Good afternoon, everyone, and welcome to Tesla's first quarter 2026 Q&A webcast. My name is Travis Axelrod, Head of Investor Relations, and I'm joined today by Elon Musk, Vaibhav Taneja, and a number of other executives. Our Q1 results were announced at about 3:00 P.M. Central Time in the update deck we published at the same link as this webcast. During this call, we will discuss our business outlook and make forward-looking statements. These comments are based on our predictions and expectations as of today. Actual events or results could differ materially due to a number of risks and uncertainties, including those mentioned in our most recent filings with the SEC. During the question and answer portion of today's call, please limit yourself to one question and one follow-up. Please use the raise hand button to join the question queue. Before we jump into Q&A, Elon has some opening remarks.
Elon?
Thank you. I think we've got a very exciting year ahead of us with 2026. We're going to be substantially increasing our investments in the future, so you should expect to see a very significant increase in capital expenditures. I think it's well justified for a substantially increased future revenue stream. Obviously, Tesla is not alone in this. I think you've seen in most, if not all, certainly the major technology companies substantially increasing their capital investments. We're going to be doing the same. I think it's going to pay off in a very big way. We're investing in and improving our core technologies, battery powertrain, AI software, AI training, chip design, laying the groundwork for significantly increased manufacturing production. We are also strengthening our supply chain across the board, batteries, energy, AI, silicon, everything.
Laying the groundwork, like I said, for what we expect to be a significant increase in vehicle production in the future. Of course, a very significant increase. Well, actually releasing Optimus, but increasing our internal production for testing, and then probably being able to have Optimus be useful outside of Tesla sometime next year. As you've heard me say a few times, I think Optimus will be our biggest product, not just Tesla's biggest product ever, but probably the biggest product ever. I remain convinced of that conclusion. On our vehicle side, it's always, I think, worth noting that a Tesla car is incredible value for money, and they're all autonomy-ready, depending on what part of the world you're in. The supervised Full Self-Driving is getting extremely good. We have just started production of Cybercab, and we'll begin production of our Tesla Semi soon.
Now, I should say, whenever you have a new product with a completely new supply chain, new everything, it's always a stretched-out S curve, so you should expect that initial production of Cybercab and Semi will be very slow, but then ramping up, and going exponential towards the end of the year and certainly next year. In fact, we'll be ramping up production of all vehicles, in all factories, to the best of our ability through the balance of this year. On the energy front, the United States and the whole world will need a lot of energy storage to meet growing electricity demand. Demand for our Megapack is very strong. We're excited to begin production of Megapack 3 later this year in our new world-class factory outside Houston.
For Full Self-Driving and Robotaxi, version 14.3 was a major architectural update, and we have a whole pipeline of major improvements to Full Self-Driving that we believe will lead to unsupervised Full Self-Driving being available anywhere in the world that it is legal to do so. Then there's a version 15, hopefully by the end of this year, but certainly by early next year. That will be a complete overhaul of the software architecture, and will run on AI4. At that point, we're really just increasing the safety level of FSD above human safety level even more, meaning I think even within version 14, we're significantly safer than human, but V15 will take that to another level. We've expanded Robotaxi to Dallas and Houston, using the same software source in the Bay Area.
The limiting factor for expansion is really rigorous validation, making sure things are completely safe. We don't want to have a single accidental injury with the expansion of Robotaxi, and we have, to the credit of the team, not had a single one to date. Optimus, we're preparing Fremont for start of production later this year with Optimus. Again, totally new supply chain, totally new technology, so therefore, the production S curve is always very slow in the beginning. We'll ramp up to significant numbers next year, and we're constructing a second Optimus factory at our Giga Texas location, and that will probably start production around summer next year. The V3 Optimus design is almost ready to demonstrate. I think we want to just make sure it's polished.
Like it works functionally, but there's some aesthetic elements that need to be finalized, and I think probably middle of this year, we should be able to show it off. We're also a little hesitant to show V3 off because we find our competitors do a frame-by-frame analysis whenever we release something and copy everything they possibly can. I think there's some value to not showing new technology until it's close to production. Congratulations again to the Tesla AI chip team for taping out AI5. That's going to be a great chip. I think probably the best AI inference chip for edge compute that exists. Certainly, I think unequivocally the best value for money. Team did a great job, and we already have a lot of momentum for designing AI6, and we've begun to discuss ideas for Dojo 3. This is all very exciting.
We've also finalized plans for the research chip fab on the Giga Texas campus, and we'll start construction of that this year. In conclusion, Tesla is working on a lot of large, ambitious projects. They're all very challenging, but I think they're going to be revolutionary. This is what the team does best, solve the hardest problems and build amazing products. I'd like to thank the Tesla team for all their hard work and thank you to all of our supporters.
Great. Thank you very much, Elon. Vaibhav also has some opening remarks.
Thanks, Travis. 2026 has had an interesting start, not just for us, but I think the world in general. On the autos business, we have seen a resurgence in demand in EMEA and certain countries like France and Germany showing over 150% quarter-over-quarter growth in deliveries. In APAC, we witnessed growth in South Korea and Japan, again, in terms of deliveries. Even out here in the U.S., we have seen a slight growth in terms of quarter-over-quarter deliveries. On the order backlog front, we ended the quarter with the highest Q1 order backlog in over two years. While the recent increase in gas prices has had a positive impact on the order rate, this improvement started before the uptrend in gas prices. This is due to the work done by the Tesla team in bringing more compelling and affordable vehicles to market.
Ten years back, when we launched Model 3 in the U.S. with the promise of $35,000 starting price, which if you adjust today for inflation, translates to about $48,000 in today's dollar terms. The starting price of Model 3 today is way less than that, while the product is way more compelling from where it started. Given this setup, we're focused on increasing our overall production volume, something that we already started in Q1. This volume increase is evidenced by the Giga Berlin reaching a record output of over 61,000 units in Q1. We plan to keep growing volumes further, not just in Berlin, but across all our factories. Our biggest limiter continues to be our battery pack capacity, and we are actively working on resolving that. Auto margins, excluding credits, improved sequentially from 17.9%-19.2%.
Note that we've had certain one-time benefits from warranty write-downs around $230 million and some relief on tariffs. We have not realized any benefit from the recent Supreme Court ruling on IPR tariffs, as there is still a lot of uncertainty around the final outcome. Both tariffs and sustained high interest rates continue to add to our automotive costs. Interest rate subvention costs are recognized upfront. If interest rates continue to rise, our cost of subvention will continue to impact auto margins. On the FSD adoption front, we continue to see improvement, reaching nearly 1.3 million paid customers globally. The bulk of the growth came from subscriptions, while upfront purchases only increased 7% as we removed the purchase option in some markets in Q1. We recently received approvals for FSD in the Netherlands.
This sets us up well for an EU-wide approval later in Q2, and we're just gated by how the regulators go about it. Additionally, we've also received approvals in China. The broader approval is still not there, but we're working with the regulators in the country, and we're hoping that we can get approval by Q3. With these approvals coming through, we expect the broader adoption of the software in the existing fleet and incremental demand for our vehicles. With all this in mind, we have evolved our vehicle sales strategy, where we now emphasize FSD as a product and vehicle as only the delivery mechanism. As we have noted previously, the energy storage business is inherently lumpy, tied to customer deployment timelines. In Q1, we deployed 8.8 GWh of energy storage, a 38% sequential decline. However, we still expect 2026 deployments to be higher than 2025.
We set yet another record with gross margins in this business over 39.5% due to some one-time benefits from certain tariff recognitions of more than $250 million from certain tariffs which we had paid in prior quarters. On a normalized basis, we continue to expect energy compression from here with increasing competition and tariff impacts. As previously discussed, tariffs in this business can have outsized impacts as most of the battery cells are procured from China. Our order backlog for this business is robust, and we're doing our best to build not just based on existing demand, but also on expected demand. Services and others improved sequentially from 8.8% to 9.2%. This includes a collection of efforts meant to support our customers, like service centers, used cars, paid Supercharging, parts sales, insurance, and even our Robotaxi business.
We're making deliberate investments in the infrastructure to help the Robotaxi in the future. We grew the Robotaxi fleet quarter-over-quarter, and we expect to keep ramping the fleet as we accelerate and get into other geographies. On operating expenses side, we did increase sequentially from a full quarter stock-based compensation expense for the 2025 CEO compensation plan, for which one milestone is still deemed probable. Additionally, our spend on AI-related initiatives, including expense on development of our own AI5 chip and new products like Cybercab, Semi, Optimus, and Megapack, et cetera, continue to be at elevated levels, and we expect this trend to continue for the full year 2026. Net income was impacted from mark-to-market charges on our Bitcoin holdings, which depreciated 22% as compared to the last quarter, and the unfavorable impact of FX, primarily from our large intercompany borrowings.
On free cash flow, we ended the quarter with just over $1.4 billion. As Elon mentioned, we are in a very big capital investment phase, which is going to start now and would last a couple of years. Based on that, our current expectation for 2026 is over $25 billion of CapEx. Just to remind you, we are paying for six factories which were going to go into operation. Some have already started, some would go into operation later part of this year. We're further increasing our investment in AI-related initiatives, including the AI infrastructure to support Robotaxi and the launch of Optimus. We've already started placing orders for the research semiconductor fab in Austin and for solar manufacturing equipment.
While this may seem a lot, and we will have the impact of negative free cash flow for the rest of the year, we believe this is the right strategy to position the company for the next era. We'll make such investments in a very capital efficient manner. We are actively working on our mission of building a future of amazing abundance. However, that requires not just a lot of investment, but an immense amount of execution. The future is going to be great, and the whole Tesla team is rising to the occasion to make this a reality. I would like to end by thanking the Tesla team, our customers, investors and vendors for having confidence in us on this journey. Thanks.
Thank you very much, Vaibhav. Now we're going to go to investor questions, starting with a question from Say.com. The first question is: when will we have the Optimus 3 reveal? Which we already touched on, but the rest of the question is: when will Optimus production start, since we ended the Model X and S production earlier mid-year? And then, what's the expected Optimus production rate exiting this year, and what are the initial targeted scales?
Well, as I was saying, what we've found is that when we've unveiled various Optimus versions, we've found out our competitors literally do a frame-by-frame analysis and copy everything we're doing. I think we want to push the Optimus 3 unveil maybe closer to production. Start of production is, we're assuming, somewhere around the late July, August timeframe. Just to inject some reality into these questions, since these questions are. Whoever did those questions does not fully understand what happens with a production line. The last S/X production will be in early May. You have to look at the entire upstream portion of the production line. You have to start with cells, battery packs, motor production, all the parts production.
We've been dismantling the S/X production line from the more basic level parts to, as you get to more larger sub-assemblies, you start dismantling the line from the small parts first, not from the final assembly first. The final assembly line, that'll be dismantled next month after the last of the S/X vehicle's done. You can't dismantle some gigantic production line overnight. It takes at least a few months to do so. You've got to install a new production line, and you've got to provide all of the wiring and communication, test out the machines of the new production line for Optimus. That also takes several months.
Frankly, if we're able to go from stopping production on one line, dismantling that entire line, reinstalling a whole new line, and turning that on in a matter of four months, that is an insanely fast speed. I don't think any other company on Earth has ever done that before, just to put things into perspective and inject some reality into the situation here. I don't know what the production rate of Optimus will be this year. It is impossible to predict these things. When you have a brand-new product and an entirely new production line, and you have 10,000 unique items, all of which have to go right to ramp production, it'll move as fast as the least lucky, slowest, dumbest part in the entire 10,000. Optimus is a completely new product with a completely new production line. It's just literally impossible to predict.
Except that I think it will be quite slow at first as we iron out the 10,000+ unique items that have to be solved for Optimus to reach volume production. Initial skills will be, obviously, we're going to start with simple skills in the factory and then build up from there.
Great. Thank you, Elon. The next question is, what milestones are you targeting for unsupervised FSD and Robotaxi expansion beyond Austin this year, and how will that drive recurring revenue?
Well, we certainly hope to have unsupervised FSD or Robotaxi operating in, I don't know, a dozen or so states by the end of this year. Initially, we're taking a very cautious approach to the rollout here. We haven't had any injuries and certainly no fatalities to date with the unsupervised FSD and Robotaxi expansion. We want to keep it that way. I think probably unsupervised FSD or Robotaxi revenue will not be super material this year, but I do think it'll be material probably in a significant way next year.
Great. Thank you very much. The next question is, when do you expect FSD unsupervised to reach customer cars?
I'm just guessing here, but probably in the fourth quarter. It's difficult to release this to everyone, everywhere, all at once because we do want to make sure that there are not unique situations in a city that particularly complex intersection. Actually, they tend to be places where people get into accidents a lot. Because perhaps there's, like I said, an unsafe intersection or bad road markings or a lot of weather challenges. I think we would release unsupervised gradually to the customer fleet as we feel like a particular geography is confirmed to be safe.
Great. The next question is, how will Hardware 3 cars reach unsupervised FSD?
Unfortunately, Hardware 3, I wish it were otherwise, but Hardware 3 simply does not have the capability to achieve unsupervised FSD. We did think at one point it would have that, but relative to Hardware 4, it has only 1/8 of the memory bandwidth of Hardware 4. Memory bandwidth is one of the key elements needed for unsupervised FSD. It's just generally a thing that's needed for AI. If you're doing an autoregressive transformer, memory bandwidth is the choke point. For customers that have bought FSD, what we're offering is essentially a discounted trade-in for cars that have AI Hardware 4. We'll also be offering the ability to upgrade the car to replace the computer, and you also need to replace the cameras, unfortunately, to go to Hardware 4.
To do this efficiently, we're going to have to set up micro factories or small factories in major metropolitan areas in order to do it efficiently. Because if it's done just at the service center, it is extremely slow to do so and inefficient. We basically need many production lines to make the change. I do think over time it's going to make sense for us to convert all Hardware 3 cars to Hardware 4, because that's what enables them to enter the Robotaxi fleet and have unsupervised FSD.
For what it's worth, in the meantime, we are going to also release a V14 version for Hardware 3. This will be a distilled version of the same V14 software that we released for Hardware 4. People should be able to start the drives from park state, and basically have all the features that V14 for Hardware 4 has. That's expected to come end of June.
Great. Thank you very much. The next question is, what enabled you to finish the AI5 tape-out early, and were there any changes to the original vision? Last week, Elon said AI5 will go into Optimus and the supercomputer, but one month ago said it would go into the Robotaxi. Has AI5 been dropped from the vehicle roadmap?
Well, the reason AI 5 tape-out finished early was because the team worked incredibly hard to make it happen. Just over time, we gathered a lot of momentum. We did have to work every weekend for six months straight, including every holiday. It was a lot of sacrifice by the team and I was there, of course, myself, every weekend. Fortunately, we didn't make any major mistakes, at least that we're aware of, that required pushing out the tape-out. The team just did a great job and worked incredibly hard is the reason. I do expect that AI5 will go into Optimus and into the data center, because it's looking like we'll be able to achieve unsupervised self-driving with AI 4 that is far greater than human safety levels. Which means it's certainly not immediately needed in the car.
At some point, I think it will make sense for us to switch to AI5 in the car, but there's not a pressing issue to do so. At some point the AI4 hardware is going to get so old that it's like, okay, the only reason they're keeping the factory open is for AI4. We are planning an AI4 upgrade to use a newer generation RAM. It'll go from 16 gigabytes to, I think, 32 gigabytes per SoC, so a total of 64 gigabytes. Probably a 10% increase in compute in trillions of operations per second and in memory bandwidth. That's AI4.1 or AI4+ probably goes into production middle of next year, I think. Samsung's doing the modifications for us, so it sort of depends on when they're able to finish those modifications and bring it to production.
Great. The next question is, now that FSD has been approved in the Netherlands and is expected to launch across Europe this summer, can you discuss your robotaxi strategy for the region?
Well, we're probably jumping the gun here on Robotaxi in Europe since it took us an immense amount of time just to get supervised self-driving approved in Europe. We don't control the regulators. We push as hard as we can, but it's ultimately up to the governments in Europe and the EU to decide what to do. Yeah, as it is, we've only been approved in the Netherlands. We expect to be approved in a lot of other countries. I think the supervised FSD goes to Brussels for EU review in May, yeah. Obviously, the next thing beyond that is to aim for unsupervised self-driving or Robotaxi in Europe. I actually don't know what the timeframe for that is, and would be somewhat at the mercy of the regulators as to when that approval would take place.
From a technology standpoint, what we deployed in the Netherlands and Europe is the same exact architecture, and the training procedure and so on, except it had more Europe data. I suspect the same thing will be true for unsupervised FSD as well. Whatever we use to solve in the U.S. will work in other places in the rest of the world too, provided we were able to add the data from the local regions.
Great. The next question is, given the recent NHTSA incident filings, can you update us on the Robotaxi safety data? If safety validation remains the primary bottleneck, why not deploy thousands of vehicles to accelerate removal of the safety driver?
Yeah. Ashok, do you want to take that?
Yeah. We are increasing the amount of our QA fleet, but we also want to use the customer fleet to give us the useful metrics back so that we can scale it safely. Like Elon mentioned, we are absolutely focused on safety. So far, we have zero incidents, and that's what the NHTSA filing also shows. In addition to safety, we are also solving some of these so-called scaling issues. For example, you do not want the Robotaxi to be stuck blocking intersections, or don't want to be dropping people off at slightly incorrect locations and so on. We are simultaneously solving the long tail of safety by monitoring the metrics across the entire Tesla customer vehicle fleet, which is close to driving 10 billion miles on FSD, in the next few weeks.
Also scaling up the amount of QA fleet that we have across the entire U.S. to accelerate our safety validation, while also scaling the rest of the factors that can throttle the increase of unsupervised vehicles.
All right. The next question is V14.3 still the last piece of the puzzle to enable large-scale unsupervised FSD and Robotaxi, or do we have to wait until V15?
Well, no, I think 14.3 is the last piece of the puzzle for unsupervised FSD. Now, the question is degrees of safety and convenience, I suppose. We have a lot of known improvements, major architectural improvements that we know would improve the probability of safety significantly. I think it's not going to make sense for us to deploy unsupervised FSD or Robotaxi at large scale when we know that there are major architectural improvements to the software that can improve safety. I think we're going to want to finish writing that software, validate it, and release it before going to large-scale unsupervised FSD, depending on what large scale means.
We are, of course, as I mentioned earlier, doing unsupervised FSD in three cities, and we'll expand on it to, like I said, probably a dozen states or more later this year. Depends on what your definition of large scale is. I do think it wouldn't be right for us to go to very large scale unsupervised FSD when we know that there are software improvements in the pipeline that would improve safety.
Yep, I'd like to note that the version of Robotaxi that's running in Austin, Dallas, Houston, et cetera, those are essentially 14.3 variants. It's obviously safe, that's why we're able to launch in those cities, and we continue to expand based on the V14.3 base for a while until V15 lands. V15 is going to be a major upgrade.
Yeah.
Great, thank you. The next two questions we've already answered about Robotaxi rollout, and the data that we're observing. We will end on the last question, which is, what is Tesla doing to scale the energy generation business with solar? Residential roof deployments have stalled. Will Tesla move to regional solar and battery farms, perhaps coupled to Superchargers? Will we deploy solar through utilities?
Yeah. The overall US residential solar market is going through a bit of a correction after the loss of the homeowner tax credit last year. We still see strong demand shaping up for the second half of the year. Tesla introduced a lease product this year that allows us to capture the tax credit ourselves and offer competitive pricing for homeowners. We have also debuted our own solar panel with superior performance and aesthetics, as well as our own best-in-class mounting system that gives us a fully integrated home energy ecosystem. We strongly believe that solar and storage markets globally will continue to grow at both residential and utility scale, and we will continue to invest in that growth.
Great. Thank you, Mike. Now we're going to move on to analyst questions. The first question is going to come from Will Stein at Truist. Will, please feel free to unmute yourself when you're ready.
Hi, can you hear me?
Yes, we can.
Great. Thanks for taking my question. Considering the various parties involved in the Terafab project, I'm hoping you can provide some details for investors about which party is going to take responsibility for each aspect of that project, funding it, designing it, building it, operating, taking production, and the like. We'd love to hear some more details.
Yeah. We're still working out the details of the Terafab deployment. In the near term, Tesla will be building the research fab on our Giga Texas campus. This is something we expect to be probably a $3 billion-ish initiative, and capable of maybe a few thousand wafers per month. It's really intended to try out ideas, the research fab, both in terms of maybe we have some ideas for improving the fundamental technology of how chips are made. There's some new physics we'd like to test out, but we also want to test out the ability to see if something is working in production. You need a few thousand wafer starts a month, to make sure that a production process is sound. SpaceX is going to take care of the initial phase of the scaled up Terafab. That's what we've figured out thus far.
Any kind of intra-company thing has to be approved by both the SpaceX and Tesla board of directors. It's got to go through a conflict resolution. It's going to have, unfortunately, a lot of complexity because we've got to make sure Tesla shareholders are served and SpaceX shareholders are served, and strike the right balance there. It takes a while to work through the kind of independent director reviews on this. That's basically what we've figured out thus far is Tesla's doing the research fab, SpaceX is doing the initial part of the large scale Terafab, and then we've got to figure out the rest.
What about Intel's involvement?
Yeah. Intel is excited to partner with us on some of the core manufacturing technologies. We plan to use Intel's 14A process, which is state-of-the-art and in fact, not yet totally complete. Given that by the time Terafab scales up, 14A will be probably fairly mature or ready for prime time, 14A seems like the right move. We have a great relationship with Intel. A lot of respect for the CEO, the CTO and the new team there. We think it's going to be a great partnership.
Yeah. The other thing on the research fab, I think we've said it before, we plan to do memory, logic, everything in the same place, including mask, because we want to have a quick iteration loop so that we can see and basically scale the technologies which we are trying to bring up.
Yeah. I think this will be unique in the world, or at least I'm not aware of any place where you have the lithography mask creation, and then logic, memory and packaging under one roof in one building. That's about the fastest I could possibly imagine doing recursive research and development and being able to try out some pretty radical ideas, some of which have. It's kind of long shot stuff, but if some of these long shots pan out, would be radical improvements in the way chips work.
Great. The next question is going to come from Pierre at New Street. Pierre, please feel free to unmute yourself.
Hey, thanks a lot for taking my question. A quick one first on FSD adoption. You have 180,000 new users, paying users this quarter. I compare that to your overall installed base, it might be 15%, but then if I shrink that to the U.S. or to North America where most of them are, it's probably more like 30%-35%. I compare that to, you probably sold about 100,000 cars in North America in the quarter, so you're winning twice more FSD users than you're selling cars. If I add to that picture the fact that I guess it's mostly Hardware 4 owners who subscribe to FSD, it sounds like most drivers in North America who have Hardware 4 would already be using FSD. Is that the right way to think about it?
The kind of success FSD is seeing today, is that the right way to think about it?
Yeah, I think you're thinking about it the right way, Pierre. The other thing which I'll share is that you can't just look at one quarter versus the other quarter in terms of churn, but we are actually seeing churn of subscribers also coming down, which again, is a reflection of the product is getting better, and obviously if subscriptions are going up, that is a good metric. The other thing also to note is that we are seeing customers actually drive longer, which again, you could correlate it. That's why you have lesser churn because people are liking the product. I've said this before, if I just use my own personal behavior, I literally get in the car, I press a button, and it just goes. Earlier I used to park, now I don't even have to park.
That is the experience which we want everybody to get, and that's why you're starting to see it in the numbers come through.
Excellent. Thank you. If I may, maybe a quick follow-up, completely different. It's more on the Optimus architecture. You talked about the partnership with xAI and Grok, and I was wondering if you can share with us anything about how the System 2 intelligence is going to be implemented. Is that going to be onboard on chips inside Optimus? Or if we should think about your fleet of one million Optimus being produced a year actually driving very significant inference demand in data centers as well for System 2 thinking.
Well, we think we can put a lot of intelligence locally in the robot, and certainly needs to be enough intelligence that if the robot gets disconnected, like if it's a bad cellular signal or there isn't Wi-Fi, Optimus can't just get stuck. It needs to have enough local intelligence that it can still do useful things, even if it loses connection, kind of like the car. The car does not need any cellular or Wi-Fi connection to be able to drive safely. Now, I guess you can think of Optimus needs kind of a manager to tell it what to do, broadly speaking, otherwise it's going to keep doing the same thing it did before. I think you need kind of an orchestration AI, which Grok would be good for orchestration.
For Optimus' voice, having a low latency intelligent voice AI, Grok is actually very good for that. If you want to talk to Optimus and have kind of a Grok level conversation, you need to connect to a Grok level AI for that. I would expect the amount of interaction, apart from the voice stuff and asking complicated questions of the robot that necessarily needs a large AI model to answer, Grok would probably have about as much interaction with Optimus as a manager would have with the people on their team. Meaning Optimus could probably work for several hours without any management oversight.
Great. The next question is going to come from Dan at Barclays. Dan, please feel free to unmute yourself.
Great. Good evening. Thank you for taking questions. Elon, your chip suppliers generally generate pretty good economics on the chip they sell. Your approach has historically been on vertical integration. Part of that has been to get better economics. I know the longer term goal of Terafab is to get the supply you need, but how much of Terafab is also motivated to get better economics on your midterm chip purchases, and how long is it going to take to ramp to get to a yield that achieves that type of economic parity?
No, Terafab is not some sort of mechanism to generate leverage over our chip suppliers. It's just literally we don't see a path to having enough sufficient quantity of AI chips down the road as we scale production to high levels. Just the rate at which the industry is growing, in logic but even more so in memory, we just anticipate hitting the wall if we don't make chips ourselves. That's the reason for the Terafab. I think that we do have some ideas for how to make maybe radically better AI chips. These are kind of research ideas there, which means long shot, but if long shot pays off, it's maybe a giant improvement. It's just easier to do that if we have our own research lab and are developing our own production technologies.
If you look long term at say, having AI satellites, making chips for those, there's just no way in hell the existing industry can keep up with that. It's impossible.
All right. Our next question is going to come from Mark at Goldman Sachs. Mark, please feel free to unmute yourself.
Yes. Good afternoon. Thank you very much for taking my question. I recognize the importance of FSD, and that FSD can help to drive vehicle sales, and nice to see some of the improvements in the FSD technology more recently with version 14. However, I'm also hoping to understand if the company's view on new vehicle models has evolved. I ask, given that, Elon, you posted on X recently that Tesla could develop a family vehicle, and there's also been some past discussion about a compact vehicle.
Well, Cybercab is the compact vehicle. It's very roomy, but it's a two-person vehicle. We do think probably most of our production long term will be Cybercab because 90% of miles driven are with one or two people. It would mean that you'd want a vast majority of your production to be Cybercab. Over time, it's going to make sense for our whole lineup to be autonomous vehicles of different sizes. I did talk a bit about this when we did the AI Day in L.A. at Warner Bros. and showed this is our current lineup and this is some idea of what our future lineup will be, which is that it's going to be almost entirely autonomous. In fact, long term, the only manually driven car will be the new Tesla Roadster.
Speaking of which, we may be able to debut that in a month or so. It requires a lot of testing and validation before we can actually have a demo and not have something go wrong with the demo. I think it will be one of the most exciting product unveils ever. I'm not sure. I don't think it moves the needle massively from a revenue standpoint, but it is very cool. I think it might be one of the most spectacular demos ever.
All right. Mark, did you have a follow-up question?
Yeah, thanks, Travis. My other question was on batteries, and the company mentioned batteries as a constraint on its growth. Can you speak more to how Tesla expects to resolve this and to what extent that might come from ramping up your own LFP and 4680 battery cell manufacturing? Or is this something that you'd expect to resolve primarily with increased sourcing from suppliers? Thank you.
Yeah. At the moment, I think the limiter is not the cells itself, it's the battery pack capacity. Like I said in my opening remarks, we're actively working on resolving this. There's more capacity being added as we speak, and I'll let Lars add a few more things to it.
Yeah, thanks, Vaibhav. As you guys may have seen in Berlin, we started launching a Model Y battery pack with our in-house 4680 cells a few months ago, and that is ramping up nicely, adding to Berlin's output and helping with the demand surge that we've seen in Europe. As well, we're adding additional capacity in our Reno facility, sort of retooling it as it's been building packs now for almost 10 years in order to put in some more efficient lines and get additional output out there. We continue to have growth in China as well, ramping in-house LFP module production and battery packs associated with that. All of those things are happening now and in the next months, and that's really plans we laid out a few months back to increase that output with the growing demand.
All right. Thank you, guys. Our next analyst is going to be Colin from Wells Fargo. Colin, please feel free to unmute yourself.
Oh, great. Thanks for taking my questions. You moved the safety driver in Austin, and you're now expanding into Dallas and Houston. What are the key safety metrics that you're tracking that gives you confidence that Robotaxi is safe enough to expand? Is it miles per intervention, miles per accident, per fatality? Where do you stand on that now?
Ashok?
Yeah. We track basically all the metrics that you mentioned. We have a pretty large QA fleet spread across all of the United States. Then we look at any intervention that could happen and then sort of simulate both in practice and also in our simulators that are very good nowadays, using neural networks, as what would have happened. Then, based on all these analysis, in the end, we make the call to expand. So far, all of the expansions have gone according to our expectations.
Yeah. I think a lot of what limits wider deployment of Robotaxi are actually not safety issues, but convenience issues, or the car basically gets paranoid and gets stuck. Because it's programmed for maximum safety. The problem is that then it sometimes gets scared to do things.
Yeah.
Like it sometimes gets scared to cross railroads, for example, or it'll get stuck at a light where the light never changes from red. There was one kind of amusing situation where a whole bunch of robotaxis got stuck in the left turn lane in Austin because, I kid you not, a Waymo had crashed into a bus. They could not turn left because the Waymo had crashed into the bus. You had this long line of, I don't know, a dozen or more Tesla robotaxis that were waiting for the bus to move, but the bus was never going to move because the Waymo was crashed into the bus. That obviously drives people crazy if there's a whole bunch of robotaxis blocking the whole road. It's a ton of things like that.
That's the single biggest thing is just the car being scared to move or getting stuck in situations like that. We've also had literal infinite loops where the car might want to make a turn into a road, but there's construction, and then it goes around the block, tries to turn into the road with construction, goes around the block, tries to turn the road. You have to stop the infinite looping, the literal infinite looping. Those are by far the issues that we have to resolve as opposed to direct safety issues.
Got it.
Great.
And then-
Did you have a follow-up, Colin?
Yeah. Last year, I asked about FSD and camera and the issues with sun glare, and you noted that there was a breakthrough with direct photon counting that addressed this issue. A month ago, there was a NHTSA filing saying that they haven't received an update when the solution was deployed and the number of vehicles. Did it require a retrofit of the camera? Is this fully deployed? I guess I'm just curious since the filing mentioned it.
Yeah. First, I want to say we did change the cameras some months ago, and those are out. The NHTSA filing is referring to older vehicles. We always work directly with NHTSA on all of the issues that they raise with us, and they're asking for quite a bit of information, and we're complying with that in as timely a manner as possible. We expect to resolve that and any of the other investigations in short order.
Yeah, we have also implemented stricter measures for the visibility of the camera. In recent software builds, if the camera is not able to see things clearly because of residue build-up or what have you, then the FSD won't be available for those cars.
You have to clean the inside of the windshield.
Great. That unfortunately is all the time we have today. We appreciate everyone's questions, and we look forward to talking to you next quarter. Thank you very much, and goodbye.