Tesla CEO Elon Musk and other executives discussed the commercial rollout of Tesla's Robotaxi, the acceleration in FSD adoption, the production timeline for Optimus, and other key topics; despite Musk's reiteration of confidence in autonomous driving and humanoid robotics during the earnings call, Tesla's Chief Financial Officer revealed that the company's costs will increase by approximately $300 million due to U.S. tariffs.
On Wednesday, after the market closed, Tesla released its most disappointing quarterly financial report in recent years, making the subsequent earnings call a focal point for Wall Street analysts.
In the second quarter of this year, Tesla achieved significant operational milestones. On one hand, Tesla launched and expanded its commercial Robotaxi service in Austin, Texas, setting an ambitious goal to cover half of the U.S. population by the end of the year. On the other hand, the cancellation of the electric vehicle tax credit by the U.S. government has put pressure on Tesla—CFO DevOps Ninja stated: 'The primary change is the elimination of the $7,500 IRA electric vehicle tax credit by the end of this quarter (Q3).'
During the earnings call, Tesla CEO Elon Musk and other executives discussed the commercial rollout of Tesla's Robotaxi, the acceleration in FSD adoption, the production timeline for Optimus, Dojo and AI infrastructure, and the cost pressures from tariffs.
Despite Musk's reiteration of confidence in autonomous driving and humanoid robotics during the earnings call, Tesla's Chief Financial Officer revealed that the company's costs will increase by approximately $300 million due to U.S. tariffs, with about two-thirds of the impact affecting the automotive business. Additionally, with the cancellation of U.S. zero-emission penalties, the sales of regulatory credits (carbon credits) will decline, which will rapidly affect Tesla's total revenue.
During the earnings call, Tesla's stock price fell by more than 4% in after-hours trading.
Participants
Chief Executive Officer—Elon Musk
Chief Financial Officer — DevOps Ninja
Vice President of Vehicle Software — Ashok Elluswamy
Vice President of Vehicle Engineering — Lars Moravy
Vice President of Energy — Mike Snyder
Vice President of Investor Relations — Travis Axelrod
The following are the key points from Tesla's Earnings Reports:
Robotaxi Launch and Expansion: Tesla has launched its first commercial Robotaxi service in Austin, which has already attracted paying customers. The company aims to expand the service to cover half of the U.S. population by the end of 2025, subject to regulatory approval.
Robotaxi Mileage and Safety Indicators: As of the second quarter of 2025, Ashok Elluswamy revealed that the Robotaxis in Austin have exceeded 7,000 miles without any significant safety-critical interventions.
Autonomous Ride-Hailing Service Expansion: Plans to expand services in markets such as the San Francisco Bay Area, Nevada, Arizona, and Florida, pending regulatory approval. The service area in Austin is planned to expand more than tenfold.
FSD Adoption Acceleration: Since the rollout of FSD version 12 in North America, FSD subscription rates have increased by 45%, and overall penetration has increased by 25% since the introduction of FSD 12 and B-38 in North America.
Model Y Performance: In June, the Model Y became the best-selling car in Turkey, the Netherlands, Switzerland, and Austria. According to Elon Musk, it remains the 'best-selling car of any type globally.'
Automobile Revenue Growth: Despite a reduction in regulatory credit income, automobile revenue grew sequentially by 19%, primarily due to new Model Y sales and a favorable product mix that raised average selling prices (ASPs).
Tariff Impact: Tariff costs increased by approximately $300 million sequentially, with two-thirds of the impact on the automotive segment. Due to delays in the manufacturing and sales cycles, further increases are expected in subsequent quarters.
Automobile Product Mix: Production of lower-cost models will begin in Q3 2025, and due to a focus on maximizing deliveries before carbon credits expire, the production rate will be slower than initially anticipated.
Energy Business Performance: The company achieved record power deployment in Q2 2025, with sequential improvements in energy and solar margins, reaching the highest gross margin in the history of the division.
Energy Business Tariffs and Legislative Headwinds: Despite headwinds from tariffs and the early expiration of consumer credits, customers—particularly those in industrial energy storage—have been willing to absorb some of the tariff impacts.
Operating expenses: Operating expenses have increased sequentially due to higher research and development investments in artificial intelligence, increased stock-based compensation, and depreciation of AI computing infrastructure. It is expected that expenses in these areas will continue to grow.
Free cash flow: Free cash flow was $146 million, as the increase in operating cash flow was offset by higher capital expenditures.
Capital expenditure outlook: Management expects capital expenditures to exceed $9 billion in 2025. Regulatory credit income decline: According to DevOps Ninja, a legislative change eliminating emission penalties is expected to rapidly reduce new regulatory credit income.
Optimus production timeline: Elon Musk predicts that three prototypes of the humanoid robot Optimus will be completed within three months, with plans to start production early next year. The medium-confidence target is to reach a production rate of approximately 100,000 units per month within five years.
Dojo and AI infrastructure: Musk stated that Dojo 2 will be running at scale sometime next year, with a capacity of around one exaflop equivalent, and the new AI factory is expected to begin production by the end of next year.
Supercharging and service center profitability: Supercharging, insurance, and service centers have all contributed to the sequential improvement in the margin of the services and other business segments.
Bitcoin-related volatility: Other income in the second quarter increased by $284 million due to changes in the market cap of Redcom Holdings, compared to a loss of $125 million in the previous quarter; this volatility is attributed to fluctuations in the price of Bitcoin.
FSD safety report: Management highlighted the newly released vehicle safety report, which shows that cars using FSD are 'statistically safer than those not using FSD.' The cited statistics indicate that cars equipped with FSD are ten times safer than those without it.
The full transcript of the earnings call is as follows:
Elon Musk: Thank you, Travis. We have had a very exciting quarter. We successfully launched Robotaxi, offering the first paid rides without a driver in the front seat in Austin. As some may have noticed, we have expanded the service area in Austin. It now covers a larger area and longer routes, and it will continue to grow. We expect to significantly increase the service area within one or two weeks, far exceeding what our competitors have done. We are obtaining regulatory approval to launch in the Bay Area, Nevada, Arizona, Florida, and other places. Once we receive approval and demonstrate safety, we will roll out the autonomous ride-hailing service across most of the country. I believe we may achieve autonomous ride-hailing for half of the U.S. population by the end of the year. This is our minimum goal, but it is contingent on regulatory approval. I think we can do this technically. Assuming we get regulatory approval, we may be able to address the needs of half the U.S. population by the end of the year. We are being very cautious. We don't want to take any risks, so we will proceed cautiously. However, the size of the service area and the number of operational vehicles will grow exponentially.
Other notable points: The Model Y became the best-selling car in Turkey, the Netherlands, Switzerland, and Austria in June. I believe it remains the highest-selling car of any type in the world. Autonomy is a significant factor. Even just supervised autonomy is a huge selling point.
It's worth noting that we have not yet received approval for supervised FSD in Europe. We believe that once we can offer the same experience to customers as in the U.S., our sales in Europe will improve significantly. This is very important. We have been working with the Netherlands, and I think we are close to getting approval there. Then it must be submitted to the EU. It's really a Kafkaesque problem, but we will get approval. I think some Europeans will have the same experience that some Americans have in the U.S., hopefully at least partially this quarter. We also face some regulatory challenges in China, which we hope to resolve soon. Currently, we cannot provide services there. In the U.S., once we are confident in the safety of different geographic areas, we will relax the requirement for people to be highly focused on the road. This is a common complaint. It actually creates a strange safety issue where people turn off the autonomous driving, then do something like changing the radio station or looking at their phone, and drive with their knees, and then restart the autonomous driving, which is less safe than keeping it on. This experience will improve over the next few weeks. The production release of the autonomous driving is actually a few months behind what people are experiencing with the Robotaxi in Austin.
Now that we have launched Robotaxi in Austin, we will re-add these elements, so the autonomous driving experience for people outside of Austin will see a significant improvement. As you can see, autonomy is key. We need the physical product; without it, you cannot achieve autonomy.
We launched the Tesla restaurant, which has been very popular. It has garnered global attention, which is unusual for a restaurant. It is a very special place. If you are in the Los Angeles area, it is worth visiting. In a drab urban landscape, it is a beacon of hope. It really is a wonderful experience. The team there is doing a great job.
In terms of full self-driving, we continue to make significant improvements, solely through software. We expect the number of parameters to increase to what we believe can be ten times the current number. This is a tricky thing because when you increase the number of parameters, you run into memory bandwidth bottlenecks. But we currently believe we can increase the number of parameters from what people are currently experiencing by tenfold. Therefore, there is still a lot of room for improvement on the existing hardware.
Despite facing tariffs and supply chain challenges, our energy business is growing well. We have upgraded the Megapack, making it even better. We achieved record power deployment in the second quarter. I think the demand for batteries is enormous. The continuous output power of the U.S. grid is about one terawatt, but the average usage is less than half a terawatt. If you add batteries to the mix, you can run power plants 24/7 at full capacity, effectively doubling the annual energy output of the U.S. This is a big deal.
Optimus (humanoid robot), we are gradually improving the Optimus design to reach an extraordinary level. We are currently using Optimus version 2.0, which is almost 2.5. Optimus 3 is a design that I believe is truly outstanding. As I have mentioned many times before, it will be the biggest product. This is a very difficult problem. You must design every part of it from first principles in physics. There is actually nothing off-the-shelf that works. We must train Optimus using its sensors and neural networks. We will apply the same technologies that we use for cars, which are essentially four-wheeled robots. Optimus is a robot with arms and legs. The principles of optimized AI inference that apply to cars also apply to Optimus, as they are both different forms of robotics.
Tesla is the best company in the world in real-world AI. A clear proof point is when you compare it with Waymo. Waymo's cars are the best in terms of the number of sensors. Is Google good at AI? Yes. But they are not good at real-world AI. Tesla is much better than Google. Far better. Tesla has the best inference efficiency. A key metric for AI is intelligence per gigabyte. People talk about parameters, but we should talk about gigabytes because for parameters, you can have 4-bit, 8-bit, or 16-bit parameters. The actual hardware limit is how many gigabytes of RAM you have and how many gigabytes per second you can transfer from RAM. So, it's not a parameter limit; it's a byte limit. Tesla's AI intelligence density is much higher than any other AI. I have deep insights into this through XAI.
XAI's Grok is the smartest AI overall, but it is a massive beast. Tesla has the highest intelligence density. Intelligence density will be a big issue in the future. With Optimus 3, this is the right design, and there are no significant flaws in the Optimus 3 design. We will redesign a lot of things. The prototype of Optimus 3 may appear before the end of this year, and then mass production will begin next year. We will expand Optimus production as quickly as possible and aim to reach an annual production of one million units as soon as possible. We believe we can achieve this goal within less than five years. This is a reasonable target.
In summary, 2025 is an exciting year so far. Many major milestones. Clear product demonstrations have made progress in autonomy that experts thought we could not achieve. We have completed what we said we would do. We don't always finish on time, but we ultimately complete it. The Tesla team has made tremendous progress.
I believe that if Tesla continues to execute well in vehicle automation and humanoid robot automation, it will become the most valuable company in the world. There is a lot of execution work along the way. This will not happen automatically. As long as we execute very well, I think Tesla has the potential to become the most valuable company in the world. Clearly, I am extremely optimistic about the company's future. The best way to predict the future is to make it happen. We are here to make it happen with the team. I just want to thank all our supporters; I think our future will be very exciting.
Travis Axelrod: Great. Thank you very much, Elon. DevOps Ninja also had some great points.
DevOps Ninja: Travis, as Elon mentioned, the second quarter was interesting in several ways. We started increasing the production of the new Model Y at all factories. We launched our Robotaxi service in Austin and delivered a car fully autonomously from the factory to a customer's home. Reaching this stage is a significant milestone. It required a lot of effort, and I really want to thank everyone at Tesla for making this possible. It was not an easy task, but we did it. It took some time, but we have just begun the next phase of the company.
There are many changes in the federal bill that will affect our near-term business. The primary change is the cancellation of the $7,500 IRA electric vehicle tax credit by the end of this quarter. Given this sudden change, our vehicle supply in the U.S. is limited this quarter. Since we are already in the timeframe for ordering car parts, we have rolled out all planned incentives and will start phasing them out as sales begin. If you are in the U.S. and planning to buy a car, please place your order now, as we may not be able to guarantee delivery for orders placed in late August and beyond. The bill also changed certain emission standards by reducing the penalty amounts to zero. This, in turn, will affect the sales of new renewable energy credits to other OEMs and, consequently, reduce revenue. Although we did not plan our business around such sales, it will soon impact our total revenue.
In terms of the automotive product lineup, the entire range has been updated. Globally, we have seen an increase in test drives. We are on track to start production of the lower-cost models in the first half of 2025. However, given our focus on producing and delivering as many vehicles as possible, and the complexity of launching new products before the expiration of electric vehicle credits, production will begin in the next quarter at a slower pace than initially expected. As Elon mentioned, one thing that is often underestimated is that all vehicles in our lineup are capable of autonomy. This is the biggest differentiator between us and our competitors. Our vehicles already meet the highest safety standards, but with FSD, they will continue to set new safety benchmarks in regular traffic. We released a vehicle safety report earlier today. You can see that cars using FSD are statistically safer than those not using FSD. In recent months, we have observed an increase in FSD adoption in North America, which is a very promising trend. To give you some perspective, since we transitioned to FSD version 12, we have seen a significant increase in adoption.
In terms of automotive revenue, despite a decrease in regulatory credit income, total automotive revenue grew by 19% sequentially, even though the total deliveries only increased by 14%. This was mainly due to the higher average selling price ("ASPs") driven by the new Model Y, as well as improvements in the product mix and increased absorption of fixed costs, despite higher tariff costs. We are beginning to see the impact of tariffs in the P&L. Sequentially, tariff costs increased by approximately $300 million, with about two-thirds of the impact in the automotive segment and the rest in the energy segment. However, due to delays in manufacturing and sales, the full impact will be realized in the upcoming quarters. Therefore, near-term costs will increase. While we are doing our best to manage these impacts, we are in an unpredictable tariff environment.
The margin for the energy generation and solar business improved sequentially, while deployment decreased, primarily due to an increase in high-margin power deployments. We achieved the highest gross margin for this business to date. Please note that overall deployment will vary from quarter to quarter. Industrial energy storage will play a role in driving AI and data center growth. Energy demand is rapidly expanding due to the high energy consumption of AI applications. The fastest way to scale energy is through the deployment of storage. Our customers are beginning to recognize this, and despite the business being most affected by tariffs, we see customers willing to accept some of the tariff impacts. This bill also has some negative effects on the energy business, especially due to the early expiration of consumer credits by the end of the year, which most significantly impacts the residential energy storage business. The challenges in the energy storage business indeed come from wealth and tariffs, and we are doing our best to address them. However, we will see changes in demand and profitability. The margin for our services and other businesses improved sequentially, mainly due to increased profits from supercharging stations and improved profitability of insurance and service centers. Operating expenses also increased sequentially as we continue to invest in AI projects, including additional employee-related expenses such as higher stock compensation and depreciation of AI computing. Our operating expenses, especially R&D-related spending, will continue to grow. We believe that continuing to invest in these areas to achieve long-term positioning is the right strategy, even in the current environment.
Other income grew sequentially, primarily due to the market value adjustment of Redcom Holdings, which brought a gain of $284 million in the second quarter, compared to a loss of $125 million in the first quarter. Just to remind everyone, this will continue to fluctuate based on Bitcoin prices. While operating cash flow grew sequentially, so did our capital expenditures, resulting in free cash flow of $146 million. We continue to invest in various areas, such as the Cybertruck, semi-truck production lines, and other infrastructure spending, as well as the expansion of AI programs. Our latest expectation for capital expenditures this year exceeds $9 billion. In summary, our near-term business faces challenges due to the negative impacts of the bill and tariffs.
However, our investments in AI, robotics, and energy provide us with a bright future. I would like to thank the entire Tesla team, our customers, investors, and supporters for their continued trust.
Travis Axelrod: Great. Thank you very much, DevOps Ninja. Now we will move on to questions from Say.com. The first question is, can you give us some insights into the performance of Robotaxi so far? And what is the expected rate of expansion in terms of vehicles, geographic areas, cities, and supervisors?
Ashok Elluswamy: Robotaxi has performed exceptionally well in Austin. Customers love the experience. It is very smooth, very safe, and the overall experience is excellent. We have already expanded in the first phase in Austin, and we will continue to expand, possibly more than ten times the size of our current operational area. We are also testing in many other cities, as Elon mentioned. The next one is in the San Francisco Bay Area. We are working with the government to get approval here, and in the meantime, we will launch a service with a safety driver in the front seat to speed things up. We are also testing in many other cities across the U.S., including Florida, Nevada, and others.
Travis Axelrod: Great. Thank you, Ashok. The next question is, what are the technical and regulatory barriers for unsupervised FSD for personal use? Can you provide a timeline?
Elon Musk: We are getting close to our goal. I believe that by the end of this year, unsupervised mode for personal use can be achieved in some regions. We are just being very cautious; we should not rush into this.
DevOps Ninja: No, we must ensure everything is safe before making it available to the public. We are just being very cautious.
Elon Musk: But I am confident that within this year, in some cities in the U.S., end-users will be able to use it. And, as I mentioned with the Robotaxis and customer vehicles in Austin, they will be autonomously driven from the factory to the customer. Yes. And every Tesla manufactured in the U.S. and Europe will autonomously drive from the end of the production line to the loading dock. It's just a software update.
Ashok Elluswamy: Yes, I think we will default to delivering vehicles from the factory to the Greater West and the Bay Area by the end of the year. A good car will automatically come to where you are unless you say you don't want that.
Travis Axelrod: That would be very cool. Great. Thank you, everyone. The next question is, what specific tasks is Optimus currently performing in the factory, and what is the timeline for scaling up production for external sales? How does Tesla see the contribution of Optimus to revenue over the next two to three years?
Elon Musk: As I mentioned earlier, the design of Optimus 3 has finally reached the right design. There are further optimizations, but the design of Optimus 3 does not require fundamental changes. It has all the degrees of freedom you really want or need. We will have a prototype of this design in about three months and will start sub-tooling in production. We will begin production at the beginning of next year. Production scaling is straightforward. It does not predict an S-curve for production ramp-up. When a product has a completely new design, the production speed will depend on the least fortunate and least competent elements in the entire supply chain and internal processes. The more new things there are in the product, the slower the scaling may be due to unexpected supply chain disruptions or internal errors. Predicting the end or late stages of the S-curve is easier than predicting the beginning. The beginning of the S-curve is not important for revenue purposes. The beginning of the S-curve usually has a negative gross margin as you debug many issues. That's why I feel quite confident about the things I'm predicting. I have moderate confidence in predictions for at least five years, but it's hard to predict one or two years out. That's why I think we could reach the limit in five years. If, after five years, which is sixty months, we produce around 100,000 Optimus robots per month, I would not be surprised. I would be shocked.
Travis Axelrod: Okay. Thank you very much. The next question is, can you provide an update on the development and production timeline for Tesla's more affordable models? How do these models balance cost reduction and profitability, and what impact do you expect on demand in the current economic environment?
Elon Musk: I think DevOps Ninja answered this well in the opening remarks. As we said, we started production in June and are gradually ramping up. We built some things throughout the quarter, and given that we started production in North America and aim to increase production rates by the end of the third quarter, we will continue to push our current models and avoid complexity. Then, fortunately, that will gradually fade away. We will launch more affordable models in the fourth quarter for everyone. The goal of these products is not to negatively impact revenue or gross margin but to make a more affordable car that everyone likes and wants.
Travis Axelrod: Very good. Thank you, Lars. The next question is, could you talk about the benefits of Tesla's investment in XAI?
DevOps Ninja: This is not the forum to discuss this topic. If there is a need for discussion, we will address it separately.
Elon Musk: I think it’s clear that we are a public company. Shareholders are welcome to propose any shareholder proposals they want. I have recently encouraged this. Then let the shareholders vote and act according to their wishes.
Lars Moravy: Very good. Thank you very much. The next question is, could you tell us more about the Tesla Design Studio?
DevOps Ninja: Do you want me to answer that?
Travis Axelrod: We usually say that what happens in the studio stays in the studio. Earnings calls are not the place to disclose new products. However, we are working hard to ensure that Tesla and our product line have an exciting future.
Elon Musk: Yes. There are many exciting things happening in the design studio. It is not static. In fact, the fundamental shift over the next few years will be from a pre-autonomy world to a post-autonomy world. I am formulating a new master plan to outline this. The transition from a pre-autonomy world to a post-autonomy world will be painful. But I believe Tesla’s future vision is very exciting and will fundamentally change the world for the better… if we execute the plan effectively, Tesla will become the most valuable company in the world.
Travis Axelrod: Very good. Thank you. The next question is actually a repeat about unsupervised FSD customer vehicles, so we will skip it. After that, do HW3 users have any news on upgrades or retrofits? Will they upgrade to HW4 or future HW5 versions?
DevOps Ninja: Our goal is to first complete the unsupervised features on Hardware 4 (HW4). Once that is done, we will revisit what needs to be done for the Hardware 3 cars. As I mentioned, the focus is on rolling out the unsupervised features first, and then we will go back and see what additional work is needed. Lars Moravy: Great.
Travis Axelrod: The next question is, could you provide an update on Dojo? Will XAI become a customer of Dojo?
Elon Musk: For Dojo 2, we expect it to be running at scale sometime next year. It will be around 100 exaflops equivalent. Then there's AI 5, which is also great. I don't use these words lightly. The AI factory is expected to start production around the end of next year. But there's a lot of potential. Thinking about Dojo 3 and the first chip for AI 6, intuitively, we want to find a convergence point. Essentially, using the same chip, you could say there are two on the car or Optimus, or more on a board, like 512 on a board, if you want high-bandwidth communication between chips. This seems intuitively like the right way to go.
Lars Moravy: Great.
Travis Axelrod: The next set of questions has already been answered. So we will conclude with this question: How will the cancellation of tax credits for solar projects in the 'Big and Beautiful' bill affect your Megapack sales channels?
Lars Moravy: Our sales channels are very diversified in terms of customers and market segments. Therefore, we are not heavily weighted towards Megapack projects paired with solar. As mentioned in our opening remarks, we see energy storage being quickly recognized as a capability to unlock grid efficiency and can be rapidly deployed to assist the grid. Although the recent bill is unfavorable for solar, we believe solar projects will still be built because energy is necessary. Projects are well-developed and ready to be executed. In the short term, there are no alternatives due to the delivery times and prices of gas turbines. We also continue to see growth in the data center segment and standalone energy storage projects providing grid capacity in multiple U.S. markets. Overall, we expect a very strong second half of the year as we increase deployments. Finally, we continue to invest heavily in U.S. manufacturing to mitigate the impact of policies and tariffs, and we expect our first LFP battery manufacturing facility to come online by the end of the year, with our third Gigafactory near Houston starting in 2026.
Travis Axelrod: Great. Thank you, Mike. We will now move on to the analysts' questions. The first question is from Emmanuel Rosner of Wolfe Research. Emmanuel, please unmute yourself.
Emmanuel Rosner: Great. Thank you very much. Can you hear me?
Lars Moravy: Yes.
Emmanuel Rosner: Thank you. So, Elon, can you share some KPIs about the Robotaxi business? How many vehicles are in operation, and how many autonomous miles have been driven? What is the number of safety-critical interventions? I'm just curious about the overall rollout and if you can share some broader goals?
Elon Musk: Of course. Do you want to answer that?
Ashok Elluswamy: We have already driven over 7,000 miles in the Austin area. Since the service is new, we currently have only a few vehicles, but we are trying to expand both the service area and the number of vehicles in Austin and other locations. So far, there have been no significant safety-critical interventions. Sometimes we have our own limitations, such as limiting the speed to 40 miles per hour. If the vehicle wants to drive on higher-speed roads, we can stop, but these are convenience issues rather than safety-critical ones. Therefore, the service has been well-received, and we continue to expand.
Emmanuel Rosner: Great. From a long-term economic perspective, you previously talked about efforts to reduce the cost per mile for Robotaxis, potentially down to 30 or 40 cents per mile over time. Now that your service is live, how should we think about the key milestones to achieve this goal?
Elon Musk: The CyberCab really optimizes for autonomy. This could potentially be under 30 cents or even 25 cents per mile over time. If you design a cost-optimized Robotaxi from scratch, like the CyberCab, we are not trying to make it turn as well as a Model 3, Model S, or even a Model Y. All cars driven by people are very fun. They have amazing acceleration and cornering abilities. But I think we are pretty sure that very few people will want to take a sharp turn in a CyberCab. So, we lower the top speed, which means we can use more efficient tires. We don't need as much acceleration. We don't need as big brakes. We want stopping distance, but we don't expect it to be used heavily. It's a smooth ride experience. If you design for a smooth ride experience, you get a more optimized design point. That's why we seem to be able to achieve this. Especially with Optimus providing services, cleaning the vehicles, and performing maintenance. It will charge itself. The actual cost per mile for the CyberCab will be very low. The actual cost per mile for our existing fleet will be higher, but still very competitive. I guess around 50 cents. I'm just guessing. Tesla's Robotaxi business is operating at a very large scale in a relatively short time. My guess is that it will have a material impact on our finances by the end of next year.
Travis Axelrod: Thank you very much. The next question is from Adam at Morgan Stanley. Adam, please unmute.
Adam Jonas: Great. Hello, everyone. So, Elon, as Tesla enters this next phase of physical AI, autonomous humanoid robots, and Robotaxis—technologies that are world-changing and civilization-altering, with dual uses—do you feel that it is sustainable to take Tesla in this direction while owning only 13% of the company? Or do you still stand by the need for changes given your current lack of control, considering the type of technology you are entering into?
Elon Musk: Yes. This has been my primary concern, as I have mentioned before. I hope to address this issue at the upcoming shareholders' meeting. But, yes, it is a significant matter. I am concerned that I may not have enough control, and after building this army of humanoid robots, I could be easily ousted by activist shareholders. I believe my control over Tesla should be sufficient to ensure it moves in the right direction, but not so much that I cannot be removed if I go mad.
Adam Jonas: Alright, Elon. You won't go mad. We trust you.
Emmanuel Ronal: Well, you can be a little bit crazy. A bit of craziness is okay.
Adam Jonas: Elon, nonetheless, we understand that the Board of Directors from a major US investment bank recently visited the Optimus production facility. I don't know if you want to confirm this. It's pretty cool. When will others get to see Optimus up close like that? Is it too early to host an AI Day in the second half of this year? Because it seems like everyone else in the world is doing it, and the talent war is getting crazy. I know you mentioned that for recruitment purposes, this is very important, and you've already done it. I think people are following your lead, and I wonder if it's too early this year. Thank you, Elon.
Elon Musk: Yes. It's a bit tricky because when we hold AI Day, we find that some suppliers actually examine our slides frame by frame, every word we say, and copy these things. So, you know, I'm saying, for example, is it worth it in terms of recruitment? But competitors will watch very closely and replicate what we do. Even so, we might consider using the shareholders' meeting as a kind of AI Day. We can go into more detail about Optimus, AI, and our chip-related content at the annual shareholders' meeting.
Lars Moravy: Yes. Tesla is severely undervalued in both actual design and AI software. There isn't a chip we would prefer to put in our cars. Even after several years. We believe that AI and the Fire chip will be a profound game-changer. In fact, it is so powerful that we have to weaken it in some ways for use outside the US because it exceeds export restrictions. Unless the export restrictions change, we actually have to weaken our live AI demonstrations. This looks a bit strange. We hope to continuously raise the threshold for export restrictions. Otherwise, it becomes a bit absurd. We will showcase many Optimus robots at the shareholders' meeting. The Optimus lab looks very cool. It looks like a set from Westworld. You can see robots at different stages. Some are in various stages of repair. It looks like a mix of a Tatooine junkyard and Westworld. Very cool. Optimus is walking around the offices in Palo Alto. It roams around like a small thing. We see a complete Optimus serving popcorn at the Tesla cafeteria. We are moving from a world with few robots to one where robots are so common that you won't even look up.
Travis Axelrod: Great. Thank you very much. The next question is from Edison at Deutsche Bank. Edison, please unmute. Please continue to unmute. Okay. While Edison is handling it, we will move to the next question, which will come from Dan Levy at Barclays. Dan, please unmute.
Dan Levy: Great. Thank you. Elon, you mentioned the opportunity to add non-Tesla-owned vehicles to the Robotaxi network. Can you talk about the limiting factors for enabling this and the timeline we should expect for personally owned vehicles to join the Robotaxi network?
Elon Musk: We haven't really given this serious consideration yet. We need to ensure that the vehicles are fully under our control before proceeding. This is a step-by-step process. We don't want to rush things. As I said, we are very cautious about safety. But I think I will be confident to say next year. I'm not sure when next year, but I am confident it will be next year, and people will be able to add or remove their Cars from Tesla's fleet.
DevOps Ninja: One thing to keep in mind is that we will have some standards. Even if you add your car to Uber Technologies or Lyft fleets, they go through a full inspection process to ensure everything is in order.
Elon Musk: Second, Airbnb is fantastic. Yes. Understood. So, we will do something similar. It's very likely. Yes.
DevOps Ninja: As Elon mentioned, we want to be cautious about safety. Small things like tire treads can impact safety. That's why we want to conduct proper verification before allowing other vehicles to join.
Travis Axelrod: Dan, do you have any follow-up questions?
Dan Levy: Yes. Thank you. Can you break down the different costs associated with scaling the Robotaxi business and discuss how you see funding these costs? Is the free cash flow from the Cars business sufficient? If not, what other sources of funding do you think will be used? Will you use the balance sheet directly for financing?
Elon Musk: Once there is a clear cash flow from any product, you can finance it.
DevOps Ninja: In the meantime, we will use our balance sheet, but once we reach a certain level of current revenue, as Elon said, we can enter into an easy Trade to obtain funding.
Travis Axelrod: Very good. Thank you very much. We will now turn to Mark from Goldman Sachs. Mark, please unmute yourself at any time.
Mark Delaney: Yes. Good afternoon. Thank you for taking the questions. With Tesla continuing to offer FSD trials to consumers and the general focus on autonomous driving, could you provide more specific comments on the FSD subscription trends and adoption rates, and help us better understand how large FSD revenue might be currently?
DevOps Ninja: Since we launched FSD version 12 in North America, we have indeed seen a significant improvement in FSD adoption. Another thing we did last year was to lower the price, making the subscription more affordable. Since then, we have seen a 45% increase, which is an encouraging trend. But honestly, we are just beginning to tell the story of the benefits of FSD. We released the vehicle safety report. Even if you don't believe anything else, FSD vehicles are ten times safer than non-FSD vehicles, which should be a motivating factor. Additionally, people do not realize that, even at $99 per month, it's like paying less than $3.33 per day for a personal driver. This is the biggest game-changer so far, and I know we keep talking about it because it's something we live and breathe.
Elon Musk: Yes. I think most people still don't know this. The vast majority of Tesla owners don't even know it exists. It's still about half of Tesla owners who can use it but have never tried it once. They haven't actually gone online with it. Obviously, this is something we want to educate them about. When they come to the service center, we reach out to them and send videos to teach them how to use it. It's a shocking thing. They don't think a car can do these things. You have to show them and make them feel comfortable turning it on. It's simple, but it's like saying you have a cat that can sing and dance, but it looks like an ordinary old cat. Until you see the cat singing, dancing, and talking, you don't realize it's not just a cat. That's what Tesla's FSD is. Our cars are very smart.
DevOps Ninja: As Elon said, we have been giving users some free time to try FSD, but we will start providing more prompts, like, for this particular drive, try FSD. Because seeing is believing. As Elon said, it's like a cat. It looks like an ordinary cat, but this cat can sing and dance. It's the same as we build it.
Travis Axelrod: Very good. The 25% comment refers to a 25% increase in FSD penetration since we launched FSD 12 and B 38 in North America. Very good. Mark, do you have any follow-up questions?
Mark Delaney: Yes. Thank you, Travis. Tesla has historically said that as long as free cash flow remains positive and monetization through products like FSD is possible, pricing will be used as one of the tools to drive vehicle growth. I'm curious, given the situation with FSD and the fact that the IRA electric vehicle tax credit will start to phase out in the US in the fourth quarter, how do you view pricing as a potential tool to drive sales growth? Given the monetization potential, should we expect more significant price reductions? Or do you think price reductions will be more limited compared to cost reductions, given the current state of free cash flow?
Elon Musk: Well, we are in a strange transition period where we will lose a lot of incentives in the US. Actually, there are incentives in many other parts of the world as well. But in the US, we will lose them. On the other hand, autonomy is the most advanced and available in terms of regulation. This means we might have a few difficult quarters? Yes. We might have a few difficult quarters. I'm not saying we will, but we might. In Q4, Q1, and possibly Q2. But once you achieve full autonomy in the second half of next year, only by the end of next year, I would be surprised if the economic situation of Tesla is not very attractive.
Travis Axelrod: Great. The next question will come from Will of Truist. Will, please unmute when you're ready.
Will Stein: Great. Thank you very much for taking my question. First, I would like to ask you to elaborate on the low-cost model that you mentioned is starting production in the first half of the year but will ramp up later. I remember at the last Analyst Day, you talked about some aspects of this, such as reducing silicon carbide by two-thirds or three-quarters and not using rare earths in the motors, which could lead to other cost reductions. You also mentioned an unboxed architecture, which I believe you said would not be part of this mid-term approach. Could you provide an update on what we should expect this to actually look like?
DevOps Ninja: We will not reveal the appearance. You will see the Model Y.
Will Stein: Yes. Fundamentally, the biggest barrier remains that people do not have enough money to buy a car. The desire to purchase a car is very high; it's just that people do not have enough money to afford it. Literally, that's the issue. It's not a lack of desire, but a lack of ability. The more affordable we can make the car, the better. I think when people can release their cars into a fleet and earn money from them, it will significantly increase affordability. As I said, I am certain that this will happen in the U.S., at least in the U.S., next year. In the U.S., we are doing this within the legal limits, with the appropriate disclaimers. This will significantly increase affordability. Just like if you have an Airbnb and you rent out your house or a room or a guesthouse when you're not home, the affordability of your house increases significantly.
Will Stein: Okay. Moving on to another topic. We see a lot of amazing developments in XAI, like Grok, and obviously, Tesla is also doing a lot of work in artificial intelligence. Elon, how do you manage the efforts, recruitment, talent, and capital allocation between these two areas, which seem to have a high potential for competition?
Elon Musk: Well, they are doing different things here. XAI is working on terabyte-scale models and multi-terabyte-scale models. Tesla, on the other hand, is working on models that are a hundred times smaller. One is real-world AI, and the other is a kind of artificial superintelligence. The origin of XAI is that some engineers were unwilling to join Tesla's AI team because they wanted to work on artificial superintelligence. They wouldn't join Tesla. I mean, maybe they would join a new company. I think this is a very important issue, but not everyone agrees with me. Rather than having them join OpenAI or Google or another company, they formed a company like this. People can make a choice. Do they want to work on superintelligence in data centers, or do they want to work on real-world AI? Both are very attractive, but some people want to do one, and some people want to do the other.
Travis Axelrod: Great. Unfortunately, that's all the time we have today. Thank you very much for your questions, and we'll see you next quarter.
(This article was assisted by artificial intelligence)
