Tesla conference call: This year will be the most important year in history, significant breakthroughs in FSD, and the robotics business may surpass automobiles in the future

On January 30th, Tesla announced its fourth-quarter results and held an analyst conference call. Tesla stated that 2025 will be the most important year in the company's history, as it will launch fully autonomous driving services and achieve other key goals. The key points from the call are as follows:

1. In the fourth quarter of 2024, Tesla achieved a record vehicle delivery volume, with an annualized delivery rate approaching 2 million vehicles. The Model Y became the best-selling vehicle globally in 2024 (not just electric vehicles, but all types of vehicles).

2. Tesla has successfully reduced the cost of each vehicle to below $35,000.

3. Tesla believes it has made significant progress in autonomous driving technology and is about to launch fully autonomous driving services that require no human intervention, which is the core driver of the company's future value growth.

4. Tesla plans to launch its unmanned fully autonomous driving service in Austin in June 2025.

5. Tesla expects that once fully autonomous driving is achieved, the utilization of vehicles will increase from the current 10 hours per week to 50 hours or more, resulting in significant asset value growth.

6. Tesla believes that the Optimus humanoid robot has enormous commercial potential and could even generate over $10 trillion in revenue, surpassing the automotive business and becoming the company's main source of income.

7. Tesla's energy storage business, including Megapack and Powerwall, is growing rapidly and will become an important part of the future grid. Tesla's energy storage business reached a record high in the fourth quarter and is expected to grow by at least 50% by 2025.

The full transcript of the call is as follows, translated by AI:

Travis Axelrod (Head of Investor Relations):

Good afternoon, everyone, and welcome to Tesla's Q4 2024 Q&A webcast. I am Travis Axelrod, Tesla's Head of Investor Relations, and joining me today are Elon Musk, Vibhav Taneja, and several other executives. We released our fourth-quarter financial results at 3 PM Central Time, and an updated presentation will be posted at the same link as this webcast. During this call, we will discuss our business outlook and make forward-looking statements.

These statements are based on our predictions and expectations today. Actual events or results may differ materially due to various risks and uncertainties, including those risks mentioned in our recent SEC filings. During today's Q&A session, please limit yourself to one question and one follow-up question. Please use the raise hand button to join the question queue. Before we start the Q&A, Elon has some opening remarks. Elon?

Elon Musk (CEO):

Thank you, everyone. So to summarize, in the fourth quarter, we set a record and delivered vehicles at an annualized rate approaching 2 million/year. So, congratulations to the Tesla team for their outstanding work in achieving record production and delivery The Model Y is the best-selling vehicle of 2024.

This is noteworthy. Not only is it the best-selling electric vehicle, but it is also the best-selling vehicle of any kind on Earth, with the Model Y taking the top spot. We remain focused on maximizing production and have clearly increased our investment in autonomous driving. But in reality, autonomous driving is about scaling things up tenfold; honestly, doubling growth is simply not enough. We have made many key investments in manufacturing, artificial intelligence, and robotics in 2024, which will yield tremendous returns in the future. Tremendous. In fact, the scale is so large that it is hard to comprehend.

I have said this before, and I still hold this view now. I see a path. I am not saying this path is easy, but I see a way for Tesla to become the most valuable company in the world, potentially several times the combined value of the current top five companies.

There is such a path. I believe this path is challenging, but it is achievable. So—this is mainly due to autonomous vehicles and automated humanoid robots. Therefore, our focus is actually developing in this direction.

We have laid the groundwork for this in 2024, and we will continue to build on it in 2025, but not just laying the foundation. We will build structures, build production lines, and prepare for what I believe will be an epic revolution arriving in 2026, with 2027 and 2028 being outstanding years. That is my prediction.

Currently, very few people understand the value of fully autonomous driving and how we will monetize our fleet. Some things I have been saying for a long time, and I know some people say Elon is like the boy who cried "wolf," several times now, but I tell you, this time there really is a wolf, and you can drive it. In fact, it can drive you. It is an autonomous driving wolf. For many people, their Tesla autonomous driving experience is like encountering a baby, thinking it will always be a baby. But obviously, it will not always be a baby. It will grow up. But if their last experience was "oh, autonomous driving is still a baby," now it has grown up.

Have you seen it? It can now walk and talk. This is exactly what we have. It is hard for people to understand this because human intuition is linear, while what we see is exponential progress. So, my primary suggestion is that anyone who doubts should just try it for themselves.

Have you tried it? When was the last time you tried? The only doubters are those who have not tried it. So, a car, typically used for about 10 hours a week, is a small fraction of 168 hours.

Once this car is autonomous, I roughly estimate that the weekly usage time will be at least one-third of the current amount.

That is, 50 to 55 hours a week. It can be used not only for goods delivery but also for transporting people. So even when people are sleeping, you can deliver goods in the middle of the night, or restock restaurants, or do anything else that people need around the clock.

These existing assets—things that can enhance their utility fivefold or more with just a software update—are now worth much more than the utility currently possessed I believe this will be the largest asset value growth in human history. There may be something bigger, but I don't know what it is. So, for those looking for past precedents in the rearview mirror, I actually think there are no precedents. So— but the reality of autonomous driving has arrived, and I repeat my advice once again: try driving, or let it drive you.

Now, it operates very well in the United States, but of course, over time, it will operate just as smoothly elsewhere. Yes. So we are working to increase annual production.

Our current constraint is the battery packs, but we are working to solve this issue, and I believe we will make progress in addressing it. Then, things will accelerate rapidly next year, and by 2027 and 2028, it will be very rapid. So, yes—let's talk a bit more about full autonomy. Our fourth-quarter vehicle safety report shows that vehicle safety continues to improve year-over-year, and regardless of whether supervised full autonomy is enabled, the safety differences are significant.

People have already seen the huge improvements brought by version 13 and its gradual updates, and version 14 will be a very significant advancement. We launched the Cortex training cluster at our Gigafactory in Austin, which plays an important role in the development of autonomous driving, and we continue to invest in training infrastructure at our headquarters in Texas. Therefore, the training demand for Optimus (the humanoid robot) could be at least 10 times that of vehicle training, requiring a sufficiently diverse range of effective roles.

You can choose—how many different roles are applicable to humanoid robots and cars.

Humanoid robots have about 1,000 times more use cases and complex tasks, which is much more complicated than cars. This doesn't mean that the scale of training will increase 1,000 times, but about 10 times. Now, you can train incrementally, which doesn't mean Tesla will spend $500 billion on training computation. We will obviously train Optimus to complete enough tasks to match the output required for the Optimus robot.

Over time, the cost of training will drop sharply. But this is an opportunity where I believe Optimus has the potential to generate over $10 trillion in revenue. It's crazy. Clearly, in this case, you can afford a lot of training computation.

In fact, even $500 billion in training computation would be a very cost-effective deal in this scenario. The future will be vastly different from the past, that much is certain. We are living at an incredible turning point in human history. So, the proof lies in the actual results.

We will launch the unsupervised full autonomy paid service in Austin in June. I have discussed this with the team, and we are confident in launching the initial unsupervised, driverless full autonomy in Austin in June. We have already conducted driverless full autonomy at our Fremont factory, with no one in the car, and we will soon do the same at our Texas factory.

So, every day, thousands of cars are driving unsupervised at our Fremont factory. They will soon do the same in Austin and at our other factories around the world, which is really cool. These cars are not just driving to the exact same place because clearly, they all want to park in the same spot The cars are actually programmed to park in the lanes needed to prepare for delivery.

They drive from the end of the production line to the destination parking spots every day, reliably doing so thousands of times a day to prepare for delivery. It's really cool. As I said, Tesla will walk into the world with self-driving cars in Austin in June. So, what I'm talking about is not an unattainable mythical scenario, but something that will happen in five or six months. Although we will initially test carefully to ensure everything goes smoothly, our solution is a universal AI solution that does not require precise local maps. So we just want to be a bit cautious.

It's not that it doesn't work outside of Austin; in fact, it does. We just want to test the waters first, ensure everything is normal, and then gradually dive deeper. Ensuring public safety and the safety of people inside the vehicle is our top priority. As for Optimus, clearly, I may sound crazy when making these revenue predictions. I understand that. But I believe they will ultimately prove to be accurate.

Regarding Optimus, there is a lot of uncertainty about the exact timeline because, for us, it’s not like a train arriving on time. We are designing this train, the station, and building the tracks in real-time.

So they might ask, why didn’t the train arrive at 12:05 on time? In fact, we are designing the train, the tracks, and the station in real-time. Wow, how can you accurately predict this? It’s almost impossible. The normal internal plan is to produce about 10,000 Optimus robots this year.

Can we successfully produce 10,000 by the end of December this year? Probably not. But can we successfully produce several thousand? Yes, I believe we can. Will these several thousand Optimus robots do useful things by the end of the year? Yes, I am confident they will do useful things. These Optimus robots in use at Tesla factories will provide feedback for the design of the second-generation production, which we expect to launch next year.

Our goal is to accelerate the production of Optimus. It may be the fastest acceleration in history. That is to say, ideally increasing by an order of magnitude each year. If we expect to increase by an order of magnitude each year, perhaps we will ultimately only increase by half an order of magnitude each year. But this is the kind of growth we are discussing.

Unnamed speaker: This is crazy.

Unnamed speaker: Crazy. Not 50%, but 500%. So these are huge growth numbers. But we really need—this is an entirely new supply chain. This is completely new technology. There is nothing off-the-shelf to use. We are trying our best in the Optimus project to use existing motors, actuators, and sensors. None of them can be applied to humanoid robots, regardless of the price. We have to start from the basic principles of physics to design everything to make it suitable for humanoid robots. And it is the most complex hand ever made, far ahead of any other hand design. Optimus will be able to play the piano and thread a needle. I mean, this is a level of precision that no one has ever achieved. So, this is really something very special. Yes, my long-term prediction is that Optimus will overwhelmingly become the value of the company Regarding energy, back to Earth. Back to Earth. Energy storage is a big issue and has become very important, and will become even more important in the future. It is a technology that can enable the grid to produce energy outputs far greater than what is currently possible. Because most grids currently lack energy storage capabilities. Therefore, they must design power plants to cope with very high peak loads, assuming there is no energy storage. Once you have grid energy storage and home energy storage, the actual total energy output of the grid each year will be much larger than people imagine. Perhaps at least double. This will drive demand for fixed battery packs, especially grid-scale battery packs, to a crazy extent—basically all the demand we can manufacture. So we have a second factory in Shanghai that has already started operations, and we are building a third factory. Therefore, we are ramping up the production of fixed battery storage as quickly as possible. Now there is a challenge, we must be careful to avoid taking money from one pocket to another, because for every gigawatt-hour of battery production per year, we must consider whether it is for fixed applications or mobile applications? It cannot be used for both at the same time. So we have to make this trade-off. Yes. But overall, whether it is mobile batteries or fixed batteries, the total demand for batteries will be very, very large in the future. So overall, 2025 will be a key year for Tesla. When people look back at 2025 and the release of supervised full-speed driving, the reality of artificial intelligence that really works in the real world, I think they may consider it the most important year in Tesla's history, even more important than our first car, the Roadster, or the Model S, Model 3, or Model Y. In fact, I think 2025 may be seen as the most important year in Tesla's history. No other company in the world is doing as well as Tesla in real-world artificial intelligence. I don’t even know who ranks second. Just like you ask, who is second in real-world artificial intelligence? I might need a very large telescope to see them. They are far behind.

Travis Axelrod, Head of Investor Relations: Awesome, thank you very much, Elon. Vaibhav also has some opening remarks.

Vaibhav Taneja, Chief Financial Officer: Yes. I will talk about things on Earth. As Elon mentioned, in the fourth quarter we set records in vehicle deliveries and energy storage deployments, achieving quarter-over-quarter and year-over-year growth in vehicle and energy storage volumes despite an uncertain macro environment. For this, I want to thank everyone at Tesla for making this happen, as well as the customers who helped us achieve this milestone. As we entered the fourth quarter, our focus was on reducing inventory levels in the automotive business. We successfully achieved this goal by ending the quarter with the lowest finished goods inventory in the past two years. This was accomplished by offering not only attractive financing options but also other discounts and programs, which impacted the ASP results. While we saw sales growth in almost all operating regions, we set new delivery records in the Greater China market. This is an encouraging trend as we achieved sales growth in a highly competitive BEV market In terms of automotive profit margins, we have seen a quarter-on-quarter decline, primarily due to lower ASP and the recognition of FSD-related revenue from future releases in the third quarter. Our cost reduction journey continues, and we have successfully lowered the overall cost per vehicle to below $35,000, mainly driven by material costs. Although we have increased depreciation and other expenses, we are preparing for the transition to the new Model Y, for which we recently started accepting orders in all markets. All factories will begin production of the new Model Y next month. While we are confident in our team's ability to rapidly scale production, please understand that this is an unprecedented transformation, and we do not know if other companies can simultaneously update all factories for the world's best-selling model. This change will result in a few weeks of production loss in the quarter. Therefore, due to idle capacity and other costs associated with ramping up production, margins will be affected, which is common in any product launch, but these issues will be overcome as production increases. We will launch several new products in 2025. We still plan to introduce a more affordable model in the first half of 2025 and will continue to expand our product line based on that. In terms of value per dollar, we believe our current product lineup is the most attractive in the industry, and it will continue to improve from here. As always, all our products are equipped with the industry's best software, autonomous driving capabilities, and the potential for full autonomy in the future. Although we offer a high-end experience, the total cost of ownership is close to that of non-premium competitors in the mass market. Energy storage deployments reached a record high in the fourth quarter but saw a quarter-on-quarter decline. This growth came from Megapack and Powerwall. Both businesses are still supply constrained, as Elon mentioned, and we are working to increase Megapack production, with the Shanghai factory coming online this quarter. While quarterly deployments may fluctuate quarter-on-quarter, we expect at least 50% year-on-year growth in deployments by 2025. Gross profit and margins for services and other businesses increased year-on-year but decreased quarter-on-quarter. This is mainly due to increased service center costs and declining profits from the used car business. Services and other businesses primarily support our new car business, especially in terms of total cost of ownership impact. Therefore, while we manage these businesses to achieve positive values on a GAAP basis, we do not expect their margins to be the same as other businesses. There is significant uncertainty regarding tariff issues. Over the years, we have made efforts to localize our supply chain in each market, but we still rely heavily on components from around the world, covering all businesses. Therefore, the imposition of tariffs, especially reciprocal tariffs, is likely to impact our business and profitability. Our operating expenses have increased both year-on-year and quarter-on-quarter. The largest driver of the increase is R&D, as we continue to invest in AI-related projects. The remaining growth comes from enhanced sales capabilities and marketing efforts for the referral program. In 2025, we expect operating expenses to increase to support our growth plans. It is worth noting that the net income for the fourth quarter was affected by a $600 million fair value adjustment for Bitcoin due to the adoption of a new accounting standard for digital assets, under which we will make fair value adjustments in each subsequent reporting period The free cash flow for the fourth quarter was $2 billion. Despite an increase of $2.4 billion in capital expenditures for 2024, we are still able to generate $3.6 billion in free cash flow for the year. The efficiency of capital expenditures is a key focus for us. Although we have invested in AI-related projects, we have made very targeted expenditures to achieve immediate benefits. The construction of Cortex has accelerated due to the launch of FSD version 13. So far, the cumulative capital expenditure related to AI, including infrastructure, is approximately $5 billion. For 2025, we expect capital expenditures to be flat compared to last year. Overall, as Elon said, 2025 will be a key year for Tesla. We have made and will continue to make many investments that will lay the foundation for the next phase of growth. This will be—now I'm about to get into the production line—it will be beyond imagination. We are just laying a solid foundation for this. That concludes my remarks.

Questions and Answers

Travis Axelrod, Head of Investor Relations: Questions and Answers

Unnamed Speaker: Great, thank you very much, Vaibhav. Now we will move into the investor Q&A session, starting with say.com. The first question is, is the unsupervised FSD still planned to be released this year in Texas and California? What obstacles still need to be overcome to achieve this goal? I think you mentioned the situation in Texas, so what about California?

Unnamed Speaker: Yes, I am confident that we will also release unsupervised FSD in California this year. In fact, I think it is very likely that we will release unsupervised FSD in many areas of the U.S. before the end of this year. As I said, we are just dipping our toes in the water, then a few toes, then the whole foot, and then the leg, ensuring everything goes smoothly. The safety standards we are looking for must be significantly higher than the average driver. So it’s not just slightly safer than human drivers, but much safer than human drivers. Because once any accident involving autonomous vehicles occurs, it immediately becomes global headline news. Although about 40,000 people die in traffic accidents in the U.S. each year, most of these do not attract any attention. But if an autonomous vehicle has a minor accident, that becomes headline news.

Vaibhav Taneja, Chief Financial Officer: We want to avoid that.

Elon Musk, Chief Executive Officer: Yes. So, in fact, the only reason we are delaying is due to excessive caution. But people can feel how the unsupervised FSD performs just by letting the car drive you around the city; you just need to look at how many times you need to intervene for safety reasons, rather than where you want to intervene or worry about. You will find that now it is almost very rare, and over time, it hardly happens anymore.

Travis Axelrod, Head of Investor Relations: Great, thank you very much. The next question is, have there been discussions with other car companies regarding FSD licensing? Elon Musk, CEO: Yes. Now we see a lot of major automotive companies showing great interest in licensing Tesla's full self-driving technology. We usually say that the best approach is to take apart our cars, see the placement of the cameras, and understand the thermal demands of Tesla's AI inference computer. This is more effective than sending some drawings of cats. And we will only consider those situations where there are very high sales volumes under licensing. Otherwise, the complexity is not worth taking on. We will not let our engineering team engage in tedious discussions with other companies' engineering teams without unsupervised full self-driving working normally across the United States. I think once it becomes obvious that unless you have FSD, you will fall behind, other manufacturers' interest in licensing FSD will be very high.

Unnamed speaker: That's great, thank you very much. The next question is, has Optimus now reached design lock, ready for production in 2025?

Unnamed speaker: Optimus has not yet reached design lock. So, as I said, we are designing this train while it is already running on the tracks, and we are also redesigning the tracks and stations.

Unnamed speaker: It is constantly evolving, yes.

Unnamed speaker: Yes, it is evolving rapidly. It is progressing quickly in a good direction. In fact, it is quite amazing. The team is doing an excellent job. I believe we have the best humanoid robotics engineering team in the world, really, and we also have all the other necessary conditions because you need a great factory package. You need excellent power electronics. You need great charging capabilities. You need outstanding communication systems that can support Wi-Fi and cellular connections. Of course, you also need real-world artificial intelligence, and then you need to be able to scale production to a significant—huge level, so you must design these things for manufacturing. In fact, what other companies are missing is that they do not have real-world artificial intelligence and do not have the capability to scale manufacturing to millions of units per year.

Unnamed speaker: I think it is an underestimated fact that the industrialization of design is completely different from mere design and production.

Unnamed speaker: Yes, prototyping is basically trivial. Prototyping is simple. Production is the challenge. I have been saying this for years. The problem is that some people have never been involved in production or manufacturing, and as a result, they think that once a spark of inspiration is designed, they can miraculously produce a million units every year. This is completely wrong. There must be a—Hollywood-style story where it is described. In fact, the problem is manufacturing, but no one mentions this because it does not fit the narrative. The Hollywood story goes like this—a lonely inventor in a garage shouts, "I've got it!" and then suddenly files a patent and can produce millions of products. But in reality, this story misses 99% of the details. Like the old saying goes, "1% of the product is inspiration, 99% is sweat." The Hollywood story is 1% inspiration, ignoring 99% sweat—that is the actual solution to how to make the initial prototype manufacturable and efficiently produced, ensuring the product is reliable, low-cost, consistent, and not easily damaged And all of this is 100 times more difficult than prototype design, at least.

Unnamed spokesperson: And then you have to get it to the destination and deliver it. Yes.

Unnamed spokesperson: Right. You have to study all these regulations, the regulatory requirements in Brazil and other countries globally, which is very challenging.

Travis Axelrod, Head of Investor Relations: Okay, thank you. The next question is also about Optimus. When will Tesla start selling Optimus, and what will the price be?

Elon Musk, CEO: Well, this year we expect to primarily complete the use of Optimus internally at Tesla, as we can obviously easily use thousands of humanoid robots in Tesla factories to handle the most boring and tedious tasks, like those tasks that no one wants to do, and we have to beg employees to do. And the robots are completely willing to do those boring, dangerous, repetitive tasks, which are exactly the tasks that humans do not want to do. In fact, this is also the easiest application scenario for us when we let Optimus do some things, like loading the body production line. If you need to transport sheets of metal to the robot welding line, and you have to keep moving things from one fixture to another, that work is very monotonous, and that is a task Optimus can do.

Unnamed spokesperson: The one running around loading welding bolts.

Elon Musk, CEO: Yes. There are many boring, tedious, and slightly dangerous jobs that are very suitable for Optimus. So we expect to use Optimus in our factories to complete these tasks, which will help us complete this improvement cycle this year. In fact, the production version two is expected to be released sometime next year. I hope it will be early next year, but it may be more around mid-next year. Next, we need to design a production line that can support the production of 10,000 Optimus robots per month, rather than a production line that produces 1,000 robots per month. So, when you design for a production line that produces 1,000 robots per month, it is actually very difficult to approach that 1,000-unit output in a short time. For any given production output, it takes time to reach its potential. The production line we are currently designing is for producing about 1,000 Optimus robots per month, the next line will be for 10,000 per month. The next line after that will be for 100,000 per month. I think the launch of version two may be around the second half of next year, but due to a lot of uncertainties, this is still a very rough estimate. But it can be anticipated that there will be exponential growth, from no one receiving humanoid robots to these robots pouring out crazily.

Vaibhav Taneja, CFO: We can't make enough.

Elon Musk, CEO: We will always be in a situation of "not being able to make enough." Even if the price is high, demand will not be an issue. And as I said, once we start to reach a stable production state of over 1 million units per year, I believe the production cost of Optimus will be below $20,000 If you compare the complexity of Optimus with that of cars, the overall quality and complexity of Optimus is far lower than that of cars. Therefore, I expect that when production reaches a level similar to that of the Model Y (over 1 million units per year), the cost of Optimus may be only about half. As for the price of Optimus, it is determined by market demand.

Travis Axelrod, Head of Investor Relations: Okay, thank you very much. The next question is, how is the mass production progress of the Tesla Semi truck, and how will it affect revenue and scale?

Vaibhav Taneja, Chief Financial Officer: I'll take that question. We just completed the topping out of the roof and walls of the Semi factory in Reno. Given the weather conditions in Reno, the progress is going well. Next, we will begin preparations for the installation of machinery. The first batch of production for the high-volume Semi design is expected to be completed by the end of 2025, with gradual increases in production starting in early 2026. But as we mentioned before, the total cost of ownership (TCO) of the Semi is almost an undeniable advantage. It is actually very similar to Optimus, primarily determined by the price people are willing to pay, and as a total cost of ownership, it is cheaper than any other mode of transportation. So, once we achieve scaled production, it will make a significant contribution to Tesla's revenue. I think it's hard to estimate the specific amount of that contribution accurately. Elon, do you have anything to add?

Vaibhav Taneja, Chief Financial Officer: No, I mean, I do think that once the Semi truck has autonomous driving capabilities, it will become very valuable. In fact, there is a shortage of truck drivers in the U.S., which is a major limiting factor for transportation. And people do get fatigued after all—I have great respect for truck drivers because it is a very tough job. But because this job is so hard, there are not many people willing to do it. In fact, I believe if I understand correctly, the number of people entering the trucking profession is far less than those leaving the industry.

Unnamed spokesperson: Yes.

Unconfirmed spokesperson: So when you think about it, well, that is indeed the case. So, when you consider that more people are leaving the trucking profession than entering it, over time, we will face a real logistics problem. Therefore, autonomous driving will be very important to meet this demand. So, um, yes... I don't know, in this context, does the opportunity of hundreds of billions a year still make sense today? I think it makes sense. It's not meaningless; it could be about a $10 billion opportunity each year. Um, maybe it will reach $1 billion a month at some point. But all of this will seem trivial compared to Optimus. So, yes, $1 billion a month is a lot, but it is also just about 1% of Optimus.

Unconfirmed spokesperson: Thank you very much. We discussed the next question at the beginning. So continuing to ask, does Tesla expect to need to upgrade vehicles with Hardware 3? If so, what is the expected timeline, and what impact will it have on Tesla's capital expenditures? I will pass this question to Costner.

Unconfirmed Speaker: They are asking some tough questions all day long.

Ashok Elluswamy, Director of Autonomous Driving Software: I think we really haven't stopped working on Hardware 3. We are still releasing software versions. We recently released version 12.6, which is like a baby version of V13, but it is a significant improvement compared to previous versions. People are still looking for ways to compress larger models into smaller ones. So we haven't given up on Hardware 3 and are still continuing to work on it. It's just that the release is slightly behind that of Hardware 4.

Unconfirmed Speaker: Great, thank you, Ashok.

Elon Musk, CEO: Yes. I mean, the honest answer is that we will have to upgrade the Hardware 3 computers for those customers who purchased the Full Self-Driving (FSD) package. That's an honest answer. It will be painful and difficult, but we will get it done. Right now, I'm a bit relieved that not many people bought the FSD package.

Unconfirmed Speaker: Okay, thank you, Elon. Next question, has Tesla given up on driving growth for the solar roof product?

Unconfirmed Speaker: Oh, Mike, go ahead. Yes.

Unconfirmed Speaker: Oh, yes, I can answer that. Yes, the solar roof is still a core part of the residential product lineup, and although it is a high-end product, it still attracts a lot of customer interest. We have made several versions of engineering improvements to make the product easier to install and distribute by reducing the number of SKUs. Recently, we have stopped doing installations directly and are driving growth through our nationwide network of certified installers. Many installers have been working with us for years, installing solar roofs.

Elon Musk, CEO: Yes. In fact, this has proven to be a better way—just to supply the roof products directly to the roofing industry. Especially when someone is re-roofing or building a house from scratch, clearly, this is the most effective time to install a solar roof, rather than installing a solar roof on a house that still has 20 years of roof life left, which doesn't make economic sense. But if it's a new house, or the roof needs to be replaced anyway, then installing a solar roof makes a lot of sense. And it's a high-end product. It's like the Model S, Model X type of products, like a high-end product. I think it looks really cool. And your house—I mean, your house generates electricity, and if you combine it with Tesla's Powerwall battery, then you can achieve self-sufficiency. Even if the grid goes down, even if it's down for a few days, your house can still operate normally. And your roof looks great. So—I recommend anyone who can afford it to install a Tesla solar roof and Powerwall. Your home's life may depend on it. And from a convenience perspective, your kids won't be yelling at you because they can't use their computers due to a power outage, or because their phones are dead and they can't make calls Vaibhav Taneja, Chief Financial Officer: That's true.

Elon Musk, Chief Executive Officer: That's true, it will indeed happen. Yes. You can't even make a call because your phone is out of battery.

Travis Axelrod, Head of Investor Relations: Thank you very much. The next question has already been discussed at the beginning, so we will skip it. The last question comes from Say.com, what are the technological breakthroughs of FSD V14? Given that V13 has already covered the technology from photons to control.

Elon Musk, Chief Executive Officer: Well, we have gone far beyond the stage of photons to output. We are now in a stage that fully relies on neural networks; it has been a while since we moved from photons to control. We are just improving the neural networks. I think we can delve into some technical details to some extent. I must say, I am still amazed at the effectiveness of autoregressive transformers in solving various problems. I mean, Ashok, is there anything you would like to add without revealing family secrets?

Ashok Elluswamy, Director of Autonomous Driving Software: I think besides what we have already released on X, well, yes, we continue to significantly scale up the model. We expanded a lot in V13, but there is still room for growth. So we will continue to scale up the model. We will further increase the context length. Currently, the memory is somewhat limited, and we want to add more memory. At the same time, we also want to provide more minutes of contextual information for driving. They will better recognize audio and emergency vehicles, adding more tricky situation data from the entire fleet, including any interventions or user interventions. We are incorporating this data into the dataset. So basically, we are expanding training computation, dataset size, model scale, model context, and all reinforcement learning objectives in every aspect.

Travis Axelrod, Head of Investor Relations: Okay. Now we move into the analyst question segment. As a reminder, please unmute yourself when asking questions. The first question comes from Daniel Rosca at Bernstein. Daniel, please unmute and ask your question.

Analyst: Hello everyone, I am Daniel from Bernstein. Elon, Tesla's stock price clearly reflects many of the anticipated benefits you mentioned today, but achieving those "difficult but achievable" goals you talked about will take some time. What different things are you currently pushing the Tesla executive team to do to accelerate innovation and realize the value you described for the company?

Elon Musk, Chief Executive Officer: Well, I mean, we are working hard to refine real-world AI, and we are making rapid progress every week, if not every month, usually there is progress every week. I spend a lot of time working with the Tesla AI team and the Tesla Optimus team. I mean, I go where the problems are, that's basically it If I haven't talked to a certain Tesla executive for a while, I would say, "Hey, we haven't seen each other in a long time." Then I would say that because their work is outstanding, I don't see them often because I go to where the problems are. If we are not at the problem, then we need to focus on what the next biggest challenge is. So, obviously, Optimus faces many challenges in how to solve these problems. Autonomous driving also has many challenges, but we are making rapid progress in both areas. Yes.

Analyst: Okay. I mean, it sounds like you are confident that the necessary conditions are in place, right? If we look back over the past 12 months, some of them may have been achieved, but what key performance indicators (KPIs) do you think can tell you whether progress is on track and whether the conditions are all in place? Or, from another angle, where do you think the most likely problems might arise? Would you say, "Hey, I need to go back and change something to better support the team?"

Elon Musk, CEO: Well, I mean, I think the predictions I make here will be very accurate. It’s worth noting that sometimes people say, "Oh, Elon is always late." That’s actually not the case; the problem is that the media only reports when I’m late and never reports when I finish early. So, of course, I’m a bit optimistic, but not to the point of being overly optimistic. There have indeed been many instances in the past where we actually finished ahead of schedule, like the completion of the Shanghai factory. In fact, factory completions are usually ahead of schedule, not delayed. Like I said, I’m very confident that we will roll out unsupervised Full Self-Driving (FSD) Teslas in Austin and several other cities in the U.S. by the end of this year. Next year, it may roll out nationwide, at least in many parts of North America. As for next year, our limiting factor may simply be regulatory issues. For example, Europe is really complicated; it’s made up of layers of regulations and bureaucracy that need to be addressed. There’s a joke that says, in the U.S. it’s innovation, in Europe it’s regulation. That is to say, guys, there are too many referees on the field. For instance, for us to just release supervised FSD in Europe, even if it works very well, we have to go through a lot of paperwork, especially in the Netherlands, which is our main regulatory body. Then the Netherlands submits it to the EU, I think in May. Then there’s a committee of EU countries. We expect it to be approved by then, but we can’t do anything to speed up the process. It actually seems like no one can speed it up. I guess all the countries have to vote in some way to make it happen before May; otherwise, it’s not possible. So, when will Europe allow unsupervised FSD? I guess, next May? I don’t know; we have to wait for when the EU meets again. Sometimes it’s a 12-month cycle, sometimes it’s a 6-month cycle.

Travis Axelrod, Head of Investor Relations: Thank you very much. The next question will come from Adam Jonas of Morgan Stanley. Adam, please feel free to unmute and ask your question Adam Jonas: Thank you, everyone. So, Elon, you once mentioned your views on LiDAR, especially for autonomous vehicles, saying that LiDAR is a crutch and a foolish thing to do. I remember you telling me that even if it were free, you wouldn't use it. Do you still think that way?

Elon Musk, CEO: Yes.

Adam Jonas: Can you elaborate on that, or do I have another question?

Elon Musk, CEO: Look, we even have radar in our cars, and we turned it off.

Adam Jonas: Got it, okay. So, you are indeed making cars, right? People might think you're crazy, but clearly, you're doing well.

Elon Musk, CEO: Obviously, humans don’t shoot lasers from their eyes when driving, unless you're Superman. But humans rely solely on passive vision when driving; they use their eyes and neural networks, brain neural networks. So biologically—the digital version of eyes and brain is cameras and digital neural networks or AI. Therefore, the entire road system is designed for passive optical neural networks. The entire world’s road system isn’t designed this way, but everyone expects it to work differently. That’s why a universal solution for full autonomous driving is so obvious. Rather than a solution for every neighborhood and every street, which is very hard to maintain, that’s what our competitors are doing. Exactly.

Analyst: I mean, LiDAR doesn’t work in the fall, guys.

Elon Musk, CEO: LiDAR has many issues. It’s not that I particularly hate LiDAR. SpaceX's Dragon spacecraft uses LiDAR to dock with the space station, and I personally led that program. I have no fundamental aversion to LiDAR; it’s just not the right solution for driving cars on the road.

Analyst: Right, you understand how LiDAR works, I get it.

Elon Musk, CEO: We actually designed and built our own LiDAR; I personally oversaw that project, and the engineering decisions were made by me. It was my decision to use LiDAR on the Dragon spacecraft, and I personally supervised that engineering project. So, I’m saying if I thought LiDAR was the right solution for cars, I would use it, but it’s not.

Analyst: Okay.

Elon Musk, CEO: Just as a follow-up, at CES, you mentioned—I’m paraphrasing—that any AI will be able to perform any cognitive task not involving atoms within three to four years. This means, Elon, that before the end of President Trump’s term, AI will rapidly enter the physical world, into the realm of photons and atoms. And I’m wondering, considering your collaboration with the government, how confident are you that the U.S. will have enough manufacturing capacity and supply chain infrastructure to deliver on your excitement about physical AI by the end of this decade? We now seem very vulnerable.

Elon Musk, CEO: At Tesla, obviously, we think manufacturing is cool. At SpaceX, we also think manufacturing is cool. But overall, for talented Americans, they are not just working at my company, surpassing me and my team; we need to make manufacturing cool again in America. Honestly, I think people should shift from law and finance to manufacturing. This is my honest opinion. It's both a compliment and a criticism. We have too many talents in law and finance in America, and there should be more talent invested in manufacturing. So yes. At Tesla, we ensure that even in the event of heightened geopolitical tensions, we can continue to manufacture our products.

Travis Axelrod, Head of Investor Relations: Okay, thank you all. The next question comes from Pierre Ferragu at New Street. Pierre, feel free to unmute your microphone.

Pierre Ferragu: Hi, thank you for taking this question. So, I have a question about deploying robotics in Austin in June. That's good news. I wonder if that means I can drive to Austin in June and personally supervise my car, or will you be using your own fleet for testing?

Elon Musk, CEO: It will be our fleet testing. This is our "trial" phase. We will be very careful to check and ensure nothing is missed. However, in June, there will be autonomous ride-hailing services in Austin. Then, we will expand to other cities in the U.S. as quickly as possible. I expect that by the end of the year, we will have unsupervised operations in several cities with our internal fleet. Then, possibly next year, people can add or remove their cars from the fleet. Just like Airbnb, you can choose to add your house or room to Airbnb's inventory or not. If you go away for a month, you can—or whatever the situation is—you can let others use your house. Then... anyway, that's for next year because we want to ensure all issues are addressed. A lot of things are not like splitting atoms; more work is needed to ensure everything operates efficiently, ensuring people can summon a car, the car can arrive at the right place, do the right thing, the payment system works properly, and billing is handled correctly.

Pierre Ferragu: Okay. So, my question is, I have a Tesla, and with FSD, I have to keep my eyes on the road at all times. It's so boring because I hardly need to intervene. The most annoying part is that I can't check my emails. Are you considering introducing a "no supervision" mode where I can take my eyes off the road to check my emails, just needing to pay attention to the road five seconds in advance and then return? Is this something you are working on? Because it seems like it's very close to realization. Can we be sure it will be achieved within this year? This is a bit of a selfish question, to be honest Elon Musk, CEO: Yes. We just think that before we allow people to view emails and text messages, we need to be very sure that the probability of injury is very low. In fact, we are currently in a situation that you may have encountered yourself, where people switch to manual driving mode to check their text messages. This way, the computer will no longer reprimand them, and then they will switch back to autopilot mode after checking their messages, which is obviously unsafe.

Pierre Ferragu: Clearly unsafe.

Elon Musk, CEO: Significantly unsafe. Much more unsafe than letting people occasionally check their messages without being reprimanded by the computer. But we just want to be cautious about this. We are currently in a state that is neither "here" nor "there." But I think this situation won't last long. Yes, we are now in a contradictory situation where people switch their cars from autopilot mode to manual mode to avoid being reprimanded by the computer, using their knees to control the steering wheel while not looking outside.

Vaibhav Taneja, CFO: As Elon said, if the system has issues, especially when people are not looking at the road, that is very dangerous. We are working hard to avoid this situation. The system's capabilities are close, but not fully realized yet. That’s why he said we are just testing the waters and gradually adapting. Thank you.

Elon Musk, CEO: In any case, it’s not far off. But we want to ensure to prove to ourselves and to regulators that the car is absolutely safer in autopilot mode than in manual mode, and that’s our goal. So this probably requires a few more months of time.

Ashok Elluswamy, Director of Autopilot Software: In terms of safety, we released our safety report today, which shows that Tesla has an accident every 5.9 million miles driven, while without autopilot, there is an accident every 700,000 miles driven.

Elon Musk, CEO: So we are developing towards an order of magnitude difference.

Ashok Elluswamy, Director of Autopilot Software: Yes, it’s about 8.5 times.

Elon Musk, CEO: Yes, that’s great.

Travis Axelrod, Head of Investor Relations: Alright, the last question comes from Dan Levy at Barclays. Dan, feel free to unmute your microphone.

Dan Levy: Great. Good evening, thank you for taking my question. You mentioned that as part of promoting sustainable energy more broadly, sustainable transportation must be promoted. I know we have heard President Trump talk about plans to overturn the electric vehicle mandate. I think there is a general belief that since regulations are the driving force behind the adoption of electric vehicles, this could slow down the adoption of electric vehicles in the U.S. So, based on your past comments on the necessity of promoting sustainable transportation, what are your views on the right policies the U.S. should adopt? Elon Musk, CEO: Right now, I believe sustainable transportation is inevitable. I am very confident that all modes of transportation will be autonomous electric, including aircraft; it cannot be stopped, just like you cannot stop the emergence of external combustion engines, steam engines, or internal combustion engines. Even if you are the biggest opponent on Earth against these new types of cars, you cannot stop the emergence of cars. The emergence of electric vehicles is unstoppable. The only limitation to the development of electric vehicles is the range issue, and that problem has now been almost solved.

Dan Levy: Okay. So, as a follow-up question, Elon, you have said that you are willing to sell cars with almost no profit to get more cars into the market. The statement released today mentioned that the acceleration of the autonomous driving process will indeed affect sales growth. So, can you talk about how you hope to put more vehicles into the market as FSD progresses to fully leverage your technological advancements?

Elon Musk, CEO: I'm not sure I fully understand the question, but we have a lot of cars; there are already millions of cars on the road.

Unnamed spokesperson: So the question is, Dan, are you asking how we combine future growth with FSD?

Travis Axelrod, Head of Investor Relations: Please continue, Dan, unmute your microphone.

Dan Levy: Yes, more of a question about, considering the progress of FSD, how aggressively are you willing to push car sales?

Elon Musk, CEO: Currently, the limitation we are trying to solve is battery production, not demand. So, right now, in the first quarter, we are undergoing massive factory upgrades, such as the Model Y upgrade, which will obviously impact production. But the issue we are facing now, in fact, I just discussed it with the executive team before the call, is how to increase the total gigawatt hours of battery production this year under any circumstances. That is our output limitation.

Travis Axelrod, Head of Investor Relations: Okay, thank you, everyone. So we will stop here for today. Thank you all for your questions, and we look forward to seeing you again next quarter. Thank you, everyone, goodbye