Tesla Shifts Focus from Compact Cars to Robotaxis and Humanoid Robots, Declaring Transition to Physical AI Company

Park Hyung-geun, Department Head, stated that Tesla is preparing to transition from automotive-centric growth to a physical AI company. The core of this shift is Tesla's internal strategic reorientation rather than external factors like SpaceX. He explained that while the company previously grew rapidly as an automotive company, it is now entering a preparatory phase to leap forward as a physical AI company.

Tesla announced that it is canceling its plans for compact car production at Giga Mexico and will instead focus on the robotaxi business. Production lines for Model S and Model X are also being converted into robot production facilities. Park Hyung-geun, Department Head, analyzed that the change in perception regarding the stagnation of the automotive market is behind this decision, stating, "From Tesla's perspective, the traditional automotive market seems to be stagnating."

Tesla's second-quarter operating margin declined to the 16% range, attributed to intensified competition with Chinese manufacturers and margin pressure from its energy business. Park Hyung-geun, Department Head, particularly highlighted the "transition to negative free cash flow." This phenomenon can be interpreted as the official start of future investments, with Park stating, "It seems plausible that serious investment in the future has begun."

This shift in cash flow suggests that Tesla is moving away from a business model reliant on car sales profits and investing in large-scale future ventures. While the company's financial health may deteriorate in the short term, it is seen as a strategic choice to secure long-term growth drivers.

The decline in profitability faced by Tesla is not merely a temporary phenomenon but rather an inevitable transitional phase occurring during the shift to a new business model. The fierce competitive environment in the Chinese EV market, in particular, is significantly impacting Tesla's automotive business profits.

Tesla is planning 'Tera,' the largest facility ever for AI accelerator chip production. A massive rendering, 5km wide and 1km long, has been unveiled, aiming for 50 times the production capacity compared to Elon Musk's 20GW target. Park Hyung-geun, Department Head, emphasized that this is an investment in preparation for the robot era, stating, "If you think about why Elon Musk is building such an excessively large factory, he has always maintained that the current automotive market is a very minor market, and the era of robots will open up in the future."

Tera is expected to be a core infrastructure of Tesla's future AI strategy, going beyond mere AI chip production. Through this facility, Tesla aims to internalize AI semiconductor production capabilities and secure the vast computational resources needed for autonomous driving and robot development.

This large-scale investment demonstrates Tesla's commitment to evolving from a mere automobile manufacturer into a comprehensive technology company based on AI technology. Tesla's strategy is to reduce external reliance and accelerate technology development by securing its own AI chip production capabilities.

Tesla concluded that it would be difficult to dominate the low-cost market due to the aggressive pricing strategies of Chinese EV manufacturers. Consequently, it shifted its business focus from the compact Model 2 strategy to autonomous driving and the robot industry. Park Hyung-geun, Department Head, explained, "China caught up too quickly in the EV market, and we realized it would be difficult to dominate the lower-tier market. I think they quickly came to that conclusion and moved on to the next stage."

This strategic shift, even abandoning the premium lineup, demonstrates how desperately Tesla is seeking new growth drivers. The profitability of the existing entry-level model market has significantly deteriorated due to strong competition from Chinese companies.

Tesla has now formulated a strategy to focus on creating new value through autonomous driving technology and robots, rather than increasing car sales. This is interpreted as an intention to find Tesla's long-term growth drivers in the physical AI sector.

Tesla's new business areas, including robot AI, Cybercabs, and Terawatt infrastructure, are expected to incur astronomical costs. Park Hyung-geun, Department Head, stated, "It demands a lot of money. As I mentioned earlier, there's investment in robot AI, Cybercabs need to be produced, and I believe Space X will share the burden for Terawatt." To cover these massive investments, maintaining stable cash flow through high margins from the automotive business is crucial.

However, a significant challenge for Tesla remains the potential delay in revenue generation from autonomous driving businesses like robotaxis due to regulatory barriers. If regulatory hurdles persist, the payback period for investments could lengthen, increasing financial burdens.

Park Hyung-geun, Department Head, analyzed that Tesla must overcome four hurdles to succeed as a physical AI company. The first is the widespread adoption of devices that can implement physical AI, such as robots and robotaxis. "First, devices are needed, that is, physical AI devices like robots and robotaxis," he explained.

The second is to mass-produce these devices to accumulate an enormous amount of autonomous driving data. Park emphasized, "Second, having only devices is not enough. It seems necessary to mass-produce them to accumulate a great deal of data." This data is essential for training AI models.

The third hurdle is building infrastructure for AI model training and deployment, and the final fourth is expanding beyond the power and space limitations of data centers, even into space. He explained that Tesla's physical AI vision can only become a reality by creating a virtuous cycle where these four elements organically interact.

Starting in 2024, Tesla implemented 'end-to-end' technology in autonomous driving, where AI is solely responsible for decision-making. Park Hyung-geun, Department Head, explained the characteristics of this technology, stating, "From 2024, Tesla applied end-to-end technology, where AI is entirely responsible for decision-making. As a result, it becomes very difficult for humans to examine the judgment process." This makes it challenging for humans to understand the AI's decision-making process, and combined with Tesla's 'vision-only' sensor approach, it faces difficulties in convincing regulatory bodies compared to competitors like Waymo.

Tesla is currently developing its third-generation Optimus robot and acknowledges the high difficulty in building its supply chain. Park Hyung-geun, Department Head, explained, "For humanoid robots, Tesla is particularly obsessed with hands. The reason for this obsession with hands is that everyone anticipates that humanoid robots will not be limited to factory applications." This demonstrates an ambition to achieve the precision of human-like hands with 22 degrees of freedom, indicating a goal to target the home market beyond mere factory deployment. Factory application is expected to serve only as a stepping stone for accumulating large-scale data.

The primary reason for the delay in humanoid robot release is the need to accumulate far more complex and extensive data than existing autonomous driving data. Park Hyung-geun, Department Head, pointed out, "For robots, it seems they need to accumulate much more data. To do that, there's clearly pressure to create a truly human-like robot from the start. I think this is why the timeline keeps getting delayed."

The precision required for humanoid robots and the difficulty of acquiring the necessary data are major factors contributing to development delays. Additionally, safety regulations and consumer acceptance issues are cited as reasons for a cautious approach to robot release.

The potential of the humanoid robot market can be seen through the case of China's Unitree. Park Hyung-geun, Department Head, mentioned, "Do you know the company Unitree in China? They are going public. They have a humanoid robot called Jiyuan, which is pretty much like a toy robot. But Unitree has already delivered about 5,000 of these robots last year, and they say they will deliver 10,000 this year, for 30 million won, which you can even buy at our supermarkets." He noted that Unitree's robot, despite its unclear practical applications, already has significant demand.

Tesla is pursuing a 'Tesla One Chip' strategy for the integration of robots, robotaxis, and data centers, with the ultimate goal of self-producing the chips they will consume. Park Hyung-geun, Department Head, explained, "Ultimately, this Tera factory I mentioned aims for Tesla to self-produce the chips they will consume."

To achieve this, Tesla aims to complete its self-production chip factory, 'Tera,' around 2036. This is part of a long-term strategy to reduce reliance on AI semiconductors and maximize cost efficiency.

Securing internal chip production capabilities reflects Tesla's strong determination to build an independent ecosystem in the physical AI sector and control the pace of technological innovation independently.

The primary economic reasons for Tesla's insistence on developing its own chips are cost reduction and independence from Nvidia. Park Hyung-geun, Department Head, explained, "Ultimately, only through self-production can costs be reduced, and the automotive business can be profitable. Tesla has already entered such a virtuous cycle. If they rely on external production, ultimately, Nvidia ends up earning all the profits."

External contract manufacturing leads to increased reliance on specific companies like Nvidia, which in turn results in high costs. Tesla aims to save approximately 10 million won per vehicle in hardware costs through self-production.

This strategy demonstrates Tesla's commitment to improving its long-term profit structure and strengthening control over AI technology development. Self-producing chips is considered an essential element for building Tesla's physical AI ecosystem.

Tesla recognizes the limitations of existing terrestrial data center construction and is exploring unconventional solutions to overcome them. Park Hyung-geun, Department Head, reported, "Elon Musk, at the Tera event, showed visions of going into space. So, he is saying, 'We will build data centers in space.'"

This is Elon Musk's strategic move to break through the space constraints required for massive data processing and storage, presenting a vision for building data centers in outer space.

The United States currently lacks large-scale manufacturing facilities, excluding Tesla, which enhances the value of cooperation with Korea. Park Hyung-geun, Department Head, mentioned Korea's strengths, stating, "The U.S. virtually has no manufacturing facilities except for Tesla. Korea produces semiconductors, manufactures cars well, and has a certain level of robot-related component ecosystem."

Korea possesses diverse manufacturing capabilities, including semiconductor production, automobile manufacturing, and a robot-related component ecosystem, allowing it to build a mutually complementary cooperative relationship with U.S. AI big tech companies. This suggests that Korea's role in producing the physical devices required by AI companies like Tesla could grow.

Particularly in the AI era, not only software but also hardware manufacturing capabilities to implement it are becoming crucial, further highlighting Korea's status. Cooperation could involve the U.S. focusing on advanced AI technology development, while Korea provides the manufacturing base to bring that technology to reality.

Such cooperation could also offer new opportunities for Korean industries, opening up the possibility for domestic manufacturing companies to grow into key partners in the global AI ecosystem.

A two-track strategy of 'internalization and cooperation' is being advocated for Korean companies to prepare for the AI era. Park Hyung-geun, Department Head, emphasized, "What our companies are discussing now is a two-track approach: simultaneous internalization and cooperation." While external collaborations, such as Nvidia partnerships, can help overcome initial learning data gaps, they come with high cost burdens.

Therefore, in the long term, improving the profit structure through securing proprietary technology is necessary. That is, a hybrid approach is crucial: simultaneously developing and internalizing core AI technologies while fostering strategic collaborations with global AI companies to create synergy.

This two-track strategy is expected to help Korean companies secure sustainable competitiveness in the AI market and reduce dependence on specific technology providers.