09/2026 08:33:20
Listen to the news

The Zhitong Finance App learned that US AI, robotics, autonomous driving and automobile manufacturing giant Tesla (TSLA.US) said that the Slovenian government has approved the use of its FSD fully autonomous driving assistance system on the country's roads, making it the sixth European country to approve the system before the EU may hold a nationwide approval vote.

“The FSD supervised version has now been approved in Slovenia and will be rolled out soon.” Tesla posted on social networking platform X on Monday that the company's CEO Elon Musk retweeted the news. The Slovenian Minister of Energy and Infrastructure Jernej Vrtovec also relayed the announcement with the caption “Continue to develop Slovenia”.

The Dutch government's European Traffic and Road Supervisory Authority previously temporarily approved the software for use on Dutch roads in early April, making the Netherlands the first EU country to allow the use of FSD. The system can fully control Tesla cars, but Tesla still requires the driver to maintain constant attention in the guidelines.

Tesla publicly released a self-produced European and North American market test data set last week. At the time, Musk said that an EU-wide approval vote could be held as early as October 6.

Tesla, under Musk's leadership, is simultaneously advancing FSD technology upgrades, European FSD and RoboTaxi entry, and automotive software monetization, and robot mass production to accelerate the implementation of the “physical AI supermap” drawn by Musk for Tesla's growth prospects. The company's second-quarter update shows that FSD v14 Lite adapts the driving capability of the AI4 platform v14 series to earlier AI3 hardware through model distillation to improve scenarios such as paralleling, pedestrian interaction, and destination parking, and has begun to be promoted to users in the US and South Korea.

According to Nvidia CEO Hwang In-hoon's opinion, “physical AI” (physical AI) emphasizes that autonomous robotics/car operating systems can sense, reason, and complete a complete set of actions in the real world, and an era where “physical AI” assists the evolution of human civilization is coming soon. “Physical AI” emphasizes enabling robotics/autonomous systems to sense, reason, and complete a complete set of actions in the real world, and these three types of capabilities are the key AI tool chain to advance the AI language model from “only dialogue” to “can work in the physical world.”

The Slovenian Traffic Safety Authority confirmed via email that the traffic regulator has approved the system, adding that while Slovenia seeks to remain open to innovation and technological developments, new technologies must be introduced in a safe manner. The agency said, “The temporary type approval for the FSD (Manually Supervised) system is a concrete example of how new technology can be gradually integrated into the transportation system under proper supervision and safety supervision.”

In addition to the Netherlands and Slovenia, four other European countries have approved Tesla's driver assistance systems, while Finland and Greece said they are considering approval.

image.png

France questioned the possibility of approving the autonomous driving system in July due to safety concerns. France said last week that it has begun testing two electric cars equipped with FSD after what the country's Minister of Transport called “constructive exchanges” with Musk.

European FSD entry accelerates, Cybercab starts the road phase

By the end of the second quarter, FSD paid users reached 1.48 million, an increase of 56% over the previous year, including one-time purchases and monthly subscription users; more than 55% of new car deliveries in North America included FSD subscriptions. Slovenia became the sixth European country to approve the system, further expanding the potential payment market, but what was approved this time is still an automated driver-assistance system that requires continuous driver supervision.

Robotaxi, on the other hand, has begun operating by modifying existing models and putting dedicated driverless vehicles into service. On September 3, Tesla's exclusive Robotaxi without a steering wheel and pedals — the two-seater Cybercab — began providing limited commercial ride-hailing services in parts of Austin. According to media reports, Tesla registered 420 autonomous vehicles in Texas at the time, 45 of which were Cybercab, but the number of registrations did not mean that all vehicles had already been put into operation for a fee. For investors, this step is related to whether Tesla can form a business that continuously charges according to the itinerary, but the US National Highway Traffic Safety Administration launched an investigation into Cybercab's self-certification of safety standards on September 4, and commercial expansion across the US still needs to cross the corresponding regulatory threshold.

Musk's “Physical AI (Physical AI)” macromap links autonomous driving, general-purpose robots, and autonomous computing power capabilities. At the G20 event on September 1, he predicted that the number of humanoid robots in the world may exceed 1 billion in the next ten years, and he also anticipated that robots will participate in manufacturing robots to further accelerate expansion.

Tesla is speeding up the construction of the corresponding foundation — the recently built Cortex 1 and Cortex 2 AI training infrastructure capacity exceeds 90 megawatts and 115 megawatts, respectively, of which the latter also supports the development and update of the fully autonomous capabilities of Cybercab, autonomous vehicles, and AI humanoid robots; Tesla's Fremont original Model S and Model X gigafactory production lines are being converted to Optimus Optimus humanoid robot production lines, and the first mass-produced humanoid robots are planned to be used for large-scale training data collection and functions development.