The following is a summary of Tesla's third-quarter 2024 earnings conference call. For an interpretation of the financial report, please refer to " "The Return of the Pie King" Tesla Finally Returns as King! 》
1. Review of Core Financial Information:
2. Detailed Content of the Earnings Conference Call:
2.1. Key Points from Executive Statements:
Business Progress
1）Automotive Sales
(1) Guidance: Despite certain external factors (large-scale wars in some parts of the world), vehicle growth next year could be between 20%-30%.
(2) Models:
- There is hope to offer more affordable models by 2025H1, and the sales of these models are highly anticipated.
- Cybercab is expected to be mass-produced in 2026, with an annual production capacity of 2 million units, produced in multiple super factories together, with a final volume reaching 4 million.
- Progress has been made in the development of the Roadster model, and the design work is nearing completion.
(3) Pricing: Committed to improving the affordability of each vehicle.
(4) Profitability: There may be a decrease in profits. Maintaining the current profit level in 2024 Q4 may be challenging, and the focus will continue to be on cost control per vehicle, striving to reduce costs without compromising the user experience.
2）Battery and Energy Storage:
2.1 Capacity:
(1) Current Situation:
- The 4680 cell is the most cost-competitive internally at Tesla. In addition to internal manufacturing, Tesla will also collaborate with external partners to ensure the capacity of the 4680 cell in energy storage.
- There are many external suppliers for energy storage, but there is also good progress internally.
- The production capacity of MegaPack is gradually increasing, with the current utilization rate reaching 40%.
- The second megapack factory in Shanghai is expected to start production in 2025Q1, with an annual capacity of 20GWh.
(2) Future Outlook: In the near future, our annual fixed energy storage capacity is expected to reach 100 gigawatt-hours, and eventually may reach several terawatt-hours per year.
2.2 Order Fulfillment: Order fulfillment has fluctuated slightly in each quarter, with a slight fluctuation in 2024Q3.
2.3 Installed Capacity: It is expected that the installed capacity of the energy storage business will increase quarter-on-quarter by 2024Q4, with a potential year-on-year doubling of growth.
3）FSD:
(1) Intervention Mileage: Compared to version V12.5, we expect the mileage between two interventions for V13 to increase by 5-6 times. In the future, we hope to increase the mileage between two interventions by several orders of magnitude. (2) Hardware capabilities: Improving hardware capabilities can allow for more end-to-end improvements in FSD at the software level. It involves analyzing how to expand neural networks and how to better train based on more data.
(3) FSD utilization: FSD utilization has significantly increased, and Tesla will introduce a 30-day trial to encourage people to try FSD, further boosting FSD utilization.
(4) Outlook for next year: We expect to see further improvements in FSD starting in Q1 next year. By Q2 (or Q3) next year, the safety of FSD driving may surpass human driving safety. It is anticipated that all existing vehicles will also be able to achieve the new version of FSD capabilities next year.
4）Robotaxi:
(1) Automatic summoning: In the future deployment, a very simple operation can bring the vehicle from the parking lot to you.
(2) Robotaxi software: In the application, you can see the current driving route and estimated time of arrival. The entire network is currently being built and is still in the early stages.
(3) Launch timing: It is expected to provide extensive taxi services to the public in Texas and California in 2025.
The approval process in California is a bit lengthy, and regulatory approval will take some time, but it should be approved within the next year.
The approval process in Texas is much faster than expected, and it should be able to start providing such calling services in Texas very soon.
5）Computing power:
Currently, Tesla has no constraints on computing power and can efficiently train to make the models more powerful.
It has already started using a GPU cluster at the Austin Gigafactory in advance and is expected to further deploy 50,000 GPU computing power in Texas by the end of this month.
6）Capital expenditure:
Capital expenditure for this quarter is approximately $3.5 billion, showing a slight increase from the previous quarter. It is expected that the full-year capital expenditure will exceed $11 billion this year.
- FSD revenue:
Due to FSD starting to be carried on the Cybertruck and FSD features being updated in North America, it contributed approximately $326 million in revenue this quarter.
[Statement]
I. Business Overview
Q3 total deliveries set a record, with many electric vehicle companies worldwide still not profitable. Our production and deliveries have been very good so far, with decent profits in the vehicle segment. However, the industry is facing challenges, and the overall environment is not favorable. Nevertheless, we achieved a new record in 2024 Q3, producing Tesla's 7 millionth vehicle yesterday.
In the energy storage business, we also achieved good performance. The energy storage business is growing rapidly like a wildfire, with strong demand for Megapack and Powerwall in terms of electricity needs.
In the robotics business, we showcased robots at the We, Robot event on October 10th. The humanoid robots walk very naturally and provide us with many convenient services, such as serving drinks and coffee. We also displayed dozens of Robotaxis, which we call CyberCabs. We also introduced Robotvans with large loading capacities. CyberCabs and Robotvans do not have steering wheels or pedals, marking a significant industry transformation In the driving process, no one can intervene manually. All vehicles at the event performed perfectly, and the humanoid robots were also very lovable.
II. Key points of each business line
Vehicles:
In terms of vehicle business, we hope to provide more affordable models by 2025H1, and we can look forward to the sales volume of models. We believe that the growth of vehicles next year can reach 20%-30%. Despite the influence of certain external factors (such as large-scale wars in some parts of the world), we still have confidence. I believe that with the further introduction of autonomous driving, a growth of 20%-30% can be achieved, and there will be even greater growth by 2026. Cybercab will be mass-produced in 2026, and we believe that 2 million CyberCabs can be produced annually, jointly produced in multiple super factories, with a final volume reaching 4 million.
Battery and Energy Storage:
Regarding the 4680 battery cells, our current operating rate is also very good and competitive. Production costs and other costs need to be considered, and in the United States, we need to consider policy incentives and taxation. Ultimately, the cost of 4680 can be achieved at a relatively low level. We have not achieved the final goal yet, but we will continue to work hard, and we are very close to the target.
The implementation of the 4680 battery cells is also remarkable, being the most cost-competitive battery cells within Tesla. We not only have many manufacturing activities internally but also cooperate with external partners to ensure a very good guarantee of production capacity for the 4680 battery cells in energy storage.
In energy storage, we need more and stronger battery cells. We currently have many external suppliers for energy storage, but we are also making good progress internally. We hope to combine internal and external production capacity.
Autonomous Driving:
Autonomous driving technology is advancing rapidly, and we hope to further improve vehicle utilization through autonomous driving. There has been a significant improvement in the mileage between two interventions. This is the case in the FSD V12.5 version, and we hope that FSD can better integrate with vehicles such as CyberTruck, CyberCab, Robotvan, etc. We hope that the relevant stack can be further improved. We hope to bring a better driving experience in city and highway driving.
FSD V12.5 is already very strong, and the FSD V13 version will be released soon. We expect the mileage between two interventions in V13 to increase by 5-6 times compared to V12.5. In the future, we hope to increase the mileage between two interventions by several orders of magnitude.
This year, FSD has made significant progress, and we expect the same for next year. We hope to achieve better performance in FSD, with a major indicator being the mileage between two interventions. By Q2 or Q3 next year, the safety of FSD driving may surpass human driving safety. Since the beginning of this year, the V12.5 version has made significant improvements, possibly by a hundredfold. The upcoming V13 version will bring even better improvements, and we expect to see further improvements in FSD starting from Q1 next year. Our technology has achieved end-to-end optimization, with further improvements in frame rate. This means that hardware capabilities have been further enhanced, which allows FSD to have more end-to-end improvements at the software level. Behind this, we need to analyze how to expand neural networks and how to train better based on more data It is expected that by Q2 next year, FSD will surpass the safety level of human driving.
For models like Model 3 and Model Y, people feel that these models have been around for a long time, and there is no expectation for them to drive on their own. However, the new models we are introducing now have significant differences in appearance and capabilities compared to previous models. The CyberTruck looks very different, and these new cars look great. If a car looks ordinary in appearance, people may not have the desire to understand what it can do in terms of autonomous driving. We hope to show everyone what a fully autonomous driving car should be like.
Software has been continuously improving, and we will launch a 30-day trial to encourage people to try FSD, which will significantly increase the adoption rate of FSD. Over the past long period, we have seen a significant increase in the adoption rate of FSD, and we hope this trend can continue.
After the We, Robot conference on October 10th, we saw more people trying FSD. They don't actually need to wait for Robotaxi or Robotvan to be mass-produced. Our existing vehicles can already provide a very good FSD experience for people. We expect that all existing vehicles will be able to achieve the new version of FSD capabilities next year.
We hope to further explore the experiences that vehicles bring to people in different scenarios. Here, we want to emphasize automatic summoning. This means that with very simple operations, the vehicle can come to you from the parking lot. In the future, we will further deploy such functions. We have also provided this vehicle capability to Tesla employees in the Bay Area. We have done a lot of work in application development, such as allowing vehicles to take people anywhere in the Bay Area. Currently, in the Bay Area, such autonomous driving is still equipped with safety personnel. In the future, we hope to truly not need safety personnel.
Our software is progressing rapidly. Some screenshots are also shown in the presentation. We hope to further test this year to see how our autonomous driving can be better deployed on the road.
Throughout the deployment process, we hope that our new vehicles, including CyberCab, can provide the best experience for people and create greater value. Whether it's navigation or media in the car, we aim to provide the best experience by offering all functions to all users through software and systems.
On the other hand, using ride-hailing apps can quickly summon vehicles. In fact, the process of navigating to the destination is also automatically determined by the vehicle. For many years, we have wanted to provide users with this Robotaxi function, where users can see the current driving route and estimated time of arrival in the application. We are currently building the entire network, which is still in the early stages. It is expected that next year, we will provide extensive taxi services to the public in Texas and California. The approval process in California is a bit lengthy, and regulatory approval takes some time, but it should be approved within the next year. The approval process in Texas is much faster than expected, and we should be able to start providing such ride-hailing services in Texas very soon No matter what, we are very excited to promote the Robotaxi calling service, which represents a profound change in the transportation industry. Tesla vehicles are not just means of transportation, but also a driving force for lifestyle changes. By then, we will become a better company.
Vehicle Safety:
We have also released the 2024Q3 Vehicle Safety Report, which shows that every 7 months we will have new initiatives and pilot routes. Data shows that in the United States, an accident occurs every 700,000 miles. However, our vehicle safety report shows that Tesla's vehicle safety level far exceeds the overall average vehicle safety level in the United States.
Computing Power:
We will continue to expand AI training capabilities to meet the needs of FSD and humanoid robots. Currently, we have no constraints in terms of computing power, which is crucial. FSD has become so outstanding that computing power is indispensable, but we also need to take some time to identify any shortcomings in this area. If there are some minor errors on a scale of ten thousand miles, we need to figure out where the problem lies, how much more time we need to collect a large amount of data, and do better training to further strengthen our software versions.
Humanoid Robots:
Regarding humanoid robots, a presentation was made at the we, robot conference on October 10th. In the future, humanoid robots should be able to achieve 22 degrees of freedom, which is twice as much as before. Such humanoid robots are very similar to humans, and we see that the computing power on which humanoid robots rely will also increase significantly. I am confident to say that in this field, Tesla is already at the forefront of the industry. We are the only company capable of deploying such technology on a large scale.
So, what are other companies lacking in this regard? I believe they lack capabilities in AI. In comparison, we have a huge advantage in achieving mass production. Specifically in the field of humanoid robots, we firmly believe that humanoid robots will eventually become the most valuable part of our business. I emphasize once again that humanoid robots may become the most valuable product in my career.
Energy Business:
As for the performance of the energy storage business, the production capacity of MegaPack is gradually increasing, and the current utilization rate has reached 40%. In addition, our second megapack factory in Shanghai will start production in the first quarter of next year, with an annual capacity of 20GWh. Speaking of the increase in energy storage scale, soon, our annual fixed energy storage capacity is expected to reach 100 gigawatt-hours, and eventually may even reach several terawatt-hours per year. To achieve a more sustainable energy development, we must mass-produce and deploy energy storage products, and this scale growth is undoubtedly necessary. Overall, I believe this field holds tremendous growth potential. We have also mentioned the use of sustainable energy generation to bring electricity into more applications, enabling electrified transportation and other applications, which is crucial for future development.
Materials:
In terms of materials, our development is not significantly constrained. Although it involves some rare materials, we have sufficient reserves, so there are no significant constraints in material supply. Looking ahead to the future of human civilization, to achieve sustainable development goals, especially in the energy field, we need to vigorously promote the widespread application of these technologies Human civilization needs to further progress, relying on technological innovations like ours in the energy storage field to help drive this process.
From a global perspective, we need to make more extensive use of solar energy. Currently, the utilization rate of solar energy is still relatively low, but in order to achieve a sustainable energy future, we must accelerate the deployment of solar energy. Overall, innovations in fields such as energy, transportation, and robotics can provide tremendous momentum for the development of human civilization and help achieve complete sustainability.
Despite many unfavorable factors in the current macro environment, I want to express special thanks to the team for their outstanding execution throughout this process.
III. Financial Performance
In the third quarter, our performance was very strong, which also proves that our strategic direction over the years has been correct. This quarter, our cash flow performance was very good, with operating cash flow reaching $6.3 billion.
Vehicle Business:
In terms of the vehicle business, we achieved both sequential and year-over-year growth, with sales increasing but also experiencing some fluctuations. Regarding vehicle financing, we note that financing incentives provided by third-party banks and financial institutions may have a favorable impact on the cost of purchasing vehicles. We also released new versions of FSD and features like Summon, which play a significant role in further advancing the vehicle business.
Carbon Credits and Regulations:
In this quarter, carbon emission credit income performed well. As of now, our revenue in this area is approximately $2 billion. We hope to continue to closely follow regulatory trends in the entire industry, especially in the Chinese market. Our business performance has been outstanding, with the development speed in the Chinese market surpassing that of the United States and Europe. Through the performance of this quarter, we have learned a lot from the acceptance of our products by users in different regions and found that our products are widely welcomed.
Pricing and Profitability:
To make vehicles more affordable, we will continue to advance this strategy. Currently, our top priority is to increase production while avoiding inventory buildup. To execute this strategy, we need to provide users with more attractive vehicle financing options. Compared to other OEMs, we believe Tesla's vehicles have significant advantages in terms of safety, performance, and software. We collaborate with different stakeholders to bring more value to users.
We have also introduced many new features, with the profit margin in this quarter increasing sequentially and production and delivery data performing well. Through localized delivery, we have effectively reduced shipping costs and tariffs. However, maintaining the current profit level in the fourth quarter of 2024 may face challenges based on the current economic environment. We will continue to focus on cost control for each vehicle, striving to reduce costs without compromising the user experience. We may see a decline in profits as we strive to improve the affordability of each vehicle. The overall trend shows a downward trend in profits, with people holding onto vehicles for longer periods and the replacement cycle being extended, especially in the United States.
As for vehicle profits, we have mentioned that vehicle profits may fluctuate during certain periods, but we will do everything possible to control costs. Although some factors are beyond our control, we will optimize cost control as much as possible within our controllable range In fact, in 2024Q3, we achieved the lowest per-vehicle cost in Tesla's history. We hope to continue to focus on the dynamic changes in costs.
Energy Storage Business
At the same time, our energy storage business is also rapidly developing. According to the actual needs of customers, we will provide them with the most suitable solutions. Order fulfillment fluctuates in each quarter, with a slight fluctuation in 2024Q3. We expect that by the fourth quarter of 2024, the installed capacity of the energy storage business will increase compared to the previous quarter, and it is expected to double year-on-year. The profit margin of the energy storage business has reached a new high, exceeding 30%. We have a strong product portfolio in the energy storage field and install energy storage equipment for more customers through extensive deployment.
Looking ahead to 2025, we will continue to meet the market demand for the energy storage business. The pipeline orders and backlog for this sector are gradually increasing, and we are fully committed to meeting the demand.
Services and Others:
Our services, including collision repairs, spare parts, and other related products, have performed well. With the increase in the size of the fleet, this part of the revenue will continue to grow.
Expenses:
Our operating expenses have shown a year-on-year and quarter-on-quarter decrease. This is due to cost-saving measures taken in 2024Q2, combined with other cost changes, resulting in satisfactory results.
Capital Expenditure
We have started using the GPU cluster at the Gigafactory in Austin ahead of schedule and expect to further deploy 50,000 GPUs in Texas by the end of this month. We are very cautious in our spending. Regardless of the investment, we will fully consider how to better utilize existing infrastructure.
Capital expenditures for this quarter are approximately $3.5 billion, showing a slight increase compared to the previous quarter. It is expected that the full-year capital expenditure will exceed $11 billion. However, we are very focused on delivery, and all work is steadily progressing. This year, we have made great progress, despite the macro environment being full of headwinds, our performance remains outstanding.
2.2, Q&A Analyst Q&A
Q: Are there plans to launch more affordable models this year?
A: We have been very clear many years ago that we will introduce new models. But now we need to focus more on the pace, as the future direction of vehicle development is fully autonomous driving. Future autonomous driving vehicles will bring disruptive changes, similar to the transition from riding horses or horse-drawn carriages to cars in history. I believe that all vehicles in the future should be autonomous and driven by electricity, and this trend is obvious.
Many car companies have not yet fully realized the importance of this trend, they have not internalized the idea that all vehicles in the future will be based on electricity and have autonomous driving capabilities. Not long ago, our Tesla production exceeded 7 million vehicles. These vehicles all have autonomous driving capabilities. Currently, we produce about 35,000 autonomous driving vehicles per week, and vehicle production and scale continue to grow rapidly. As for when the $25,000 model will be launched, this is not the most important issue at the moment. The key lies in the development trend of the entire automotive industry. For designs like Cybertruck that look unusual or even bizarre, we believe it is a symbol of futurism This car features revolutionary design elements, showcasing our creativity and future vision in design work. By introducing such a model, we aim to provide users with a better experience.
We aim to offer a variety of models to provide users with the best autonomous driving experience. Technological advancement is crucial for industry development, and our other major advantage lies in manufacturing capability. In any case, we will move forward according to the established roadmap. As for the launch of the $25,000 model, it is not particularly meaningful to focus solely on its timing. We should focus more on the overall travel experience and consider how to provide a better way of travel through vehicles. While we emphasize repeatedly that we will strive to reduce the cost per vehicle, the specific timing of launching a model is not important. The most important thing is to reduce the cost per mile of travel.
We have several models and multiple projects in progress, hoping to enable more users to enjoy optimized autonomous driving capabilities through these models. As long as new demands arise, we will make every effort to meet them. Meeting user needs is always our top priority.
Q: How to solve the issue of long waiting time at Tesla service centers?
A: Our goal is to solve the problem at its root. Although there may be some issues within the factory, these issues will not affect our users. And after vehicle delivery, if any issues arise, we are confident in providing the best service. However, I must emphasize one point: the best service is no service at all. If the factory can solve all problems and ensure that there are no issues to be resolved after vehicle delivery, the importance of service centers will be greatly reduced. This also means that in the event of issues during vehicle use, the waiting time for users at service centers will be very short.
Our vision is to solve problems at the source and strive to minimize issues during vehicle use. In Q3 and Q4 of this year, we will add 70 service centers, with the scale of each individual service center in North America significantly expanded. Our service network has doubled in size compared to last year. I believe we need more specialized service personnel to provide better service to users. In addition, we hope to reduce subsequent maintenance costs through more early initiatives.
In fact, a car consists of roughly ten thousand components. Assuming the cost of each component is $4, the total cost of the entire car will be as high as $40,000. If we expect to reduce the cost to $35,000, we must reduce the cost of each component by $0.5, or 50 cents, among the ten thousand components. Obviously, achieving cost reduction in so many details is not easy, as many components are indispensable.
We are excited about the design of CyberCab and look forward to introducing more revolutionary products in the future. Clearly, no other car manufacturer is currently engaged in the work we are doing. We believe our vehicles are transformative, akin to buying a walking robot that relies on wheels during movement. At Tesla, we will delve into how to better coordinate numerous components, how to effectively integrate resources in the production process, and we also strive to achieve the fastest speed in the delivery process No other car manufacturer has such a comprehensive and systematic approach as we do, considering all aspects of design, manufacturing, production, and delivery logic. Just imagine, starting from scratch, from designing individual components, to subsequent manufacturing processes, to the engineering technologies to be used in manufacturing, and finally to vehicle delivery, we can handle all these processes properly. Because we can manage all these processes, we can demonstrate unique advantages in terms of cost, efficiency, and time.
In the battery field, the development speed of the 4680 battery cells is extremely rapid. These battery cells are not only widely used in vehicle manufacturing, but also in energy storage businesses, such as in products like Megapack and Powerwall. In fact, at Tesla, the work we are engaged in is indeed challenging and full of innovation. We are always pushing forward. As a large-scale company, we have multiple business segments, each dedicated to achieving outstanding development.
Q: Progress on Semi?
A: It is expected to start trial production in the second half of 2025 and gradually expand full-scale production in the coming years. The demand for Semi is very high, and from an economic perspective, the transportation cost per mile of Semi is much lower than that of diesel trucks, making it highly cost-effective.
Semi has been tested by some cooperative fleets and companies, and has received excellent feedback. Especially, drivers from PepsiCo are highly satisfied with Semi, to the extent that they do not want to go back to driving traditional diesel trucks.
Moreover, FSD (Full Self-Driving) technology will also be applied to Semi, and Tesla plans to further enhance its autonomous driving capabilities through hardware and software upgrades in the future to improve vehicle safety and driver comfort.
Q: Progress on Roadster?
A: Roadster is a model that we attach great importance to. We hope that through this model, we can drive future energy towards a more sustainable direction.
In the development of the Roadster model, we have made some progress, and the design work is nearing completion. I believe this product will be very eye-catching. At the appropriate time, we will share more relevant information.
Q: When can Robotaxi be deployed on a large scale? And what will be the specific deployment method? Initially, will it be considered to start deployment from the Tesla fleet and then allow users to subscribe to include their vehicles in the Robotaxi fleet? In addition, as the hardware gradually upgrades from the current HW3 to the more advanced AI5, the hardware performance will become even stronger. Does this mean that with the continuous hardware upgrades, the capabilities of Robotaxi will also be further enhanced?
A: Regarding the above questions, we currently do not have a clear answer. In the hardware versions HW3 and FSD 12.5, we have witnessed the powerful hardware capabilities of the vehicles. In the future, we look forward to further exploring optimization solutions for vehicle deployment. Regarding the deployment of the Robotaxi fleet, both of the above-mentioned modes are possible. Here, I would like to emphasize the development of FSD and hardware We expect to gradually improve the performance of HW3, HW4, and future hardware versions such as AI5 to achieve a higher level of safety. For unsupervised Full Self-Driving (FSD), we also expect it to achieve outstanding safety. For users who obtain new vehicles through upgrades, system upgrades are particularly necessary. Overall, we are committed to continuously improving vehicle performance and hope to provide users with more choices after the widespread deployment of the Robotaxi fleet.
It is possible that HW3 may not reach the safety level required for unsupervised FSD. If this turns out to be the case, we will offer free upgrades to users who have FSD functionality on HW3.
Q: It's very interesting about the expansion of computing resources. You mentioned that you currently do not feel limited by computing. I would like to know how you utilize these additional computing resources. Are you expanding to build larger models, such as extending from GPT-3 to GPT-4 like OpenAI? Or do you already have a fixed model framework and mainly accelerate the learning process by increasing computing resources to improve the reliability of the model?
A: We are rapidly advancing AI-related work and plan to launch new projects in California next year. Our goal is to achieve true unsupervised FSD, which requires AI to have extremely high capabilities.
Real-world artificial intelligence is different from large-scale language models (LLMs) because it needs to process a large amount of contextual information. For example, Tesla uses internal and external cameras to achieve autonomous driving based on computer vision technology, which requires a large amount of data and training. In terms of data, our data volume is extremely large, reaching gigabytes or even higher in a short period of time.
Although the resource requirements for training unit models and real-world AI training are not yet clear, we are always committed to reducing the computational power requirements on vehicles. We expect vehicles to have strong capabilities themselves while maintaining a low computational burden during use. We believe that through more training, the resources needed for vehicles to handle different tasks will gradually decrease, just like human drivers, where beginners need full concentration while experienced drivers can handle tasks more easily.
In terms of inference, although the demand for inference computing is relatively large (e.g., 10 kilowatts), it is still small compared to our GPU cluster. Currently, we have no constraints on computational power, but how to process and filter a large amount of video training data to identify the most critical information is a challenge. The team will further enhance the system's reliability through larger-scale training and organizing and filtering video data. Additionally, Tesla is not limited to improving computational power but also enhancing system performance through better data processing methods. Furthermore, we have excellent simulation methods that can provide more data and metrics, which is a significant advantage for us.
Q: I see that many states require safety drivers to be in the vehicle from a regulatory perspective. What are your thoughts on this?
A: Indeed, each state has different requirements, with some states requiring safety drivers and others not. It involves determining whether safety drivers are required based on how many miles and how long they need to drive. Regardless, safety is of utmost importance, and we will ensure full compliance Next year, we will further follow up on regulatory dynamics and see what new regulatory requirements there will be next year.
Q: Currently, many investors are unclear about how xAI's operating model specifically benefits Tesla. Some people believe that there is a competitive relationship between these two companies, such as competition for talent and technical resources. How do you explain the collaborative relationship between Tesla and xAI and their future development direction?
A: Although xAI is beneficial to Tesla in many ways, including improving training efficiency and scale, their goals and problems to be solved are different. Tesla's AI work is mainly focused on practical applications, such as autonomous driving, which requires efficient inference computing power, and Tesla excels in this area. Unlike xAI's more general artificial intelligence work, Tesla's AI is designed for vehicle operations/robotics fields.
Tesla's AI focuses on real-world applications, such as visual and audio processing for vehicles, which require highly efficient inference computing.
Tesla's AI chip design and team are world leaders in the industry, and currently no cloud computing company can match them. Although xAI is not perfect, it has played an important role in helping Tesla improve AI training methods and make operations more robust. In conclusion, although Tesla and xAI address different problems, their collaboration is very helpful for Tesla's goals.
Risk Disclosure and Statement of this article: Dolphin Research Disclaimer and General Disclosure
