On March 1, Tesla held its 2023 Investor Day. Musk and Tesla executives made a collective appearance and explained the underlying logic of Tesla's business and products from various angles. This is a Q&A part after Musk's announcement of the Mexican Gigafactory and its executives.
The video is about 44 minutes long and has about 11,000 words.
Source video: Tesla / Translation & Subtitle: Rubble Villager
List of executives and timestamps
17 people from left to right (personally organized, information may not be accurate):
Martin Viecha: Director of Investor Relations
Brandon Ehrhart: General Counsel and Company Secretary
Laurie Shelby: Vice President of Environment, Health and Safety
Mike Snyder: Senior Director of Megapack
Tom Zhu: Global Vice President, Head of Greater China
Karn Budhiraj: Vice President of Supply Chain
Roshan Thomas: Vice President of Supply Chain
Rebecca Tinucci: Senior Director of Charging Architecture
Zack Kirkhorn：CFO
Elon Musk：CEO
Drew Baglino: Senior Vice President of Power Systems and Energy Engineering
Ashok Elluswamy: Director, responsible for FSD
David Lau: Vice President of Software Engineering
Pete Bannon: Vice President of Silicon Engineering and Low-Voltage Electronics
Colin Campbell: Vice President of Power Systems Engineering
Lars Moravy: Vice President of Vehicle Engineering
Franz von Holzhausen: Chief designer
00:37 Elon announces the Mexican Gigafactory
03:28 Rod Lache: Timeline for the next generation of cars
05:44 Adam: Mining and refining of lithium
09:18 Ben Kallo: What more can Tesla do to drive renewable energy?
11:51 Philippe Houchois: Two-way charging
13:57 Philippe Houchois: How many models are needed to reach 20 million cars a year
14:50 George Gianarikas: Market share growth in China; Geopolitical implications
16:42 Colin Rusch: Speed of learning; Demand issues
20:14 Emmanuel Rosner: Near-term priorities for next-generation vehicles; Conflict with the needs of existing models; The most difficult thing is production
27:30 Dan Levy: Cost strategies and regional variations
31:33 Overraka(?) :capital expenditure
33:41 Overraka(?) : Multi-stroke reconstruction of FSD
34:57 Alex Potter: An update on dry cell electrodes
39:13 Chris: Details of the next generation of cars (Elon refused to answer)
40:08 Will Danoff: When will AI help produce cars
The rubble is focused
Elon
* (Q&A session) should be a discussion about the long-term future, not just focusing on the production of the rest of the quarter, let's try to direct the question to long-term value creation.
* Tesla's next Gigafactory, near Monterrey, Mexico.
* We don't want to get into mining or refining, but if necessary, we will. Refining capacity is the main bottleneck.
* If there is an entrepreneur who wants to win without losing money, there is no doubt that it is refined lithium, or an anode cathode, or any material of lithium battery.
* How many models are needed to achieve an annual production of 20 million vehicles? Elon: Not many, maybe ten.
* To a large extent, demand is a function of affordability, not expectations, which is critical.
* We found that even small changes in prices have a large impact on demand, very large. It's helpful to know this.
* As long as we improve the affordability of cars, demand will explode. The difficulty lies in the production of the car and the entire supply chain that supplies the car.
* The most important of these problems is the production of battery cells. In fact, we deliberately overproduce or oversupply cells beyond what the car needs, because if it is less than the car needs, then the factory will come to a standstill.
* We can oversupply cells and battery packs and then regulate the production of stationary energy storage, which is much easier than regulating car production.
* We acquired Maxwell at the time, indeed only for dry electrode technology. But it also shows how big the gap is between small-scale feasibility and large-scale viability.
* In terms of training, we have one of the largest neural network training systems in the world. We expect to increase the capabilities of the system by an order of magnitude by the end of this year. And then maybe by the end of next year, with some combination of Nvidia and Dojo, it could be another order of magnitude.
* For the next generation of vehicles, we will arrange a suitable product launch event.
I don't think AI can help us produce cars in the foreseeable near future.
Drew Baglino
* We do have large lithium suppliers at the moment. They understand the refining program at our Corpus facility and the technology we're trying. We let them know about this because we thought these technologies were fundamentally more scalable. As these technologies are validated, we plan to share them with them.
* Probably in the next two years or so, we will bring this feature (bidirectional charging) to all of our vehicles.
* Elon has told the team many times that scrapping equipment or wasting money is acceptable, wasting time is not.
* [When it comes to battery production] our internal goal is roughly to increase capacity by 1,000 weeks per quarter, and that's our pace.
Tom Zhu
* As long as we provide valuable products at affordable prices, there is no need to worry about demand. That's basically the philosophy we follow.
* (Answering a U.S.-China geopolitical question) As long as we are needed in this country, I don't see any risk.
* We localize as much as possible. For example, in China, more than 95% of our supply chain is localized, from Tier 1 to Tier 2 to Tier 3 suppliers, which brings huge savings in production costs.
Lars of Moravia
* The next-generation platform isn't just one car, it's multiple models, and that's a segment where we'll try to focus on affordability and how much we expect.
Full text of the new plant announcement and Q&A
Elon:
Okay, that's probably the most important announcement today. We are pleased to announce that we will be building a gigafactory in Mexico.
Obviously, we'll be holding a big opening and breaking ground, but we're excited to announce that Tesla's next Gigafactory will be near Monterrey, Mexico. We are very excited about this.
I want to emphasize that we will continue to scale up production at all of our existing plants, including California and Nevada, and obviously Texas, as well as Berlin and Shanghai. We intend to increase the capacity of all our plants, and the Gigafactory in Mexico will complement the capacity of other plants.
To clarify, this is not a transfer of capacity from one place to another, we are purely trying to increase the total global capacity.
In short, we are excited to announce this news. In fact, I believe that the governor is on the scene, not very clear, welcome! I believe the Secretary of State is there too.
In short, we look forward to its completion, and we will hold a large event for the groundbreaking and opening.
Speaking of which, I think we can move on to the Q&A. Obviously, we have enough bench depth. There may be too many people on stage, but we will try to answer questions within reason.
This should be a discussion about the long-term future, not just focusing on production for the rest of the quarter, let's try to steer the question to long-term value creation. Okay, let's start asking questions
Martin Viecha:
Rod？
Rod Lache:
Hi, I'm Rod Lache at Wolfe Research.
You've talked about your plans for the next generation of cars, powertrains, and batteries, and it's very exciting, and I hope you'll talk about the timeline for deploying these new products.
It sounds like the plan isn't just a new car, but a paradigm change in how the vehicle is assembled and assembled. Will this affect all the work you're doing so far?
The Model Y, which is being produced here, will be produced in a very different way in the future. Can you give us a feel for what's going to happen next and what the timeline for achieving it all looks like?
Elon:
Okay, I'll go about it briefly, but in general, the vehicles of the future will use the most fundamental architectural changes.
Remaking the molds in the factory means shutting the factory down for a long time, and I hope it doesn't.
However, the production method of Model Y has different differences, there are rear body integrated casting, front body and rear body are cast in one piece, and there is a structural battery pack. Then, there are some smaller improvements. But I think the vehicles of the future will contain these really big changes.
Do you have anything to add? Lars, maybe you have something to add?
Lars of Moravia:
I?
Yes, I mean... I agree with you one hundred percent, Elon.
It's easy to add innovation to a new car, and in the long run, we will obviously bring innovation. We've been discussing this, but we don't want to shut down the factory. In terms of timelines, we will move forward step by step as quickly as possible.
As always, Elon has predicted that our next generation will be produced at the Mexican plant, but so will our other plants. Everything is about keeping all factories running.
We expect that this will be a huge product, and we will be advancing rapidly in the next few years.
Lars Moravy: Vice President of Vehicle Engineering
Martin Viecha:
Next, I give Adam a question.
Adam:
Thank you. First of all, congratulations Mexico, come on! Great, congratulations!
Elon, about applying first-principles thinking and innovation to an area that is not currently under your control, that is, the extraction and extraction of some key raw materials.
I'm sure a few years ago, you had a patent on sodium chloride to extract lithium from clay, amphibole clay and this class of substances.
Maybe some innovation is needed to wake up the mining industry and reduce its costs, what are your plans for that? Because it seems like this is a real limiting factor.
Elon:
Well, at any given moment, we're going to try to address what we think is the limiting factors. We want to do the least possible and we don't want to go into mining or refining, but if necessary, we will.
I believe that there is a need to focus on refining capacity. We need to produce a large number of anodes, cathodes, lithium hydroxide and lithium carbonate. Refining capacity is the main bottleneck.
That's why we're building a lithium refinery in Corpus Christi.
Drew Baglino:
In terms of mining companies, and looking at this part of the value chain, we do have large lithium suppliers right now. They understand the refining program at our Corpus facility and the technology we're trying.
We let them know about this because we thought these technologies were fundamentally more scalable. As these technologies are validated, we plan to share them with them. Because as Elon said, we don't really want to do these things.
We do it because these things aren't moving fast enough. If we can demonstrate that progress can be made faster, our aim is to transfer that knowledge to current suppliers
Similarly, we're experimenting with clay craftsmanship, and we're continuing to do so, and that's the same at this point.
We also work with our suppliers to conduct trials and share knowledge. Our purpose is to help the world do this better. The core problem is the extraction of lithium or other elements from the ore.
Elon:
Obviously, next to this building, we are building a cathodic processing facility. Go a little further and you'll see another large building that is used to refine the cathode.
But as I said, we really wish other companies were doing it. We do it because we have to, not because we want to.
Drew Baglino:
In the United States, there is no large-scale cathode production, but it is needed.
I repeat, if we do, we will try to do it from first principles. We have tried a range of new approaches and we are confident that they will work. Once these methods are validated, we want to bring them back to our suppliers so they can build new facilities faster with less investment.
Drew Baglino, Senior Vice President of Power Systems and Energy Engineering, from right
Martin Viecha:
Well, the next one is Ben.
Ben Kallo:
Hello, I'm Ben Kallo from Baird.
Similarly, for renewable energy, since this is an important pillar of Master Plan 3, can Tesla do more to accelerate the development of the entire renewable energy industry?
Elon:
Well, I don't know what else we haven't done, but I can say that if there is an entrepreneur who wants to win without losing money, there is no doubt that it is refined lithium, or anode cathode, or any material of lithium battery.
Yes, I think we are already doing everything we can.
Drew Baglino:
In terms of renewable energy, what we can do with Tesla is this: the more we reduce the cost of energy storage, the more flexible the load we bring to the grid by allowing the car to charge at the right time, the more valuable renewable energy is.
I'm not kidding, because there's a lot of wind energy right now in Texas. At night, there is almost too much wind energy because they don't want to shut down nuclear power (?).
Bringing truly low-cost energy storage into the grid will make renewables more valuable, which will ultimately accelerate their deployment. I think that's what we're focused on.
Martin Viecha:
Next, Philippe?
Philippe Houchois:
Thanks, I'm Philippe Houchois from Jefferies, I have two questions.
First, every company in this industry wants to be Tesla. Every car manufacturer is trying to emulate what you're doing, what you've accomplished. But they are not focused on achieving the highest possible growth with as few models as possible.
I want to know, since your goal is to reach 20 million units by 2030, how many models do you need to reach that goal, and how does that fit into your plans for mega-scale? How do you manage this?
There's also the fact that probably consumers don't want to see Tesla or the same Tesla on every street corner. I just wanted to see what you guys think about this.
My other question is about bidirectional charging.
You talked a lot about making the world use more renewable energy and better use of cars. In my opinion, two-way charging is a way to make better use of the car. But in the past, you didn't seem to be willing to do that. I would like to know what your latest views on this topic are.
Drew Baglino:
Regarding two-way charging, we didn't intentionally decide not to do it, just to say that at the time, it wasn't a priority.
As we continue to improve the power electronics of our vehicles, we find different ways to charge in both directions while also reducing the cost of vehicle power electronics.
In everything Tesla does, the goal is usually to get more for less. We are carrying out a technological transformation of power electronics. Probably in the next two years or so, we will bring this feature to all of our vehicles.
Well, that's all I want to say.
Two-way charging, you can use the car's electricity to feed back to the grid
Elon:
I don't think there are many people who want to use two-way charging unless you have a power wall, because if you unplug your car, your house goes completely black, which is extremely inconvenient.
Drew Baglino:
The main value of this thing comes from charging the car at the right time, not transmitting energy in reverse.
Elon:
If you have a power wall that can withstand the load of your home, then you can use your car as a supplementary energy source to the power wall. Then, you won't be maddened by unplugging your car and the house losing power.
I think there is some value in this matter as a supplementary energy source in the future. If you have a power wall, then the people in the house will always have convenient access to electricity.
Martin Viecha:
There is also the question of the number of models.
Elon:
Oh, how many models.
There won't be many, maybe ten? I don't know, not a lot.
The situation in the traditional automotive industry is that people have nothing to do. People have nothing to do but turn upside down and do some random combinations, and the cars are almost exactly the same...
I mean, how many variants of the model do you see on the road?
There may be hundreds.
Are these variants all good?
No, most of them are not, they are just mutating for the sake of variants.
But look at the evolution of mobile phones, there used to be hundreds of flip phones, and now? The car of the future will look like this.
Martin Viecha:
George？
George Gianarikas:
George Gianarikas of Canaccord Genuity.
I have a question about your market share growth plans in China. And, will the political tensions between the U.S. and China affect your long-term grand plans? Thank you.
Elon:
Hey Tom, come on, don't stress too much.
Tom Zhu:
Well, in fact, our market share in China is still growing strongly. In particular, earlier this year, we made price adjustments. After that, a lot of demand was generated, which actually exceeded our capacity.
As Elon said, as long as we offer valuable products at affordable prices, we don't have to worry about demand. That's basically the philosophy we follow. We want to do everything we can to reduce costs from the supply chain, from improving factory efficiency, and pass on this value to our customers. I think as long as we continue to do that, I won't worry too much about China's market share.
Then, geopolitics, no one can know, let's do our best. In fact, we create a lot of jobs for the local community, and we're also creating a lot of jobs in our suppliers and factories. We contribute a lot to the local economy.
I think as long as we're needed in this country, I don't think there's any risk.
Tom Zhu: Xiaotong Zhu, Global Vice President, Head of Greater China
Martin Viecha:
Colin?
Colin Rusch：
Oppenheimer的Colin Rusch。
One of the things that stood out to me during the presentation was the operational efficiency metrics that you talked about. Can you talk about the speed of learning and how to track it? From an innovation perspective, when you go into other fields, how do you think as an organization?
Elon:
Does anyone want to try to answer?
Drew Baglino:
I'll talk about the more abstract point, which is that you can't improve the things that you can't measure.
At Tesla, we are like ruthless measuring machines. Once we started measuring those metrics that helped with operational efficiency, we had a path to improvement. I think the key is to have a good yardstick.
Elon:
Okay, and I should say that I've said this many times over the last few years, but sometimes I need to say it again: people's expectations of owning a Tesla are very high. The constraint is their ability to shell out money for a Tesla, not whether they want one.
It is easy for all of you here to lose sight of this. If your income far exceeds the price of the car, then you focus on price-performance without thinking about affordability. But for the vast majority of people, buying a car is driven by affordability. That's why we can't simply double the price of a car.
Or you can imagine the extreme situation, if there is a car that people expect infinitely, but it costs $10 million, it doesn't make much sense, because most people can't afford it.
To a large extent, demand is a function of affordability, not expectations, and that's key.
One thing we weren't sure about before was what exactly was the price elasticity of Tesla's demand, and how much would demand increase when we lowered prices? We have found that even small changes in price have a big impact on demand, very large. It's helpful to know this.
And then, autonomous driving is a very, very important proposition, because the assets you own could potentially gain five times the utility they currently have.
If a passenger car is used for 10 or 12 hours a week, it also pays a lot of parking costs, while a self-driving car may use 50 to 60 hours a week and can reduce a lot of parking costs. If all this holds true, this could become the largest asset value increase in history overnight as fleets turn on autonomous driving.
Martin Viecha:
Immanuel?
Emmanuel Rosner：
Thank you very much, Emmanuel Rosner from Deutsche Bank.
As you begin to roll up next-generation vehicles, what are your recent priorities in terms of market segment or vehicle?
The slide you showed showed had two wrapped cars, one of which looked like a van and the other that looked like a smaller car, which I guess could be the Model 2. What are your near-term priorities for upgrading the next generation of vehicles?
How do you ensure that as costs fall by 50%, you don't cannibalize the demand for existing vehicles?
On the right are the two wrapped next-generation cars mentioned by Emmanuel
Elon:
From the expectation of wanting to own our vehicle, the demand can be limitless. At present, it is almost the same as infinity. Affordability is key.
Basically, as soon as we make cars more affordable, demand will explode. The question is, how do we produce cars? The difficult part is producing cars. I cannot stress this enough, the difficulty lies in the production of cars and the entire supply chain that accompanies them.
This is an exceptionally difficult logistical challenge, and all the parts and materials that must be assembled on the vehicle must be scaled up with the car, all of which require exponential ramp-up.
Whatever the reason, as long as one link goes wrong: earthquakes, floods, fires, revolutions, I remember I went through all of that, and the parts of the supply chain were broken and you stopped.
The difficulty of producing a car far outweighs other factors, as well as the corresponding supply chain.
Which of you would like to talk about supply chain issues?
Karn Budhiraj:
In my talk, I mentioned that perfection is just a passing grade, and we really need everything to happen perfectly. And strategies to mitigate various risks – some anticipated, others not – do depend on the circumstances. The supply chain we manage requires expertise and in-depth understanding to come up with these strategies.
As Elon said, we've been through it all, from recurring tariffs, then ocean logistics, chip shortages, COVID, floods, a fire in Japan and the closure of a factory. Malaysia once broke out the new crown, and the back-end of many chips stagnated. This was also a storm that Elon had intervened in, which was nerve-wracking, but in the end we survived.
Third from left, Karn Budhiraj: Vice President of Supply Chain / Roshan Thomas, fourth from left: Vice President of Supply Chain
Roshan Thomas：
Yes, as hard as it is, I think we've been laying the groundwork to reach out to all levels of the supply chain as closely as possible, to build control, to guide the different levels of the supply chain. This is how we try to mitigate the corresponding risk.
Similarly, dual supply chain, triple supply chain, maintaining redundancy is also how we have been trying to mitigate risks.
Lars of Moravia:
Yes, when it comes to vehicles, Model 3/Y is an architecture and S/X is an architecture. Our next-generation platform will not only target one market segment, in fact, we are considering all the existing segments that we have not yet entered, as well as future markets, and co-designed with our supply chain partners to allow us to quickly enter these segments to meet our needs.
Echoing Elon's point, if we make a car exciting and affordable at the same time, many times, it doesn't necessarily matter which segment it is in, because it's the car you want. We saw this on the Model 3, where many people thought the sedan wouldn't be a big hit, but we sold a huge amount.
The next-generation platform isn't just a car, it's multiple models, and that's a segment that we'll try to focus on affordability and expectation.
Elon:
Yes, there is an old saying, win battles by tactics, wars by logistics. The logistical challenges here are enormous.
When we start to take up a significant percentage of the industry, we can't oversupply, which is unrealistic. Some of these of the dual supply chain, triple supply chain mechanism you are talking about, are suitable for small-scale projects, but not for large projects. Because using the three-supply chain is like an airplane with three engines, and if any one engine hangs, the plane crashes.
We have to either oversupply, which increases capital expenditures, bring idle suppliers, build large warehouses, or design for a certain level of reasonable overloading, and then there are rush costs, because it is inevitable that something will go wrong and we will have to temporarily airlift supplies.
The speed of progress is indeed the speed at which we scale with 10,000 logistics problems.
Among these problems, the most important is the production of battery cells. In fact, we deliberately overproduce or oversupply cells beyond what the car needs, because if it is less than the car needs, then the factory will come to a standstill.
But what do we do with all the extra cells? Easier to regulate production are power walls and megapacks, fixed energy storage. In this way, we can oversupply cells and battery packs, and then regulate the production of stationary energy storage, which is much easier than regulating car production.
Strategically, I think that's a good thing. The Megapack's capital expenditure is small compared to cars. In addition, the demand for megapack is almost limitless. Basically, we can sell as many megapacks as we can as long as it is competitive with the utility grid.
Yes, the demand is almost limitless. In other words, the demand has a few terawatt hours, and we still have a long way to go to reach a few terawatt hours per year.
Martin Viecha:
Dan？ Next question, Dan asks.
Dan Levy：
Thank you, Dan Levy of Barclays.
We know that the competitive landscape varies from region to region, and the cost landscape varies from region to region. We see that the cost advantage of the Chinese Gigafactory is much higher than that of the Fremont factory. Obviously, their dynamics are quite different.
So to what extent do the cost strategies you develop today vary by region? Or are there more globally consistent ways to reduce costs?
Tom Zhu：
Yes, I think our strategy in this area is pretty consistent, which is that we localize as much as possible. For example, in China, more than 95% of our supply chain is localized, from Tier 1 to Tier 2 to Tier 3 suppliers, which brings huge savings in production costs.
We have a highly localized workforce, and we also have access to skilled labor in the Yangtze River Delta region, and I mean Shanghai in particular. Of China's more than 30,000 employees, there may be fewer than 20 expatriates. We had deep localization there to build that cost structure. And we try to replicate this approach in different gigafactories.
"China Speed", it took only nine and a half months to build the Shanghai Gigafactory
Of course, the supplier base is different, and the labor market varies from region to region and country to country, but we strive to localize. I think that definitely gives us an advantage to compete with automakers around the world.
Another point, like Zach said, is that we have a direct sales model, which also saves a lot of operating expenses, and we have full control over all expenses across the company. And we don't spend money on marketing or advertising. Every penny we save becomes the value we can provide to consumers. In this part, we have also followed a fairly consistent strategy for many years.
Lars of Moravia:
Of course, Colin, Pete and I talked about production, whether it's produced here, or in Europe or Asia, it applies everywhere. We are thinking fundamentally about this issue and are not targeting any region.
Zack Kirkhorn:
I would like to add to what Tom talked about about localization and Lars' point. In terms of our plants, processes, and cost reduction solutions, we are moving towards greater standardization.
When we think about the next generation of platforms, we're always thinking about the capacity we want to build on this platform, how many individual factories we need to build, and what is the fastest way to expand our construction footprint globally.
For the Model Y, our current approach is that each plant is an incremental improvement over the previous one. This requires engineering time and engineering costs for each plant design. Then, you'll notice that the factories are slightly different from each other.
As we moved to next-generation platforms, I remember Lars you used the term copy-paste, and in order to get things right the first time, we definitely learned some lessons and needed to make some adjustments. But we need to be as standardized and general as possible so that we can scale as quickly as possible.
Zack Kirkhorn：CFO
Martin Viecha:
Overraka (?) over here.
Overraka(?):
Thanks, I have two questions, one for Zach and one for Ashok.
To Zach's question, how do you see the long-term growth and operating margin of the business? 150 billion to 175 billion in capital expenditure, is this basically a guide for capital spending until 2030?
To Ashok's question, you highlighted the multi-trip reconstruction, how is this different from Mobileye's REM map? And can you tell us about the AP4 hardware?
In his speech, Zack gave financial guidance corresponding to long-term business goals for 2030
Zack Kirkhorn:
With regard to operating margin, we intend to continue to improve operating expenses as a percentage of total revenue over time. We will continue to reduce the cost of our products and strive to maintain gross margins at a healthy level.
From the hardware part of the business, we expect to continue to be healthy over time. Then on top of that, there is also a software section. Elon has commented several times on the impact FSD software has had on the economic indicators of the business. In this area, profit margins, operating margins will have a great impact.
Specific to your question about capital expenditure guidance, the purpose of that slide is not to provide specific guidance, but simply to be transparent and convey our internal rough estimates.
Providing context for the targets of 20 million vehicles and 1 terawatt-hour of energy storage per year, we believe these targets are quite feasible relative to the expected cash income of our business. This number may change as we learn more. We may choose to increase our vertical integration efforts in some areas, and we may find efficiency issues that can be improved in others. This also goes back to the whole conversation earlier about whether we do mining or not. But our point is to say that, based on our predictions, this is entirely possible.
Ashok Elluswamy: Director, responsible for FSD
Ashok Elluswamy:
The important point I want to say about multi-stroke reconstruction is that we want to automatically label most of the data. We can collect data from the fleet and then reconstruct the local area, which can serve as oversight for these video footage.
We want to build a scalable autonomous driving system, and we don't want to rely on HD maps or other mechanisms, although we can easily build a system with the same technology. But this technology is mainly for automatic annotation, providing accurate three-dimensional labels for all video sequences involved in the reconstruction.
Elon:
I don't know how much we've talked about this, but in terms of training, we have one of the largest neural network training systems in the world.
We expect to increase the capabilities of the system by an order of magnitude by the end of this year. And then maybe by the end of next year, with some combination of Nvidia and Dojo, it could be another order of magnitude. I'm not sure if the scale is known, it's huge.
During the presentation, Ashok introduced the current training cluster
Martin Viecha:
Maybe bring up the last two questions, one from Alex and one from Christ.
Alex Potter：
Okay, Piper's Alex Potter.
I certainly want to ask Drew or anyone else on stage about the latest situation with dry battery electrodes, and if we want to try to overfeed the cells and scale them, there's a lot of variability, it's a complex orchestra of supply chains.
But at least for me, a lot of the key lies in the dry cell electrodes. How are you going?
This was the part that attracted me the most about the factory tour, seeing through that window that this was obviously not a scientific project, but any information you would like to share: production, progress, where you are today, how it compares to the progress you had hoped for.
Thank you!
Drew Baglino:
As you can see, this is a real factory that produces a large number of dry electrodes in an automated way. We have come a long way.
People may have different opinions, but we are perfectionists, we have clear end goals, and we are making progress towards those end goals every week, whether it's the speed of the mold, the output of the production line, or the downstream process. We are not standing still in the rate of progress, both in terms of anodes and cathodes.
One of the big benefits of oversupply, which Elon mentioned, is that it gives us the opportunity to experiment and improve, rather than sticking to the approach we happened to take when we started a year and a half ago.
Elon has told the team many times that scrapping equipment or wasting money is acceptable, but wasting time is not. We've been experimenting with the way we are most likely to succeed.
In fact, in the factory here, you see more than one anodic production line, they are actually operating two production lines, and the final process step of the powder into the mold is slightly different in between. And it's a competition, which line will have higher output and which line will perform better. We are in a position to do that, and we are taking advantage of that to advance technology as quickly as possible.
Elon:
The dry electrode problem is indeed a rather difficult one. We bought Maxwell only for dry electrode technology. But it also shows how big the gap is between small-scale feasibility and large-scale viability.
In February 2019, Tesla acquired energy storage company Maxwell Technologies at a valuation of $218 million
We have a very talented team of engineers working to scale the dry electrode process and make it reliable and consistent. We've been working on this for quite some time, and there's a good chance we'll be able to reach mass production this year.
Drew Baglino:
Yes, we are basically ramping up production every week. Our internal goal is roughly to increase capacity by 1,000 weeks per quarter, and that's our pace.
I think what you're seeing is basically one ton of raw material per hour per line. It's hard to understand what this really means, unlike "Oh, this technology works in the lab." When processing tons of raw materials per hour, there are all sorts of problems. Even if only 0.1% of the dust escapes to the equipment area, it can cause dust problems, which can short-circuit some electronic devices, problems like this.
At lab scale, you won't even notice this problem. But after processing thousands of tons of raw materials in a matter of months, you're like, oh, there's a new failure mode here. That is where we are at the moment. But we're solving these problems, and the team is working on, and it's making progress every week.
Martin Viecha:
The last question comes from Chris.
Chris:
Thank you for taking my questions, I have a few follow-up questions about the next generation of cars.
First, when will we see it, even if it's just a prototype?
Second, are there any details to share in terms of size, content, and performance?
And then thirdly, I remember you mentioning that in addition to Mexico, other factories will also produce the next generation of cars. Can we understand that it is possible for you to start this product in an existing plant before the new plant in Mexico is built?
Elon:
I think we may have to refuse to answer that question. We'll have a suitable product launch arranged, and if we answer your question, that's a spoiler.
Maybe one more question... Does anyone want to ask?
Martin Viecha:
Will，Will Danoff。
Will Danoff：
I have two questions, it was indeed an impressive afternoon, thank you very much.
Elon, I'm curious because you've doubled and will double again. Can you talk about how you're managing a larger business? What you need to do to manage a bigger business.
Second, I'm curious what you think, how generative AI, and all the rapid breakthroughs of the last few months, can help you make cars and make them easier to produce? Thank you.
Elon:
I don't think AI can help us produce cars in the foreseeable near future. At that point, none of us here will have to work anymore.
Unidentified persons:
This is a big problem.
Elon:
We just have to lie flat.
I'm a little bit worried about AI, and I think it's something we should focus on, and I think we should need some kind of regulator to regulate the development of AI and make sure it's working in the public interest, because it's a pretty dangerous technology.
I'm afraid I may have done something to hasten it up, and that is... I'm not sure.
AI technology is obviously useful, like the autonomous driving we're working on. Some believe that autonomous driving is a general artificial intelligence (AGI) type of problem. I don't think it's a general-purpose AI problem, but it certainly requires very complex neural networks because road systems are designed for eyes and biological neural networks. Naturally, this corresponds to cameras and digital neural networks.
If you look at the neural network architecture in our car, frankly, it's crazy. There is a network above the network, and there is a network above the network.
Ashok Elluswamy:
When you open this network architecture, even the visualizer crashes. It is so complex that there are currently no tools capable of visualizing it.
Elon:
Yes, it's hard to imagine, indeed, it looks crazy. I guess when we're driving, something like this is happening in our brains, it's amazing.
I think Tesla is doing good work on AI.
This question makes me nervous and I don't know what to think and say.
Martin Viecha:
Well, it's already seven o'clock, I think we have so much time, thank you very much for coming!
Elon:
Thank you!
(above)
![Yang Dayong, Vice President of Changan Automobile: Determined to build a new digital travel technology brand[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)