Whip Bulls reported that on April 24, Tesla's U.S. stock market released its financial report for the first quarter of 2024 after hours. Revenue was $21.3 billion, compared to the consensus of $22.34 billion, compared to $23.329 billion in the same period last year. Adjusted earnings per share were $0.45 versus the consensus of $0.59. Net profit was $1.129 billion, compared to the consensus of $1.53 billion. Gross margin was 17.4% versus the consensus of 16.5%.
At the earnings conference, Tesla held a conference call, and the following is the transcript.
Note: The content is an AI translation and has not been proofread for reference only.
Martin Vicha
Q1 2024 Q&A webcast. I'm Martin Viecha, VP of Investor Relations, and I'm joined today by Elon Musk, Vaibhav Taneja, and a few other executives. Our Q1 results were announced in our published update around 3:00 p.m. CT, with the same link as this webcast.
During this conference call, we will discuss our business prospects and make forward-looking statements. These comments are based on our projections and expectations for today. Actual events and results could differ materially due to a number of risks and uncertainties, including those identified in our most recent filings with the SEC. During the Q&A portion of today's call, please limit yourself to one question and one follow-up.
Please use the "Raise Hand" button to join the queue for questions. But before we get to Q&A, Elon made some opening remarks. Elon?
Elon Musk – CEO and Product Architect
Thanks, Martin. To recap, in the first quarter, we overcame some unforeseen challenges and completed an upgrade to the updated Model 3 in Fremont. As we can all see, global EV adoption is under pressure, with many other contract makers reducing EV production in favor of plug-in hybrids. We don't think it's the right strategy, and EVs will eventually dominate the market.
Despite these challenges, the Tesla team has performed well in a tough environment, especially with Megapack's energy storage deployment reaching an all-time high in the first quarter, resulting in record profitability in the energy business. And that number is likely to continue to increase in the coming quarters and years. It will increase. We actually knew it was going to happen, and it was going to be much faster than we expected in the automotive industry.
We also continued to expand our AI training capabilities in Q1, more than doubling the amount of training compute consecutively. In terms of new product roadmaps, there's been a lot of talk over the past few weeks about our upcoming family of vehicles in the future. We have updated our lineup of future vehicles to accelerate the launch of new models, with production previously mentioned starting in the second half of 2025. As a result, we expect it to be more like early 2025, or even later this year. These new vehicles, including more affordable models, will use aspects of the next-generation platform as well as aspects of our current platform, and we will be able to produce them on the same production lines as the current vehicle range.
Therefore, it does not depend on any new factory or large-scale new production line, it will be produced more efficiently on our existing production line. We believe that when fully realized, our production capacity will exceed 3 million vehicles. As for FSD Version 12, which is pure AI autonomous driving, if you haven't experienced it yet, I highly recommend you to try it, it has far-reaching implications. And the rate of improvement is fast.
We've now enabled all the features of Hardware 3 in North America for all cars equipped with cameras and inference computers. So, I think, it's rolled out to about 1.8 million cars, and so far, we're seeing about half of the people using it, and that percentage is increasing every week. As a result, we have now driven more than 300 billion miles with FSD V12 since the introduction of Full Autonomous Driving (Supervised Full Self-Driving). It is clear that a vision-based end-to-end neural network approach is the right solution for scalable autonomy.
This is indeed the way humans drive. Our entire road network is designed for biological neural networks and eyes. Therefore, cameras and digital neural networks are the natural solution for our current road system. To make it easier to use, we've reduced the subscription price to $99 per month, so it's easy to try out.
As we announced, we will be showcasing our dedicated robo-taxi, or Cybercab, in August. Yes. With regard to AI computing, over the past few months, we have been actively working to expand Tesla's core AI infrastructure. There was a period of time when our training progress was limited.
At this point, we are no longer limited by training, so we are making rapid progress. We've installed and debugged, which means they're actually running 35,000 H100 computers or GPUs. The word GPU is wrong. They needed a new word.
I always cringe when I say GPU because it's not. GPU bracket – G stands for graphics, it doesn't do graphics. But you know, there's about 35,000 H100S active, and we expect that number to probably reach around 85,000 by the end of the year, and training, just for training. We make sure that our training is as efficient as possible.
This is not only related to the number of H100s, but also to the efficiency with which they are used. So, all in all, we're very excited about our roadmap for autonomy. I think this should be obvious to anyone using version 12, and it's only a matter of time before we surpass humans in a very short period of time in terms of reliability. We are indeed moving towards the future of electric vehicles and autonomous driving.
I go back to what I said a few years ago, and in the future, non-self-driving gasoline cars will be like riding a horse and using a flip phone. In hindsight, this will become very obvious. We will continue to make the necessary investments to drive Tesla's future growth and profits, and I would like to thank the Tesla team for their incredible execution during this period and look forward to everything we have planned for the future." Next.
Martin Vicha
Thank you very much, Vaibhav also gave your opinion.
Vaibhav Taneja——首席财务官
It's important to acknowledge what Elon said. From our automotive business perspective, we did see a sequential decline in revenue, primarily due to seasonality, uncertainty in the macroeconomic environment, and other reasons that Elon mentioned earlier. Excluding the impact of the Cybertruck, the automotive margin fell from 18.9% to 18.5%. The impact of pricing actions was largely offset by lower unit costs and revenue recognition for autonomous driving features in certain vehicles in the United States
This feature was not available before. In addition, while we did experience higher costs due to the Model 3's ramp-up in Fremont and supply disruptions in Berlin, those costs were largely offset by cost-cutting initiatives. In fact, if we exclude the ramp costs of the Cybertruck and Fremont Model 3, there is a slight improvement in automotive margin revenue. Currently, the Model Y in Austin and Berlin is already very close to Fremont in terms of standardized cost per vehicle. We were able to reduce costs without sacrificing quality, thanks to the amazing effort put in by the team in executing Tesla's relentless pursuit of efficiency across the business.
We're also seeing an increase in interest in credit as other OEMs withdraw their investments in EVs, which means a steady stream of revenue for us. Clearly, seeing others retire from EVs isn't the future we want. We prefer that the entire industry go all out. On the demand side, we've taken a variety of initiatives, including lowering the price of purchase and subscription options and introducing an attractive Model 3 rental special rate in the United States
$299 per month and offers attractive financing options in some markets. We believe that our promotional activities, combined with attractive financing, will go a long way in expanding our reach and driving demand for our products. Our energy business continued to make meaningful progress with a record margin of 24.6%. We expect energy storage deployments in 2024 to grow by at least 75% compared to 2023.
As a result, this business will begin to make a significant contribution to our overall profitability. Please note that there are some fluctuations in our storage deployments due to various factors beyond our control, so deployments may fluctuate on a quarterly basis. In terms of operating expenses, we have seen sequential growth from AI initiatives, continued investment in future projects, marketing, and other activities. In the first quarter, we had negative free cash flow of $2.5 billion.
The main drivers are an increase in inventory due to a mismatch between build and delivery, as mentioned earlier, and an increase in our capital expenditure on various initiatives, including AI calculations. We expect inventory growth to reverse in Q2 and free cash flow to return to positive again. As we prepare for the next phase of growth in the company, we must make the difficult but necessary decision to reduce the headcount by more than 10%. On an annual run rate, savings are expected to be well over $1 billion.
We also have a strong focus on capex efficiency and utilizing our installed capacity in a more efficient way. The savings from these initiatives, including cost reductions, will help improve our overall profitability and ultimately enable us to scale our investments in AI. In conclusion, the future is extremely bright, and the journey to achieve this goal while being challenging will be very rewarding. Once again, I would like to thank the entire Tesla team for the excellent results achieved.
We can open it up for Q&A.
Martin Vicha
Good. Let's start with an investor Q&A. The first question is, what is the current status of the 4680 and what is the current output? Lars?
Lars Moravy——车辆工程副总裁
Of course. The 4680 production is up about 18 percent, up 20 percent from the fourth quarter, surpassing the Cybertruck's [inaudible], as we posted on X, about 7 GWh per year. We anticipate battery production ahead of the Cybertruck's capacity growth throughout the second quarter, and we added the third of our four production lines in Phase 1 while maintaining several weeks of battery inventory to ensure we are ahead of the production line. As we are ramping up production, the cost of goods sold continues to decline rapidly week after week, driven by increased throughput and increased throughput across the line.
Therefore, our goal, and what we hope to do, is to beat the supplier cost of nickel-based batteries by the end of this year.
Martin Vicha
Thank you. The second question is about Optimus Prime. So, what is the current state of Optimus Prime? Are they currently performing any factory tasks, and when is mass production expected to begin?
Elon Musk – CEO and Product Architect
We were able to do simple factory tasks, or at least, I should say, factory tasks in the lab. We do think we're going to have a limited production of Optimus in a natural plant, with a useful task done by the end of the year. And then I think maybe we can sell it before the end of next year. These are just speculations.
As I said before, I think Optimus Prime will be more valuable than everything else combined. Because if you have a sentient humanoid robot that can navigate reality and perform tasks as required, there aren't any meaningful limits to the scale of the economy. So, that's what's going to happen. I think Tesla is in the best position among all humanoid robot manufacturers to be able to achieve mass production through effective reasoning about the robot itself.
I mean, and that's perhaps a point worth emphasizing, that Tesla's AI reasoning efficiency is far superior to that of any other company. No company can match Tesla's reasoning efficiency. We had to do this because we were limited by the inference hardware in the car. We have no choice.
But it will be beneficial in many ways.
Martin Vicha
The third question is, what is the current assessment of the regulatory approval pathway for unsupervised FSD in the United States, and how should we consider the appropriate safety thresholds compared to human drivers?
Lars Moravy——车辆工程副总裁
I can get started. Several states have passed self-driving car laws. These states are paving the way for operations, and data from such operations is guiding the wider adoption of driverless cars. I think Ashok can talk a little bit about our approach to security, but we expect that these states and the work being done and the data that we provide will pave the way for broad regulatory approvals in the United States
At least in other countries as well.
Elon Musk – CEO and Product Architect
Yes. Self-driving car companies carving out a path through the regulatory jungle can actually be very helpful. So, that's actually quite helpful. They have apparently been operating in San Francisco for a while.
I think they got the approval of Vale City. As a result, these approvals are happening rapidly. I think if you have large-scale, statistically significant data that conclusively shows that self-driving cars have half the accident rate than human-driven cars, I think that's hard to ignore because at this point, stopping autonomy means killing people. So, I don't really think there's going to be a major regulatory hurdle, there's hard data that autonomous vehicles are safer than human-driven cars.
In my opinion, it's a lot like an elevator. The elevator used to be operated by a person with a relay switch. But sometimes, the person gets tired or drunk, or just makes a mistake and sends someone in half between the floors. So, we just got in the elevator, pressed the button, and we didn't think about anything.
In fact, it's a bit weird if someone were standing there with a relay switch. That's how cars work. All you need to do is use your phone to summon a car. You go in.
It takes you to your destination. You get out.
Vaibhav Taneja——首席财务官
You don't even have to think about it.
Elon Musk – CEO and Product Architect
You don't have to think about it, it's like an elevator. It will take you to your floor. That's it. Don't think about how elevators work or anything like that.
I should clarify that Tesla will operate the fleet. So, you can think of it as a combination of Airbnb and Uber, like Tesla, which means that Tesla will own a certain number of cars and operate in a fleet. There will be a certain number of cars, and then there will be a bunch of cars owned by the end user. End users can add or subtract their cars from the fleet at any time, and they can decide whether to let friends and family use only the car or just buy a car that a five-star user or anyone can have at any time and the car comes back to them, just like Airbnb is completely theirs.
You can rent or not rent your room at any time. So, as our fleet grows, the number of cars will reach 7 million, eventually 9 million, and eventually tens of millions worldwide. With a constant feedback loop, every time something goes wrong, it's added to the training data, and you get this training flywheel, like Google search has that kind of flywheel, it's hard to compete with Google because people are constantly searching and clicking, and Google is getting a feedback loop. So, the same goes for Tesla.
But the scale may be difficult to understand, but it will eventually reach tens of millions. I think the AWS element also has some potential, if we have very strong inference power, because we have hardware 3 in the car, but now all the cars are made with hardware 4. The Hardware 5 is beautiful and well designed, and should be used in cars, hopefully by the end of next year. When the car is not moving, it is possible to run distributed inference.
So, it's kind of like AWS, but it's distributed inference. It's like it takes a lot of computers to train an AI model, but the amount of computation required to run it is a lot less. So, if you can imagine a future, maybe there will be a fleet of 100 million Teslas, and on average, they might have a kilowatt of inference computing power. This is equivalent to 100 gigawatts of inference computation distributed around the world.
Integrating 100 GW of AI computation is quite difficult. Even in the future of autonomous driving, cars may no longer be used for 10 hours per week, but for 50 hours per week. This still gives the car inference computer more than 100 hours a week to do other things. And it seems to be wasted if you don't use it.
Martin Vicha
Ashok, do you want to talk about process and safety?
Ashok Elluswamy——自动驾驶软件总监
Yes. We've spent years validating security every single week, we've trained hundreds of neural networks that can produce different driving car trajectories and replay them through the millions of clips that we've collected from users and ourselves. Our simulation system also tries to create and test it in a way that is close to fashion. Some of them are validated, and we provide them to our QA network.
We have hundreds of these locations in different cities like San Francisco, Los Angeles, Austin, New York, and more. They're also driving it and collecting real-world mileage, and we've got estimates of key events, if they have a net improvement from the previous week's build. Once we're confident that the build is a net improvement, then we start delivering to early adopters, like the initial 2,000 employees, who want to build it, and they'll provide feedback like if there's an improvement or if they've noticed something new that we've not captured in our own QA process. Only after all of this has been verified will we go to external customers.
Even when we go outside, we have real-time dashboards to monitor every critical event happening in the fleet, sorted by its importance. That's why we're constantly looking for improvements in build quality and security. Then, any failures like Elon mentioned, we get the data back, add it to the training, and improve the model in the next cycle. So we have this continuous feedback loop of issue, fix, evaluation, and then rinse and repeat.
Especially with the new V12 architecture, all of this is automatically improved without much engineering intervention, because engineers don't have to be creative and don't have to like the way they write algorithms. It is mostly self-learning based on data. So, you'll see, okay, every failure or something like that is the way a person chooses, is how you drive an intersection or something like that, and they're going to get the data back. We add it to the neural network, and it automatically learns from the data that has been trained, as opposed to some engineer saying, oh, here, you have to spin the wheel so much or something like that.
There are no hard reasoning conditions. Everything is a neural network. Very soft. It is probabilistic, so it will adopt a probability distribution based on the new data obtained.
Elon Musk – CEO and Product Architect
Yes. We do have some insight into what will happen in three or four months, because we have advanced models that are much more capable than the ones in the car, but they have some problems that we need to address to make them fixed. As a result, the car's capabilities will change significantly, but it will also have some quirks that need to be addressed before it can be released. As Ashok said, we have to be very careful with what we post to our fleets or customers in general.
So if we look at 12.4 and 12.5, they can actually even be said to be version 13, version 14 because it's very close to the overall retraining of the neural network, which is very different in each case. So we have a good idea of the condition of the model, how the car will perform after three or four months.
Ashok Elluswamy——自动驾驶软件总监
Yes. In terms of scale loss, people in the community usually talk about model scale loss, where they massively increase the size of the model and then get a performance boost accordingly, but we also figure out the scale loss and other access in addition to model-side scaling, also data scaling. You can increase the amount of data that you use to train the neural network, which will also bring similar benefits, you can also scale by training the computation, you can train it for a longer period of time, and one more GPU or more dojo nodes, which can also provide better performance, and you can also do architectural scaling, which is to use a better architecture to achieve the same amount of computation, resulting in better results. So, combined with model size scaling, data scaling, training compute scaling, and schema scaling, we can basically infer that, okay, by continuing scaling based on this ratio, we can refine the big performance in the future.
Obviously, it takes time to do experiments, because it takes weeks to train, it takes weeks to collect tens of millions of video clips and process all of that, but you can estimate trends based on future progress based on what we've seen in the past, and based on past data, they're usually correct.
Martin Vicha
Good. Thank you so much. I'm going to talk about the next question, which is can we get the official timeline for the $25,000 vehicle?
Lars Moravy——车辆工程副总裁
I think Elon mentioned that in his opening remarks. But as you mentioned, we're updating our future vehicle lineup to accelerate the rollout of low-cost vehicles in a more capex-efficient way. Our mission is to provide our customers with the most affordable cars as quickly as possible. We've built these new cars on existing lines and open capacity, which is a major shift to leverage all of our capacity at marginal capex before we spend high capex.
Elon Musk – CEO and Product Architect
Yes. We'll talk about this further on August 8th. But really, the way to think about Tesla is almost entirely to solve the problem of self-driving and be able to enable self-driving for a huge fleet of vehicles. I think when the day comes when you can do unsupervised and fully autonomous driving, it's probably going to be the biggest asset value appreciation in history.
Lars Moravy——车辆工程副总裁
Five million cars.
Elon Musk – CEO and Product Architect
Yes. In about a year it will reach 7 million cars, then 10 million, and finally, we are talking about tens of millions of cars - not finally, like before the end of the year. Yes. But I think that in this decade, the number of cars will reach tens of millions.
Martin Vicha
Thank you. The next question is, how is the Cybertruck ramp going?
Lars Moravy——车辆工程副总裁
I can take that too. A few weeks ago, Cybertruck was selling 1K per week. This happened within the first four to five months since we implemented the SOP at the end of last year. Of course, mass production is the most important thing.
That's what drives costs, so our costs are coming down, but growth is still being challenged by a lot of new technologies, some vendor constraints, etc., and it's going to continue to grow this year, focusing only on cost efficiency and quality.
Martin Vicha
Good. Thank you. The next question is, have any traditional automakers contacted Tesla to discuss a possible future authorization of FSD?
Elon Musk – CEO and Product Architect
We are in dialogue with a major automotive manufacturer about FSD licensing.
Martin Vicha
Thank you. The next question is about robo-taxis. Elon has already spoken about this. So, we'll have to wait until August.
The following questions are about the next generation of cars. We've already talked about this. So, let's move on to the semi-finals. What is the timeline for the half-scale scale-up?
Lars Moravy——车辆工程副总裁
As a result, we are completing the engineering of semi-finished products to leverage the lessons we've learned from our fleet, our pilot fleet, and our PepsiCo fleet to achieve super cost-effective high-volume production, which we will be slightly expanding this year. At the same time, as we demonstrated at the shareholder meeting, we have already started to build a factory in Reno. Our first vehicles are scheduled to be launched by the end of 2021, with external customers starting in 2026.
Martin Vicha
Good. There are a couple more questions. So, our favorite question is, can we make the FSD transfer permanent until FSD fully achieves level 5 autonomy?
Lars Moravy——车辆工程副总裁
Yes.
Martin Vicha
Good. Next question. How was the production growth of the Lathrop and what do you think was the running rate of the Megapack at the end of the year?
Unknown speaker
Yes. Rathrop is proceeding as planned. We have a second GA production line that allows us to increase our exit rate from 20 GWh per year at the start of the year to 40 GWh per year at the end of the year. The line has been put into service.
There really is nothing restrictive on the ramps. Given the long sales cycle for these large projects, we can usually have order visibility 12 to 24 months before the ship date. As a result, we were able to plan our build several quarters in advance. As a result, this allows us to scale up our factory to accommodate the growth of our business and orders.
Finally, we would like to thank our customers around the world for their trust in Tesla as a partner in these incredible projects.
Martin Vicha
Let's answer the analyst's question. The first question comes from Bernstein's Tony Saconaghi. Tony, please go ahead and unmute.
(Tony Sacconaghi) —AllianceBernaghi—that's what we'd like to see.
Thank you for your question. I was just wondering if you could elaborate on the type of new car you talked about today. Given that the existing models will run on the same line, are these similar to tweaks to existing models?or are these new models?How should we think about them in the context of the Model 3 Highland update, and what will these models look like relative to it?Due to the tight schedule, the Model 3 Highland will require a lot of work and equipment reconstruction. Maybe you can help put this all in context.
I have a follow-up, please.
Elon Musk – CEO and Product Architect
I think we've already said that we're in that regard. So, what's your next step?
(Tony Sacconaghi) —AllianceBernaghi—that's what we'd like to see.
Elon, it's more personal for you because you're leading a lot of important companies right now. Maybe you can talk about where your heart is in relation to your interests. Do you expect to reduce your involvement in Tesla in the next three years?
Elon Musk – CEO and Product Architect
Well, because it takes up most of my time at work, and I work almost every day of the week, I rarely take a Sunday afternoon off. I'm going to make sure Tesla is very prosperous. It is indeed thriving, and it will be so in the future.
Martin Vicha
Good. Thank you. Let us give the floor to Adam Jonas of Morgan Stanley. Adam, please continue.
Go ahead and unmute.
Adam Jonas – Morgan Stanley – Analyst
Good. Great. Hey, Elon. As a result, you and your team expect sales growth in 2024 to be significantly lower than in 2023.
But how confident is your team in a 0% plus increase? In other words, does the statement leave room for year-over-year sales declines?
Elon Musk – CEO and Product Architect
No, I think this year we will have higher sales than last year.
Adam Jonas – Morgan Stanley – Analyst
Good. My follow-up, Elon, is about future products. If you've identified execution, let's say you've identified execution on the next generation of cheaper vehicles, more radical gigabit castings, I don't want to say one, but it's closer to, say, one-piece construction package, unboxed, 300-mile range, $25,000 price point, all of these features are unique to you. How long will it take for your best Chinese competitors to replicate the cheaper, better cars you can offer in a few years?How long will it take for your best Chinese competitors to replicate this? Thank you.
Elon Musk – CEO and Product Architect
I mean, I don't know what our competitors can do unless we're doing a relatively better job than them, because if you look at the decline in sales of our competitors in China compared to the decline in our sales, we're down less than they. So, we're doing a great job. But I think Cathy Wood said it best. In fact, we should be seen as an artificial intelligence or robotics company.
If you rate Tesla the way you would a car company, you have to — fundamentally, it's just a false framework, and if you ask the wrong question, the right answer is impossible. So, I mean, if somebody doesn't believe that Tesla is going to solve the problem of self-driving, I don't think they should be an investor in that company. It's like that, but we'll do it, we'll do it, and then your car will go from 10 hours a week, like an hour and a half a day, to maybe 50 percent, but at the same cost.
Vaibhav Taneja——首席财务官
I think that's the key thing to keep in mind, right, especially if you look at FSD Supervised, if you're not a believer in autonomy, that should give you a look back at this coming. In fact, the situation is getting better day by day.
Elon Musk – CEO and Product Architect
Yes. If you haven't tried FSD 12.3 yet, like I said, 12.4 would be significantly better, and 12.5 would be even better. We have visibility into these things. Then you really don't understand what's going on.
It's impossible.
Vaibhav Taneja——首席财务官
Yes. That's why we can't just think of it as a car company, because a car company has just one car. But here, we're not just a car company, because cars can be autonomous. Like I said, it's happening.
Ashok Elluswamy——自动驾驶软件总监
Yes. Other than that, the Testa AI community is growing like it's growing fast.
Elon Musk – CEO and Product Architect
Yes. I mean, we're putting real cars into cars. So, we're like telling us about the future wagon that you're building. I thought, well, actually, it doesn't need a horse, that's the point.
That's the point.
Martin Vicha
Good. Thank you. The next question comes from Piper Sandler's Alex Potter. Alex, go ahead and unmute.
Alex Porter – Piper Sandler – Analyst
Great. Thank you. Yes, so I totally agree. This paper depends entirely on artificial intelligence, the future of artificial intelligence, fully autonomous driving neural network training, all of that.
In this case, Elon, you talked about your desire to gain 25% voting control of the company. I totally understand why this is the case. So, I don't have to ask this. I ask you if you've come up with any mechanisms to ensure you get this level of voting control.
Because if you don't, then the core part of the paper may be at risk. So, do you have any additional comments on this topic?
Elon Musk – CEO and Product Architect
Well, I think no matter how Tesla is - even if I get kidnapped by aliens tomorrow, Tesla will solve the problem of autopiloting, maybe a little slower, but it will at least solve the problem of autopilot of the vehicle. I don't know if we're going to win in terms of Optimus Prime or future products, but even if I disappear in the automotive space, Tesla has enough momentum to solve the problem of autonomous driving. Yes, there is a lot more we can do in the future. For Optimus Prime I would be more silent, if we had a super-sentient humanoid robot that could follow you indoors and you could escape, we were talking about a Terminator-level risk.
Yes, I would feel uncomfortable. If there is no meaningful impact on how it is deployed. Shareholders have the opportunity to approve or re-approve this competition because I can't say that. It's a fact.
They have a chance. Yes, we'll see. If the company generates a lot of positive cash flow, we can obviously buy back shares.
Alex Porter – Piper Sandler – Analyst
Good. It's actually all very useful background. Thank you. Maybe there's one last question, which I'll pass on.
OpEx is decreasing, thanks for quantifying the impact there. I am also interested in a potentially more qualitative discussion of the impact of these layoffs. What types of activities might you sacrifice as a result of parting ways with these people?
Vaibhav Taneja——首席财务官
So, you know, like we said, we've laid off employees across the board. As the company continues to grow, there will be some level of layoffs. There will be some duplication of work in some areas. So, you need to go back and see where all these pockets are and throw them away.
So, we're basically going through this exercise, and we're like, hey, how do we adapt this company to the next phase of growth? The way to think about this is any tree that grows. It needs to be trimmed. This is the pruning exercise we go through. Ultimately, we will be stronger and more resilient to the future, because the future is indeed bright.
As I said in my opening remarks, we just need to get through this period and get to the goal.
Elon Musk – CEO and Product Architect
Yes. As far as I know, we are not going to give up anything important. From 2019 to now, we have experienced a long period of prosperity. So, if a company makes 5% mistakes in its organization every year, it adds up to inefficiencies.
We've made a few corrections along the way. But now is the time to restructure the company for the next phase of growth, and you do need to restructure it, just like humans, when we start with a cell and a kind of zygote, a kind of blastocyst, you start to grow arms and legs, and then briefly out. and you have a tail.
Alex Porter – Piper Sandler – Analyst
But you flicked your tail off.
Elon Musk – CEO and Product Architect
Hope you get your tail off. And then you're a baby, and basically, you have to be an organism – a company is kind of like a biological growth. If you don't restructure it for different stages of growth, it will fail. If you have 10 units, 100 units, 1 million units, 1 billion units, 1 trillion units, then you can't have the same organizational structure.
Humans have about 35 trillion cells and don't feel like a human being. But you're basically a walking colony of cells, about 35 trillion, depending on your weight, and the number of bacteria is about three times that number. So, in any case, you have to restructure the company to achieve a new phase of growth, otherwise you won't be able to achieve this growth.
Martin Vicha
Thank you. Let's give the floor to Mark Delaney of Goldman Sachs. Mark, please go ahead and unmute.
Mark Delaney – Goldman Sachs – Analyst
Yes. Good afternoon. Thank you very much for your questions. The company had previously described potential FSD licensing discussions at an early stage, but some OEMs weren't really convinced.
Can you elaborate on the extent to which the licensing opportunity that you mentioned today is progressing, and what do you think Tesla needs to achieve with this technology in terms of product milestones in order to successfully reach a licensing agreement?
Elon Musk – CEO and Product Architect
Well, I think we just need — just need to be obvious that our approach is correct. I think so. I think we have 12.3 now, and if you just let the car drive you around, it's clear that our solution can do autonomous driving with a relatively low-cost inference computer and a standard camera. No lidar, no radar, no ultrasound.
Nothing.
Vaibhav Taneja——首席财务官
Automotive manufacturers don't need to do heavy integration work.
Elon Musk – CEO and Product Architect
Yes. So it's really just a case of having them use the same camera and reasoning computer and license our software. Once it's clear that if you don't have this in your car, no one wants your car anymore. It's a smart car.
In fact, I remember that at that time Nokia was the king of the market, and yes, the phone was crushed. Then I saw that they launched a smartphone that was basically a brick with limited features. Then there are iPhones and Androids, and people still don't understand that all phones are like this. There will be no flip phones.
There will be a niche product or home phone. Yes. Not even exactly. When was the last time you read an entire book? It's like an idea.
Yes.
Vaibhav Taneja——首席财务官
In hotels, sometimes in hotels.
Elon Musk – CEO and Product Architect
Yes, hotels have it. People don't understand that all cars need to be smart cars, otherwise you won't sell them, or the car won't – no one will buy it. Once that becomes apparent, I don't think licensing is optional anymore.
Lars Moravy——车辆工程副总裁
It became a method of survival.
Elon Musk – CEO and Product Architect
Yes, absolutely. Get a license, otherwise no one will buy your car.
Vaibhav Taneja——首席财务官
I mean, the other thing that I would add is that in our conversations with some of these OEMs, I just want to point out that they spend a lot of time in the product lifecycle. They're talking about a few years before putting it in their product. We may be able to reach a licensing agreement sooner than that, but it will take a while. So, that's the biggest difference between us and them.
Elon Musk – CEO and Product Architect
Yes. I mean, actually, the agreement signed now will lead to the possibility of three years of its use on the car.
Vaibhav Taneja——首席财务官
That's early.
Elon Musk – CEO and Product Architect
It's basically like lightning.
Vaibhav Taneja——首席财务官
It's an eager OEM.
Elon Musk – CEO and Product Architect
Yes. So I wouldn't be surprised if we did sign the agreement. I think there's a good chance that we'll sign one agreement this year, maybe more than one. But yes, it can take up to three years to integrate with the car.
Even if all you need is a camera and our reasoning computer. So, just talking about a huge design change.
Vaibhav Taneja——首席财务官
Yes. Again, to clarify, this is not something we have to do. It's something they have to do, and it takes time. Right, Mark?
Mark Delaney – Goldman Sachs – Analyst
Yes, very helpful. My next step is to better understand Tesla's pricing approach in the future. Previously, the company had said that price reductions are driving incremental demand as car prices come down, especially for vehicles that are able to earn IRA credits and some of the rental offers offered by Tesla. Do you still think it makes sense to make incremental price reductions to existing products from now on, and can companies make sense to reduce prices from now on and maintain positive free cash flow per year on top of their current product mix?
Elon Musk – CEO and Product Architect
Yes. I think we can achieve meaningful positive free cash flow.
Lars Moravy——车辆工程副总裁
I think Vaibhav said in his opening remarks that just like we're trying to reduce costs, we're basically offsetting the price reductions, like we're trying to get them back to the customer.
Elon Musk – CEO and Product Architect
Yes. I mean, at the end of the day, like any company, if you're selling a quality product at a great price – if you're offering a quality product at a great price, then the sales are going to be really good. This is true in any region. So, over time, we really need to constantly make sure that our products are affordable.
Moreover, the price is affordable for people. So, it's not – you have to address value for money and basic affordability. Basic affordability issues are sometimes overlooked. If someone earns hundreds of thousands of dollars a year, they don't think about cars in terms of basic affordability.
But the vast majority of people live on wages. So, if the monthly lease refinancing cost is $10, either way, it actually makes a difference. The important thing is to keep people affordable and keep lowering prices -
Lars Moravy——车辆工程副总裁
More accessible.
Elon Musk – CEO and Product Architect
Yes, exactly. Make the price more affordable, value for money, and improve over time.
Lars Moravy——车辆工程副总裁
People can also kick their butts if they want to buy it.
Elon Musk – CEO and Product Architect
Yes. It's going to be a great product and it's also affordable. The bar for quality products is constantly improving. So, you can't just be set in stone.
You have to keep improving the car, raising the price, and at the same time raising the cost of production, and that's what we're doing.
Vaibhav Taneja——首席财务官
Yes. In fact, like I said in my opening remarks, just like the revamped Model 3, the updated Model 3 is a really great car. I don't think people even fully understand the engineering effort that went into it, and Lars and the team have actually released videos to explain how different the car is. I mean, it doesn't look and feel the same.
Not only does it look and feel different, we've added a lot of value to it, and you can lease it for a low price of $299 per month.
Lars Moravy——车辆工程副总裁
Airless.
Vaibhav Taneja——首席财务官
Yes.
Martin Vicha
Good. The next question is from George from Canaccord. George, please go ahead and unmute.
George Gianarikas —— Canaccord Genuity——分析师
Hello. Thank you for answering my question. First of all, can you help us understand some of the timing of the rollout of FSD in other regions, including clarifying your recent comments on China?
Elon Musk – CEO and Product Architect
You mean new markets? yes, we are — there are a lot of markets where we don't currently sell cars that we should sell. We'll see an acceleration of this.
George Gianarikas —— Canaccord Genuity——分析师
New markets for FSD?
Elon Musk – CEO and Product Architect
Yes. So think of end-to-end neural network-based autonomy, just like humans, which works well in virtually any market without modification. So we plan to release it as a regulated autonomous system, subject to regulatory approval, to any market where we can get regulatory approval, and we think that includes China. So, yes, just like humans, you can rent a car abroad, and you can drive well.
Obviously, if you live in that country, you will drive better. So, we will make the car drive better in these other countries with country-specific training. But it can drive well almost anywhere.
Vaibhav Taneja——首席财务官
The driving basics are basically the same everywhere. Just like a car is a car, a traffic light is a road, and a road is a road.
Elon Musk – CEO and Product Architect
It understands that it should not bump into things, regardless of the rules of the road.
Vaibhav Taneja——首席财务官
Exactly. There are some road rules that you need to follow. In China, you should not cross a solid line to change lanes. I think in the United States, that's a suggestion.
In China, if you do this, you will be punished heavily. We have to do more, but mostly on a smaller scale. It's not like a whole change of the stack or anything like that.
Martin Vicha
Hey George, do you have a follow-up?
George Gianarikas —— Canaccord Genuity——分析师
Yes. So my follow-up has to do with Q1 deliveries, and I'm curious if you think the supply constraints that you mentioned throughout the launch were impacting the results, maybe you can help us quantify that? is that why you're confident about volume growth in 2024?
Vaibhav Taneja——首席财务官
Yes. I think we did talk about that in our opening remarks. A lot of different things happened in the first quarter. Seasonality is a big factor, from the ongoing pressures of the macroeconomic environment.
Our factories were attacked. We've had a Red Sea attack, we're lifting the Model 3, we're boosting the Cybertruck. All of these things are happening. I mean, it almost feels like the culmination of all this activity in a limited period of time.
That gives us confidence that, hey, we don't expect these things to happen again.
Elon Musk – CEO and Product Architect
Yes. We think the second quarter will be much better. Yes.
Lars Moravy——车辆工程副总裁
Just one thing after another. So crazy.
Elon Musk – CEO and Product Architect
Yes, exactly. Just – if your cars are parked on a ship, they obviously can't be delivered to people. If there's an overdemand for the Model 3 and Model Y in a market, but you don't have it there. This is an extremely complex logistical situation.
So, I'll also say – we've really overcomplicated the sales process, and we've simplified the sales process a lot in the last week or so. As a result, buying a Tesla becomes too complicated, and you should be able to buy the car in a minute. So, we come back – you can buy a Tesla from a rather complicated interface in under a minute.
Martin Vicha
Good. Thank you. Let's listen to Oppenheimer's Colin Rusch. Colin, please unmute.
Colin Rusch – Oppenheimer & Co. – Analyst
Thank you so much, guys. Given Tesla's true quest to become a leader in artificial intelligence in the physical world, in your review of distributed inference, can you talk about what this approach unlocks beyond what's currently happening in vehicles?
Elon Musk – CEO and Product Architect
Ashok, do you have something to say?
Ashok Elluswamy——自动驾驶软件总监
Yes. Like Elon mentioned, just like a car, even if it's a full robo-taxi, it could be used for 150 hours a week.
Elon Musk – CEO and Product Architect
That's my guess, it's about a third of the week.
Ashok Elluswamy——自动驾驶软件总监
Yes. Maybe more or less, but there's definitely going to be some time in that world for charging, cleaning, and maintenance, and you can do a lot of other work, and even now, we're seeing, for example, that LLM companies have these batch workloads, and they send a bunch of documents that run through a fairly large neural network and require a lot of computation to handle those workloads in chunks. Now that we have paid for the calculations of these cars, it is wise to use them instead of letting them be like buying a lot of expensive machines and then leaving them idle. We don't want that.
We want to use the computer as much as possible, close to essentially 100% of the time, to get the most out of it.
Elon Musk – CEO and Product Architect
I think it's similar to Amazon Web Services, when Amazon started out as a bookstore, people didn't think AWS would be the most valuable part of Amazon. So, no one notices this. But they found that they had too much compute because their compute demand would spike to extreme levels for a short period of time during the year, and then sit idle for the rest of the year. So, what should they do to pull the extra calculations for the rest of the year?
It's kind of – yes, monetizing it. It seems easy to say, well, if we have millions, if we have millions, if even tens of millions, of cars in which the computers sit idle most of the time, then we are likely to let them do something useful. And then I mean, if you get to the level of 100 million cars, and I think we're going to get to that at some point, then you have a kilowatt of available computing power, and maybe your own hardware 6 or 7 at that time. So you really — I think you can have about 100 gigawatts of useful calculations, which is probably more than anybody, any company, probably more than any company.
Ashok Elluswamy——自动驾驶软件总监
Yes, probably because driving a car no matter what it takes a lot of intelligence. When it's not driving a car, you just use that intelligence for something else, to solve a scientific problem or answer a stupid question for someone else like a human.
Elon Musk – CEO and Product Architect
We've looked at how to deploy workloads to these compute nodes.
Ashok Elluswamy——自动驾驶软件总监
Unlike laptops and mobile phones, it is completely under Tesla's control. Therefore, it is easier to distribute the workload across different nodes rather than asking for user permission on your own phone, which would be very tedious.
Elon Musk – CEO and Product Architect
Well, you're just draining your phone's battery, so technically, I guess Apple is going to have the most distributed computing, but you can't use it because you can't get it — you can't just run the phone at full power and drain the battery. So, for a car, even if you're a kilowatt-class inference computer, that's insane power compared to a cell phone. If you have a 50 or 60 kWh battery pack, it's not a big deal whether you're plugged in or not. It can be plugged in or unplugged, just like you can run for 10 hours and use 10 kWh of computing power.
Lars Moravy——车辆工程副总裁
We have a built-in liquid cooling and thermal management system. Yes, that's exactly for data centers. It's already in the car.
Elon Musk – CEO and Product Architect
Exactly. It is distributed generation – distributed power supply and distributed cooling. That's already paid.
Ashok Elluswamy——自动驾驶软件总监
Yes. I mean, people are underestimating the cost of distributed power and cooling.
Vaibhav Taneja——首席财务官
Yes. Capital expenditures are shared around the world. In some way, everyone owns a small portion, and they may be able to make a small profit from it.
Elon Musk – CEO and Product Architect
Yes.
Colin Rusch – Oppenheimer & Co. – Analyst
Thank you so much, guys. My follow-up was a bit mundane. Looking at the 4680's ramp-up, can you talk about how close you are to your target production and when you can start accelerating the incremental capacity expansion of the technology?
Elon Musk – CEO and Product Architect
You know, we've made good progress on that. But I don't think that's very important, at least in the short term. As Lars said, we think it will surpass the competitiveness of suppliers by the end of the year. Then we will continue to improve.
Lars Moravy——车辆工程副总裁
Yes. I mean, I think it's also important to note that the ramp now is related to the Cybertruck ramp. So it's like we're not going to build 4680s randomly, unless we have somewhere to put them. Therefore, we will make sure to be cautious about this.
But we have also made significant investments with all of our battery suppliers. They're great partners, they've done great development work with us, and a lot of advancements in technology and chemistry, they're also in their cells.
Elon Musk – CEO and Product Architect
Yes. I mean, a big part of the Tesla 4680 is using internal batteries to hedge against what might happen to our suppliers, because for a while, they were in a very difficult position because every big automaker put in a lot of battery orders. As a result, the price per kilowatt-hour of lithium-ion batteries has reached insane numbers, insane levels.
Vaibhav Taneja——首席财务官
疯子.
Elon Musk – CEO and Product Architect
Yes, just madmen. So, okay, we've got to do some hedging here to deal with the number that cost per kilowatt-hour is twice what we expected. If we have in-house battery production, then we can hedge against demand shocks, where we have too much demand. That's really the way to think about things.
We're not trying to take on a whole bunch of problems just for fun. We started the battery program to solve the problem of the crazy increase in cost per kilowatt-hour from suppliers due to huge orders placed by every automaker on the planet.
Martin Vicha
Good. Thank you. The last question comes from Ben Carlo of Baird. Ben, continue, unmute.
Ben, you're still silent.
Elon Musk – CEO and Product Architect
Well, I want to say it again, I would highly recommend that anyone who is considering Tesla stock should drive a 12.3. If you don't, you really can't get to know about this company.
Martin Vicha
Good. So, since Ben isn't unmuted, let's try Shreyas Patil from Wolfe Research. One last question.
Shreyas Patil——沃尔夫研究——分析师
Hey, thank you very much. Elon, during last year's investor day, you mentioned that the cost per car sold for the next generation of cars will be 3% lower than the current 50% and Y. I think that means about $20,000 in cost of sales. About one-third of this comes from the out-of-the-box manufacturing process.
But I'm curious, if you see an opportunity for some of the other drivers around powertrain cost reduction or material cost savings, can those opportunities be largely transferred to some of the new products that you're talking about launching right now?
Lars Moravy——车辆工程副总裁
Well, of course. I mean, in a nutshell, yes, I mean, like the out-of-the-box manufacturing approach is certainly great and revolutionary, but with that comes some risk because of the new production line and not, but all the subsystems that we're developing, whether it's the powertrain, the drive unit, the battery improvements in manufacturing and automation, the thermal system, the seats, the integration of internal components and the reduction of low-voltage controllers, all of these are transferable, and that's what we're doing, trying to get that into their products as quickly as possible. So, yes, for engineering work, we're not going to try to throw it away and cough, we're going to take it and take advantage of it and use it to make the most of the cars that we're going to make in the future.
Shreyas Patil——沃尔夫研究——分析师
Good. Great. And then on the topic of 4680 batteries, I know you mentioned that you do think it's more of a hedge against the rising cost of batteries from other OEMs. But even today, you still seem to have a cost advantage over some other automakers.
I wonder if there's an opportunity to convert 4680 batteries, and sell them to other automakers, and really generate an additional revenue stream, given the rationalization of the car manufacturing program that you're talking about right now. I'm just curious if you have any thoughts on this?
Elon Musk – CEO and Product Architect
Great. What seems to be happening is that the network is missing something, and battery orders from other automakers have dropped dramatically. As a result, we are seeing suppliers selling at more competitive prices, more competitive than in the past. It's clear that a lot of our suppliers have overcapacity.
Unknown speaker
Yes. This is coming beyond what Elon said - by the way, in addition to what Elon said about 4680, from a supply chain perspective, what 4680 does for us is help us understand the supply chain upstream of the battery supplier. So, a lot of the 4680 deals that we've made, we can also supply these materials to our partners, which will reduce Tesla's overall costs. So we're basically plugging ourselves into the upstream supply chain by doing that.
So, in addition to the excess capacity that these suppliers have, this is also good for reducing overall pricing.
Elon Musk – CEO and Product Architect
Yes. No, I mean, it's obviously going to rise and fall. So there's going to be a boom and bust in battery production, production over supply, and then supply over production, and so on, I don't know, DRAM or whatever. But just like what is true today may not necessarily be true in the future, there will be some kind of boom and bust cycle here.
In addition, government incentives come with additional complexities such as the Inflation Reduction Act, the IRA – I always find it a funny name or a funny name. Yes, like the IRA, the Russian Internet research institute.
Vaibhav Taneja——首席财务官
Segregated Retirement Accounts.
Elon Musk – CEO and Product Architect
Yes, exactly. Ross IRA. It's like the situation with Spider-Man, where the IRA wins. But the incentive structure complicates.
As a result, demand for batteries produced in the U.S. is stronger than outside the U.S. But how long an IRA lasts, I don't know.
Unknown speaker
That's why we have to [inaudible] how to hedge against all this.
Martin Vicha
Good. Thank you so much. That's all we have for today. But at the same time, I would like to make a brief announcement.
I wanted the investment community to know that about a month ago, I met with Elon and Vaibhav and announced that I was leaving the field of investor relations. I'll be here for a few more months or so. So don't hesitate to contact us. But after seven years of sprinting, I'm going to take a break and spend quality time with my family.
I would say that these seven years have been the greatest privilege of my career. I'll never forget the memories I had from the beginning of production hell, looking at the company from the inside and seeing what it has become today. Especially a big thank you to the people in this room and the people outside of the dozens of rooms I've worked on over the years. I think the strength and teamwork of the Tesla team is unlike any other team I've seen in my career.
Elon, thank you so much for giving me this opportunity in 2017. Thank you for regularly seeking feedback from investors and debating with me.
Elon Musk – CEO and Product Architect
Yes. Well, I mean, I reached out to you because I think your analysis of Tesla is the best I've ever seen. Yes, thank you for helping Tesla get to where it is today for seven years. It's been a pleasure working with you.
Martin Vicha
Thank you so much. yes, thank you to all the thousands of shareholders we've met over the years who walk around the plant and love all the interactions, even the tough ones. Yes, looking forward to the next three months of calls, but I'll listen on the other side. Thank you so much.
Duration: 0 minutes
Call Participants:
Martin Vicha
Elon Musk – CEO and Product Architect
Vaibhav Taneja——首席财务官
Lars Moravy——车辆工程副总裁
Ashok Elluswamy——自动驾驶软件总监
Unknown speaker
(Tony Sacconaghi) —AllianceBernaghi—that's what we'd like to see.
Adam Jonas – Morgan Stanley – Analyst
Alex Porter – Piper Sandler – Analyst
Mark Delaney – Goldman Sachs – Analyst
George Gianarikas —— Canaccord Genuity——分析师
Colin Rusch – Oppenheimer & Co. – Analyst
Shreyas Patil——沃尔夫研究——分析师
![Tesla's Chinese version of FSD was exposed to use Baidu's high-assist map?autonomous driving is finally coming?[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)