Welcome to we lecture 5, it's your host, Dylan Loomis, quick shout out to my newest patron, Manuel. Thank you for choosing to support the channel. The reason you will not see me on screen today, I did some hill sprints outside this afternoon and it's allergy season here and you can probably hear it in my voice. My eyes will not stop watering so just much easier not being on screen for today's video. TSMC just held its 2024 technology symposium and we actually learned some things about Tesla's dojo. TSMC revealed some new semiconductor technology and detailed some new chip packaging strategies and the latter could accelerate a trend toward processors made from more and more silicon such a system. Tesla's next generation dojo training tile is already in production. Now I have to say where they're saying such a system Tesla's next generation dojo training tile is already in production. They could just be referring to the first generation of dojo and for some reason calling it next generation. However that seems unlikely because the current dojo tile that's been in production that we've known about for a while was version 1 so to refer to that as next generation doesn't make a lot of sense but I just wanted to put that out there. And even if the next generation of dojo is not in production yet it will be eventually if he go back to the Q4 call last year Elon said we have plans for dojo 1.5 dojo 2 dojo 3 and whatnot.
And in 2027 TSMC plans to offer technology for more complex wafer scale systems than Tesla's that could deliver 40 times as much computing power as today's systems. For decades chip makers increased the density of logic on processors largely by scaling down the area that transistors take up and the size of interconnects but that scheme has been running out of steam for a while now instead the industry is turning to advanced packaging technology that allows a single processor to be made from a larger amount of silicon. And that idea is what TSMC is now calling its system on wafer. That's important because AI training is the obvious first application for wafer scale tech but it's not the only one and may not be the best. Another promising application is for data centers. Such a system could cut the number of advanced networks which is in a very large 16,000 rack data center from 4,608 down to just 48. The core of Tesla's dojo is the D1 chip which was designed by Tesla and yes the current generation is indeed manufactured by TSMC using the 7 nanometer process.
And listen I know Elon said dojo is a long shot but there's also a chance he was downplaying it publicly on purpose to of course maintain relations with Nvidia. And as we just talked about in a recent episode the word on the street is that Tesla's dojo is already accounting for about 21% of the training happening for Tesla's FSD. And now we learn that Tesla's next generation of dojo, whatever that may look like is already in production straight from TSMC. I'd also caution everyone is careful with comparing different metrics or compute capacity of a chip like the Nvidia H100 with dojo because Tesla's dojo is designed specifically to train FSD and video data. Additionally one of the defining characteristics of dojo is its extreme scalability.
Tesla told us it'll be an order of magnitude higher from an interconnected bandwidth than typical distributed systems. Here it is, it was optimized for neural network training workloads and built from the ground up with large systems in mind. So as Tesla's data collection scales with FSD rolling out globally that should in theory continue to benefit dojo. So basically with dojo the chip to chip and board to board communication that bandwidth should be much better than some of the off the shelf options. It's certainly encouraging to hear Tesla's dojo supplier planning to make big leaps forward potentially confirming Tesla's next gen dojo is in production and I'd also remind everyone Tesla's surviving at all was also a long shot and dojo does not have to be better than Nvidia at every head to head metric to make a real contribution given the specificity of the chip and further as it scales it keeps Tesla from paying those 75% margins to Nvidia.
Those add up quickly when you're spending billions of dollars each year on compute and as Tesla's data training grows and dojo adds more chips to the training stack it'll continually boost the efficiency of the training since that's how it was designed and that's really where it will shine.
Bill Gurley put forth what I think really should be some conventional wisdom saying VC investors would be thrilled to have all CEOs sign up for Elon and Tesla like comp packages that deeply align CEO interest with shareholder interest a.k.a. performance based compensation plans that way you don't have CEOs like Peter Rawlinson taking home nearly 400 million dollars in compensation when they haven't come close to profitability nor have they sold that many cars.
There were some reports today about Tesla's Megapack factory in China beginning construction we've already heard about this for weeks but as you can see from the latest video from WUWA it's really still just a muddy field construction has not begun yet. There may have been some minor prep work that's already taken place but for me I'm expecting production to begin sometime optimistically toward the end of quarter one 2025 but as you can see there's clearly still a long way to go so just be careful with all of those reports about construction beginning.
Joe Tettmaier reported it looks like a major media event for VIPs to formerly commission Tesla's Megapack at Giga Austin. This project will provide backup power for the factory at Giga Austin and will be able to support the grid as well.
Joe Tettmaier报告说,看起来像是一场为要员正式委托特斯拉在Giga Austin的Megapack的重大媒体活动。这个项目将为Giga Austin工厂提供备用电力,并且还可以支持电网。
Top Gear put out another video with a cyber truck this time they showed some new accessories this was just a prototype but it's a cooler that fits right in the front and through a software update the front sensitivity has been fixed and adjusted. Most of the video was talking about camp mode and the cyber truck tent but I learned that Tesla designed this tent in partnership with Hyne Planet. If you're not familiar they're a company who does higher end camping gear so I thought that was a cool fun fact. It was also nice to see Martin Vieja dunking on the industry talking about that video saying regular software updates for the win. How is this not an industry standard 12 years since Model S launched? Honestly though to Martin I would say because designing and building a computer on wheels with your own hardware and software stacks is really hard and requires a lot of talent from software engineers and it involves risk and most of the legacy companies are not built to do any of the above.
Top Gear发布了另一个视频,这次是与一辆电动皮卡有关,他们展示了一些新的配件,这只是一个原型,但是有一个可以完全嵌入前部的冷却器,并通过软件更新修复和调整了前部的灵敏度。视频大部分是在讨论露营模式和电动皮卡帐篷,但我了解到特斯拉与Hyne Planet合作设计了这个帐篷。如果你不熟悉,他们是一家生产高端露营装备的公司,所以我觉得这是一个有趣的事实。看到马丁·维埃哈在评论视频时批评了这个行业,说定期的软件更新胜出。为什么12年过去了,Model S推出以来,这还不成为行业标准呢?但老实说,对于马丁,我要说,因为设计和构建一辆带有自己硬件和软件堆栈的轮式计算机确实很困难,需要来自软件工程师的很多才华,并且涉及风险,大多数传统公司没有能力做到这些。
S&P Global came out and reaffirmed Tesla's BBB credit rating so nothing has changed. Now it is true that last summer Elon said why ESG is the devil and he quote posted from S&P Global. Tobacco companies are crushing Tesla in the ESG ratings. So yes I'm fully aware that S&P Global when it comes to ESG ratings is an absolute scam. However this is different when it comes to credit ratings there are sometimes things we can actually learn.
They said we expect Tesla's EBITDA margin to remain strong over the next two years. Tesla appears likely to stem further declines for the remainder of 2024 with better manufacturing efficiency. Tesla's strong liquidity provides its sufficient financial flexibility though we still expect free operating cash flow to decline over the next two years relative to 2022 and 2023 levels. And yes there is a difference between free cash flow and operating cash flow. So the easiest way to think about that free cash flow is after a company's capital expenditures. For Tesla's case in quarter one they spent a billion dollars on that AI infrastructure. Operating cash flow on the other hand does not account for any capital expenditures throughout the year so it's really just focusing on the actual business operations. Tesla provides its own chart for this. The blue is the operating cash flow and the red is going to be free cash flow. As we talked about after the earnings call Tesla maintaining 200 million dollars of operating cash flow even despite all of the one-offs in quarter one and the inventory build is an impressive feat. Back to S&P despite Tesla's expected decline in operating cash flow Tesla's cash flow metrics will remain better than most investment grade automakers through 2025 with nearly 27 billion dollars in cash and with their estimates for positive cash flow for the rest of 2024 and solid access to the ABS markets we believe Tesla will maintain strong liquidity. Tesla's cash balances remain well above our established thresholds which are around 15% of sales for Ford and GM to peers that contend with higher industry cyclicality.
They said we revised our management and governance score too moderately negative from neutral. That's to reflect a lack of board effectiveness, key man risk and exposure to an uncommon high level of contingent liabilities or lawsuits relative to the company's size and peers. The last thing I wanted to highlight they said though none of these lawsuits or investigations so far have had any material impact on Tesla's brand or strong financial condition. But yes if you wanted to say Tesla's credit rating doesn't really matter because they don't plan to access capital and the debt markets that's totally fine but again the overall outlook financially for Tesla from S&P is quite strong. And keep in mind that certain funds or investors will only invest in investment grade companies.
I also wanted to highlight that S&P report and the free cash flow versus operating cash flow so when you see charts like this about BYD's free cash flow going negative for quarter one you can say to yourself well hey maybe BYD actually had a large capital expenditure during quarter one that caused this to go negative just like happened to Tesla. So my point with all of this is on the surface to look at a free cash flow number going negative and then to argue well the sky is falling for whatever company it's just not enough analysis because you have to go further and figure out were there any capital expenditures during the quarter. But don't misunderstand me here I'm not criticizing AJ in the comments he did point out BYD's CapEx for Q1 but I've seen plenty of people throwing around these negative free cash flow numbers for BYD and Tesla without any context about CapEx and it's just not the right way to do it.
In case you saw this article from Reuters talking about Tesla retreating from the next gen giga casting process which was supposed to be one giga cast for the whole vehicle it's mostly a nonsense article. The trouble here is we don't know how far Tesla ever got toward committing to a one-piece casting in the first place. I'm sure they were exploring and talking to suppliers but that doesn't ever mean they actually committed to it. I also saw a bunch of people criticizing Reuters because they were talking about Tesla using a three-piece cast the front the rear and the middle and while it's probably more accurate to say to just for the front and the rear if you go to Tesla's investor day you can see on the screen they are talking about the unboxed process having a front rear and other casting they said the step back on giga casting occurred last autumn.
I did think it was worth sharing though because there is a chance that Tesla has explored the single giga cast and realized that maybe the upfront investment just is not worth it from a cost savings perspective and engineering it would be too difficult so yeah ideally would Tesla have a single cast for the next gen platform maybe but you have to ask the question at what cost.
Bloomberg said Tesla's top HR executive Ali Arabolo has left the company at this point it's not clear if she was fired or if she left on her own. I'm just speculating here but if she was the individual tasked with letting a lot of these people go maybe she just didn't want to deal with it. Just remember with Tesla saying they're laying off more than 10% maybe up to 20% of staff over the coming days and weeks will actually find out which departments have been reduced.
Stopping by Ali's LinkedIn page she had worked at Tesla for over six years in HR. It's probably a good time to travel back in time to look at something Elon said six years ago. Difficult but necessary Tesla reorganization underway my email to the company has already leaked to the media here it is. Here's the beginning of it Tesla has grown and evolved rapidly over the past several years which has resulted in some duplication of roles and sub job functions that while they made sense in the past are difficult to justify today.
In the next part of the email he called it a company wide restructuring. So here's proof this has absolutely happened before at Tesla however it's not all great as Elon did say in that email I also want to emphasize we're making this hard decision now so that we never have to do this again. Well that was certainly a bit over optimistic. As part of Tesla's spring update the auto park features are getting some updates to begin it no longer takes up the entire left portion of the screen. They're also saying that for model SNX if you're doing a K turn it'll be able to automatically shift to drive or reverse on its own. The word is this is also in development for the model 3 performance so hopefully it rolls out to other models as well.
For the last few years Amazon zooks has been totally in the shadows very few updates. Now we're learning that by the end of this year they're expecting zooks to begin a commercial driverless service. It's not a done deal but they said the experience could start with a robot taxi service offered as part of Amazon's prime membership program. To date zooks has been providing ride healing services for employees and it plans to launch that commercial robot taxi service in Las Vegas this year. There have been some limitations in terms of what circumstances zooks can operate in and NHTSA is still examining whether the company's robot taxi conforms with the FMVS standards.
Tesla stock closed the day at $179.99 down 1.8% while the NASDAQ was down 0.33%. It was a lower volume day trading about 13 million shares below the average volume the past 30 days. Hope you guys have a wonderful day please like the video if you did you can find me on X linked below and a huge thank you to all of my patreon supporters.