Tesla FSD v12: Breakthrough We've Been Waiting For?

Hey everybody RobMower here, welcome back to Tesla Daily today. Of course we are going to be talking about FSD Beta version 12. We've also got updates on hardware 4, project highland, Cybertruck, and a few other items as well. Tesla stock starting off the week in green just barely up 10th of a percent to close at $238.82 while the NASDAQ was up 8-10th of a percent on the day to day.

大家好，我是RobMower，欢迎回到《特斯拉日报》。当然，今天我们要讨论的是FSD Beta 12版。我们还会向大家更新硬件4、高地项目、Cybertruck和其他几个项目。特斯拉股票在本周开盘时微微上涨了0.1%，收盘价为238.82美元，而纳斯达克指数当天上涨了0.8%。

Alright FSD Beta version 12 as promised Elon did due to livestream for version 12 on Friday, of course right after Tesla Daily had ended, but we'll catch up on that today. He did also participate in brief twitter space prior to that livestream starting. A few interesting details from that that we'll talk about, but let's just start off with the FSD Beta livestream itself, which was a hardware 3 vehicle running FSD Beta version 12 in Palo Alto for about 45 minutes, with Elon just picking random destinations and dropping pins and seeing how the car performed.

根据承诺，埃隆发布了FSD Beta 12版本，并在周五通过直播展示了该版本，当然是在Tesla Daily结束后进行的，但我们将在今天补充这方面的内容。在直播开始之前，他还参加了一个简短的Twitter Space。从那段时间中得到了一些有趣的细节，我们将讨论其中的一些，但首先让我们来谈谈FSD Beta直播本身。直播中使用了一辆硬件3的车辆，在帕洛阿尔托行驶了大约45分钟，埃隆随机选择目的地，设定目标点，看车子的表现如何。

My reaction and the reaction I'm seeing from others I think this was ahead of what we probably expected for FSD Beta version 12 at this point. Not perfect, there was one disengagement and another time where it sounded a turn way too long and most people probably would have intervened for that. But in general, seemed very steady, seemed very comfortable, Elon remarked on that a number of times just how comfortable and smooth it was. And although it probably wasn't the most challenging FSD Beta drive that's ever been attempted to say the least, there were challenging scenarios with construction, with roundabouts, with pedestrians.

我的反应和我从其他人身上看到的反应告诉我，在当前的FSD Beta 12版本中，它超出了我们的预期。虽然不是完美的，有一次出现了人工干预的情况，还有一次转弯时间过长，大多数人可能会干预。但总的来说，它显得非常稳定、非常舒适，埃隆多次提到它有多么舒适和流畅。虽然这可能不是有史以来最具挑战性的FSD Beta驾驶，但也有一些具有挑战性的情景，比如施工、环形交叉口和行人。

And although this was Palo Alto and you can see Tesla's driving around constantly, so Tesla's got a lot of data here, the routes did seem to be random, so not pre-planned, which is nice to see. So overall, a really nice drive, probably pretty similar to what we might see from 11.4.7, which at this stage I think would be great for version 12, as obviously it's still an unreleased version and there are reasons for that. But again, probably a little bit closer to being ready for that than what maybe we would have expected going into it.

虽然这是在Palo Alto，你可以见到特斯拉不断开车，因此特斯拉在这里有很多数据，但路径似乎是随机的，没有预先规划，这很不错。总的来说，这次驾驶很不错，可能与11.4.7版本差别不大，这对于12版本来说应该是很好的，因为显然它还是一个未发布的版本，有其原因在。但是，可能比我们预期的更接近准备好发布的状态。

Most impressive to me though, rather than the drive itself, was the commentary that Elon provided during the drive, particularly noting that in the current version, there are over 300,000 lines of code, referring there to either entirely or primarily C++ programming, relying on a lot of heuristics that Tesla programs in to make decisions on how to drive. And as we know, with version 12 Tesla trying to remove that and go fully end-to-end neural network from perception to vehicle planning, basically going from a raw photon input to vehicle control output with essentially only neural networks in between.

对我来说，最让我印象深刻的不是驾驶本身，而是埃隆在驾驶过程中提供的解说。他特别指出，在当前版本中，有超过300,000行代码，主要使用C++编程，依赖许多启发式算法来决策驾驶方式。而且，正如我们所知，在版本12中，特斯拉试图将其移除，并完全实现端到端的神经网络，从感知到车辆规划，基本上是从原始光子输入到车辆控制输出，中间仅使用神经网络进行处理。

So 300,000 lines of code Elon in this drive said that this version has quote unquote basically none of that. Although we knew Tesla was heading in that direction with version 12, this to me is one of the areas where things were ahead of my expectations. I figured okay, Tesla will head that direction, maybe it's a multi-step process where pieces of it happen at different times, with Elon saying that there's basically none of that, maybe there still is a little bit of that happening, but largely, he said that those 300,000 lines of code basically just removed, meaning this version of version 12 is already pretty much end-to-end.

所以埃隆在这次驱动中说，这个版本基本上没有那些30万行的代码。虽然我们知道特斯拉在12版本中朝着那个方向发展，但在我看来，这是其中一个超出了我的预期的领域。我想，好吧，特斯拉可能会朝着那个方向前进，也许这是一个多步骤的过程，其中的一部分在不同的时间发生，埃隆说没有那个，也许仍然有一点这样的情况，但他表示，这30万行的代码基本上已经被移除了，这意味着这个12版本已经基本上实现了端到端。

That's extremely exciting, especially if the capability of this version 12 is already similar to the version 11.4.7 that we've got now, which although it ends up looking pretty similar on a live stream, this is just a fundamentally completely different architecture from current version. Elon specifically said that all of this is just a neural network making decisions based off of the weights of the neural network based off of the training video data that Tesla has supplied to it. There's no programming of speed bumps, but it slows down. There's no programming to pull over to the side of the road into destination, but it does and it parks there. There's no code to tell it what a roundabout is or how to drive through it, but it does and on and on and on throughout this entire drive.

这真是令人非常兴奋，尤其是如果这个版本12的能力已经类似于我们现在拥有的版本11.4.7，虽然在直播中看起来很相似，但实际上它与当前版本完全是基于不同的架构。埃隆特别表示，所有这些仅仅是一个神经网络根据神经网络权重以及特斯拉提供的训练视频数据做出决策。它没有编程来处理减速带，但它会减速。没有编程告诉它要靠边停在目的地，但它确实这样做，并停在那里。没有代码来告诉它什么是环岛或如何行驶通过它，但它确实这样做，并且在整个驾驶过程中一直如此。

Like this is pretty mind-blowing stuff, think about it from the perspective of Tesla's vehicle perception, and we've known that that's used neural networks for a long time. Think about if that had to use specific lines of code, how many lines of code Tesla would need to just even interpret the scene as opposed to a neural network that's been trained already on millions of those scenes. When you think about it from sort of that image analysis perspective, it should be so clear why the neural network is so much more efficient, faster, more effective, and probably most importantly can be improved more quickly than trying to hard code that type of an interpretation.

这真是令人震惊的东西，从特斯拉的车辆感知的角度来看，我们已经知道它使用了神经网络很长一段时间了。想象一下，如果不使用神经网络，特斯拉需要多少行代码才能解释场景，而神经网络已经在数百万个这样的场景上进行过训练。当你从图像分析的角度来思考时，很清楚神经网络为什么更高效、更快、更有效，并且可能最重要的是能够更快地改进，而不是试图编写那种类型的解释的硬代码。

So it's a little bit more intuitive when you're thinking about images or video. It's just applying that same sort of logic to decision making now rather than perception. Not as intuitive, but that is the same fundamental shift in architecture that we are now seeing for planning and control of the vehicle with FSD beta version 12.

所以当你考虑图像或视频时，它会更加直观一些。现在，我们将同样的逻辑应用到决策制定上，而不是感知。虽然不太直观，但这正是我们现在在FSD beta 12版本中看到的车辆规划和控制的基本架构转变。

And taking that line of thought even a bit further, not only will this fundamentally change the decisions that the Tesla, the vehicle itself, is making, but also its interpretation of planning and control of other actors in the scene, whether it's vehicle, pedestrian, whatever else, those things are also all going to be influencing the neural network and the output from it to actually control the vehicle.

更进一步思考，这不仅将从根本上改变特斯拉车辆自身所做的决策，还会影响其对场景中其他参与者的规划和控制的理解，不论是车辆、行人还是其他任何事物，这些都将影响神经网络和其输出来实际控制车辆。

So this really is mind-blowing stuff and again, completely different than current architecture, but you look at the video, it kind of looks the same. I think most people just say, Oh yeah, another Mafestee Beta version, you know, there's still a disengagement. It's not driving around downtown Manhattan during rush hour, etc, etc. But again, all that is failing to recognize how truly different this is.

所以这真的是令人震惊的东西，而且与目前的架构完全不同，但你看视频时，它看起来有点相同。我想大多数人只是说，“哦，是又一个Mafestee Beta版本，你知道的，仍然存在分离感。它还没有在曼哈顿市中心的交通高峰期四处行驶等等。”但是，再次强调，这种看法忽视了这种差异的真正程度。

So anyway, hearing that that is really the stage that Tesla is at right now, that was the most exciting part for me. And my hope, which I think is at least somewhat rooted in reality, is that this architecture should allow Tesla to improve the rate of progress on FSD. And Elon did say in the live stream that he feels good that Tesla is now at a place where they have a very rapid and virtuous cycle of updating the weights of the neural networks with high quality video data. That may now be the biggest challenge with FSD is acquiring that data. Tesla's very well positioned to do that, refining that data, making sure that it is truly high quality data, and then training on that data, which we know is going to require a lot of compute and Tesla working really hard on that.

无论如何，听到特斯拉目前正处在这个阶段，这是我最兴奋的部分。我希望，我认为这至少在一定程度上是现实根基的，这个架构应该能够让特斯拉提高全自动驾驶（FSD）的进展速度。埃隆在直播中确实说过，他对特斯拉当前的状态感到满意，因为他们能够以极高质量的视频数据更新神经网络的权重的速度非常快，这应该是特斯拉目前在FSD中面临的最大挑战，就是获取这些数据。特斯拉在这方面有很好的定位，他们会不断优化这些数据，确保数据质量真正高，然后在这些数据上进行训练，这我们知道需要大量的计算资源，特斯拉正在努力实现这一点。

So we've talked a lot about the compute constraint. I think that's pretty well documented. The refining to get high quality data, that's been a little bit less of a focus Tesla has talked about things to automate that at AI day. But that is an area that is a part of the progress path and now becomes a bigger part that is just kind of worth keeping an eye on listening to the things Tesla has to say about that when we consider how things develop from here.

所以我们已经详细讨论了计算约束的问题。我认为这已经有着相当完善的记录。为了获取高质量的数据，优化仍然是特斯拉关注较少的部分，他们在AI Day上提到了自动化的事情。但这是进展道路上的一部分，现在变得更重要了，因此值得留意特斯拉对此的看法，以了解事情从这里如何发展。

Now some more context I want to pull in here. Elon did as I mentioned do that Twitter space before this FSD beta drive. He mentioned on that that Tesla is probably going to spend around $2 billion on training this year alone. And he said they'll probably spend similar or probably even more next year, a decent amount of that should be hardware. So that may last a while. But that's a pretty decent amount of Tesla's expenditures. Said that they are going to continue buying Nvidia alongside trying to ramp dojo. Said that they are doing some training on dojo already. We did know that previously. Hopefully that will ramp up your significantly. And then also mentioned that they are going to be starting today actually a new 10,000 unit cluster of H100s from Nvidia.

现在我想再提供一些背景信息。正如我之前提到的，埃隆（Elon）在进行FSD beta测试驱动之前，在Twitter上进行过一次其它的讨论。他在那次讨论中提到，特斯拉今年可能会花费大约20亿美元用于培训，明年可能会花费类似甚至更多的资金，其中相当一部分将用于硬件。这可能会持续一段时间。但这是特斯拉支出的相当大一部分。他说他们将继续购买NVIDIA的产品，同时努力推进Dojo。他还提到他们已经开始在Dojo上进行一些培训。我们之前就已经知道这一点了。希望这将显著提升效果。他还提到，实际上他们将从今天开始使用由NVIDIA提供的一批新的H100集群，总计10000个单元。

If we go back to the graph that Tesla had posted about getting to 100 exa flops of compute over the next year or so, they put this in units of A100. So I'm not sure for Tesla's purposes how they would compare the H100s to the A100s. But it should be at least two times. Potentially more feel free to share your thoughts on that. But if we assume that's equivalent to 20,000 A100s, this should at least be doubling Tesla's compute capability from just a few months ago.

如果我们回顾特斯拉发布的有关在未来一年左右达到100艾克赫计算能力的图表，他们将其以A100单位表达。所以我不确定特斯拉在设计上如何将H100与A100进行比较。但至少应该是A100的两倍。有可能更多，欢迎分享你的想法。但如果我们假设这相当于20,000个A100，那么从几个月前开始，特斯拉的计算能力至少增加了一倍。

It's even more exciting now that we know more about version 12 and can see that it should be able to take even better advantage of that additional capacity. And as we previously talked about, things are going to continue to scale extremely quickly for compute for Tesla from this point. Tesla's head of AI infrastructure tweeted about this saying Tesla AI 10K H100 cluster go live on Monday. Due to real world training, we may have the largest training data sets in the world hot tier cache capability beyond 200 petabytes orders of magnitudes more than large language models join us and quote another member of the Tesla AI team shared a similar thought and also noted that version 12 and then training is compute bottlenecked.

现在我们对第12版本有更多了解，可以看出它应该能更好地利用额外的容量，这让人兴奋不已。正如我们之前谈到的，从现在开始，特斯拉的计算能力将会以极快的速度不断扩大。特斯拉的AI基础设施负责人在推特上发文称，“特斯拉AI 10K H100集群将于周一投入使用。”由于实际训练，我们可能拥有世界上最大的训练数据集，热层缓存能力超过200PB，比大型语言模型更大数个数量级。特斯拉AI团队的另一名成员引用了这段话，并提到第12版和训练过程存在计算瓶颈。

So extremely exciting times for FSD for Tesla AI right now. I know sometimes the progress and the consumer product can feel a little bit slow, but I do think there's a lot of fundamental progress happening behind the scenes that isn't always immediately apparent. I mean, this drive is a great example of it.

现在对于特斯拉AI的全自动驾驶技术来说是非常令人兴奋的时刻。我知道有时候进展和消费产品可能会感觉有点慢，但我确实认为在幕后有很多根本性的进步，这并不总是立即显现出来。我指的是，这次的驾驶就是一个很好的例子。

Oh yeah, another disengagement same old FSD. But that is definitely not the case. I'm actually a little bit glad that there was an intervention because obviously Tesla community can hype things up at times. It was good for us to see that this is still a work in progress. There's going to be a lot of work to do still from this point. But hopefully that work is being done on the right architecture. And there's a reason that Tesla's going to spend, you know, forward $5 billion on this over the next couple of years.

哦，又一次取消参与同样古老的全自动驾驶系统。但事实并非如此。对于有干预的事情，我其实有点高兴，因为特斯拉社区有时会过分炒作。我们很庆幸看到这仍然是个正在进行的工作。从现在开始，还有很多工作要做。但希望这项工作是在正确的架构上进行的。特斯拉要在接下来的几年内投入至少50亿美元在此上面，这是有原因的。

There is still a lot of work that needs to be done and that can be painfully slow, especially if expectations are high. So we've got to keep that in mind as well. But a lot to be excited about. I would also say that that point on the amount of money that Tesla's investing in this, it's a great reminder why automargans today, they're really overemphasized. Not that they're not important.

还有很多工作要做，而且可能会进展缓慢，特别是当期望值很高的时候。因此，我们必须牢记这一点。但有很多令人兴奋的地方。我还想说一下特斯拉在这方面投资的金额，这是一个很好的提醒，为什么当今的汽车巨头被过分强调。并不是说它们不重要。

We've talked about this, but what's most important is that Tesla is able to fund these types of projects, fund the increase in scale that the business is going after. So that both with autonomy, with electric vehicles, and with the projects that Tesla has on the future roadmap that they can bring those to life. No one today or in 2021, when the stock was going way up, cared about what Tesla's margins were on the model aspect in whatever random quarter in 2012, none of that stuff matters at all once the new business lines take over.

我们已经谈论过这个问题，但最重要的是特斯拉能够为这些项目提供资金，并支持他们追求的规模扩大。这样，无论是自主驾驶、电动汽车还是特斯拉在未来计划中的项目，都能够实现。在当下或者在2021年，当特斯拉的股票价格不断上涨时，谁会关心特斯拉在2012年某个季度的型号方面的利润率呢？一旦新的业务线占据主导地位，这些都将变得无关紧要。

Those things were important at the time for Tesla to get where they are and just kind of an overall sense of health of the business, not saying it's not important. But oftentimes those things get easily overemphasized. So just a good reminder of where Tesla is heading.

这些事情对于特斯拉在当时取得现在的地位以及企业的整体健康状况非常重要，这并不是说不重要。但是这些事情往往容易被过分强调。因此，这只是一个提醒我们特斯拉正朝着何方发展的好机会。

A couple other interesting details, then we'll move on. Elon did say that FSD version 12 is being tested all around the world. He mentioned specifically in New Zealand, Thailand, Norway, Japan, but other countries as well. And it sounded like international expansion is a relatively high priority item for Tesla. So I think many people will be happy to hear that.

接下来还有几个有趣的细节。埃隆确实表示，全自动驾驶（FSD）的12版本正在全球范围内进行测试。他特别提到了新西兰、泰国、挪威和日本，当然还有其他国家。而且听起来，特斯拉对国际扩张相对而言是一个较高优先级的事项。因此，我认为很多人会对此感到高兴。

And then Elon did have a couple of comments on hardware for said hardware for software will lag hardware three by at least another six months as our focus needs to be on getting FSD on hardware three working super well and provided internationally. So definitely disappointing to hear that for hardware for vehicles. Elon did reply to a comment from dirty Tesla on X saying that this really would be something more like six months rather than something like two years.

然后埃隆对于硬件有一些评论，他说硬件对于软件来说会落后至少六个月，因为我们的重点需要放在使硬件三上的全自动驾驶功能能够在全球范围内运行上。所以对于汽车的硬件来说，这无疑是令人失望的。埃隆还回复了Dirty Tesla对于X车型的评论，表示这个时间大概是六个月而不是两年。

And that maybe it would be less on the Twitter space he mentioned that they just had to do a lot of retraining because of how the sensor inputs come into the neural networks. Also an interesting tidbit on hardware for he mentioned that hardware three is all being done on computing hardware in the vehicle that's only drawing 100 watts of power. Obviously impressive. We have known that though, but on hardware for he said that that will draw a little bit more power he didn't specify how much. But interesting to hear that Tesla's decided that those trade offs are worth it from a little bit more power hungry hardware perspective.

也许在Twitter空间上，他提到，他们不得不进行大量的重新训练，因为传感器输入如何进入神经网络。另外一个有趣的细节是他提到硬件三是在车辆的计算硬件上完成的，只消耗100瓦的电力。显然令人印象深刻。我们已经知道这一点，但关于硬件他说，那会消耗更多的电力，但没有具体说明数量。但有趣的是听到特斯拉决定这些权衡从更需要更多电力的硬件角度来看是值得的。

All right, moving on from FSD we've got a couple quick updates on Highland and Cybertruck really quickly on Highland. There have been a few more sightings coming out of China of Tesla testing model three vehicles that are covered in the front and the back. I don't think we've got too many new rumors over the last few days but more of these sightings. So we'll keep an eye on that. And for Cybertruck of course more of these sightings a lot on transport trailers. Seen a lot of that already but Ezekiel Overstreet posted on X a drone shot of 13 different Cybertrucks in the same frame in the parking lot at Giga Texas. And of course that's after some have already been shipped out.

好的，离开FSD，我们快速更新了一下Highland和Cybertruck的情况。关于Highland，中国出现了几次更多的特斯拉Model 3测试车的目击报道，这些车辆的前后部被遮盖。过去几天里，我不认为我们有太多新的传闻，只是更多的这些目击报道。所以我们会继续关注这个情况。至于Cybertruck，当然还有更多这样的目击报道，特别是在运输拖车上。我们已经看到了很多这样的报道，但Ezekiel Overstreet在X上发布了一张无人机拍摄的Giga Texas停车场里同时出现了13辆不同的Cybertrucks的照片。当然这是在一些Cybertrucks已经被运出之后的情况。

So pretty high number to be seeing all in the same place. There's also a newly wrapped version of the Cybertruck that we're seeing this one rather than being an F-150 wrap seems to be a Toyota Tundra wrap. So it looks like Tesla continuing to have a little bit of fun there and maybe antagonize some of their biggest competitors in that segment.

在同一个地方见到这么多印度的非常高的数字。此外，我们还看到了一辆全新包装的Cybertruck，与F-150的包装不同，看起来更像是一辆Toyota Tundra的包装。所以看起来特斯拉继续在那个领域玩一些小把戏，或许在与他们最大的竞争对手之间制造一些对抗。

Next we've got a great update from Joe Tagmyer. He made his way out to the Corpus Christi Tesla lithium refinery location. Of course the location that Tesla just broke ground on earlier this year. So we get to see a little bit of the progress. Definitely seems to be coming along pretty quickly and also gives us a nice point of reference for hopefully some more updates on this in the coming months.

接下来是Joe Tagmyer给我们带来的重要更新。他前往了科珀斯克里斯蒂特斯拉锂电矿砂精炼厂的所在地。当然，这个位置就是特斯拉今年年初刚刚开始动工的地方。所以我们能够看到一些进展。这个项目似乎进展相当迅速，也为我们希望在接下来的几个月内得到更多更新提供了一个很好的参考点。

Next we've got an update on Megapack. I think this was from Friday but some of your merit noting on X that Tesla has dropped the price of one Megapack from about $1.9 million to about $1.5 million. So about a 22% reduction. This does seem like a big price cut. However it's worth noting that Tesla did have different prices based on the number of Megapacks being ordered. So across the entire quantity spectrum I'm not sure what the average price change would be. Tried to look your way back machine but it didn't work. But may not be a significant for larger quantities I'm not sure. And then it's probably also worth noting that although the price previously would have indicated something like $479 per kilowatt hour. If we look at the quarterly filings Tesla hasn't really been earning close to that on energy sales at least in terms of what's being recognized.

下面是关于Megapack的最新消息。我认为这是从上星期五得到的消息，但有关于X公司的一些要注意的事项，特斯拉把一个Megapack的价格从大约190万美元降低到大约150万美元。所以大约降低了22%。看起来确实是一个很大的降价。然而值得注意的是，特斯拉根据订购的Megapack数量有不同的价格。所以对于整个数量范围，我不确定平均价格的变化是多少。我试图在网上查找之前的数据，但没有成功。但对于较大数量来说可能并不重要，我不确定。另外还值得注意的是，虽然之前的价格可能表明每千瓦小时479美元左右。但如果我们查看季度报告，特斯拉在能源销售方面实际上并没有接近这个数额，至少在被认可的范围内。

Last quarter for example Tesla deployed 3.65 gigawatt hours of energy storage which at $479 a kilowatt hour would be over $1.7 billion in revenue but Tesla's total revenue reported for energy including solar was only $1.5 billion. So I don't know exactly what causes that difference whether that's driven by bulk orders or revenue recognition or something else. But because of that I think it's difficult to assess exactly how big of a price cut this may end up being in reality. Something we'll definitely keep an eye on and then it looked like Tesla also shorted the estimated delivery windows. I think that's more likely to be a reflection of good news in Tesla's production plans than bad news in their order rate. Although obviously interesting to see that paired with a lower price in the design studio which could suggest more of the latter. But at the same time price decreases could correspond to increased production rates and lower cost of goods sold per unit from that increased production. So could either be really good news here or not so great news but because energy has been doing so well I lean more towards the optimistic take.

举个例子，上个季度特斯拉部署了3.65千兆瓦时的储能产品，按每千瓦时479美元计算，这将带来超过17亿美元的收入，但特斯拉报告的整体能源收入，包括太阳能在内，只有15亿美元。所以我不确切知道导致这种差异的原因是大宗订单、收入确认还是其他问题。但正因为如此，我认为很难准确评估这种定价下调实际上会产生多大的影响。这是我们一定会关注的问题，此外，看起来特斯拉还缩短了预计交付时间窗口。我认为这更有可能是特斯拉生产计划方面的好消息，而不是订单速度的不佳消息。尽管在设计工作室里价格下调与此同时出现，这可能暗示了后者的情况。但与此同时，价格的下降可能与生产速度的提高和单位产品的销售成本降低有关。所以，这可能是一个好消息，也可能不那么理想，但由于能源的良好表现，我更倾向于乐观的观点。

Alright last couple of things here we've got an updated note from Alex Potter of Piper Sandler. They have reiterated their overweight rating and $300 price target for Tesla stock after a visit to Gigaboolet. I don't think there's anything on production rates or anything like that but a couple of the key takeaways just in general from the note that I thought were interesting. Piper Sandler says that relative to price cuts adds maybe a cheaper way to boost orders and that tests are ongoing. So enough that information would be coming from Tesla or just their observation of some of the things that we have seen. They also note that rising EV inventory may be due to Tesla's price cuts with other brands unable to compete. And on Berlin they mentioned that Berlin based production of battery cells is still on hold because of the implications of the Inflation Reduction Act from North American battery production but that the buildings are ready. So we've talked about that before but something to continue to keep an eye on.

好的，最后还有几件事情。我们收到了Piper Sandler的Alex Potter的最新消息。他们在参观了Gigaboolet后，重申了对特斯拉股票的超配评级和300美元的目标价格。我不认为有关生产速率之类的信息，但是我认为这份报告中几个重点的观点很有意思。Piper Sandler认为，与价格削减相比，提高订单可能是一种更便宜的方法，并且测试仍在进行中。因此，这些信息可能是来自特斯拉，也可能只是他们对我们所见到的一些情况的观察。他们还指出，电动汽车库存的增加可能是由于特斯拉的降价，而其他品牌无法竞争。关于柏林，他们提到，由于北美电池生产的通货膨胀减缓法案的影响，柏林的电池单体生产目前还停滞不前，但建筑物已经准备就绪。所以我们之前谈论过这个问题，但继续关注是很重要的。

Alright last item for today just a quick look at the calendar. I don't think we looked at this on Friday so a lot of economic reports coming up this week. I want to particularly highlight the PCE Personal Consumption Expenditures report on Thursday morning. The Fed likes that one a lot and we're already almost into September here and looking forward to the next FOMC meeting in mid-September. So we'll see how that one comes in and how expectations for the September meeting change off of it. Then also of note here you may see the GDP release for Q2. For some context on that GDP does get released three times and this is the second release so not the final not the first but rather an updated preliminary number for Q2. The first time around this came in at 2.4% so see if any major changes on that but probably not quite as notable as the first GDP release.

好的，今天最后一个项目，快速查看一下日历。我认为我们上周五没有看过这个，所以这周将有很多经济报告。我特别要强调周四早上的个人消费支出报告。美联储非常喜欢这个报告，我们已经快进入9月了，期待着9月中旬的下一次FOMC会议。所以我们将会看到这个报告的结果以及对9月会议的预期将会有怎样的变化。此外，你可能还会看到第二季度的国内生产总值（GDP）发布。为了提供一些背景，GDP会有三次发布，这是第二次发布，不是最终版本也不是第一次发布，而是第二季度的一个更新的初步数据。第一次发布时为2.4％，所以看看是否有任何重大变化，但可能不像第一次GDP发布那样引人注目。

Alright that is it for today then as always thank you for listening make sure you're subscribed and signed up for notifications. You can also find me on the X at Tesla podcast and we'll see you tomorrow for the Tuesday August 29th episode of Tesla Daily. Thank you.

好的，今天就到这了，一如既往地感谢你的聆听，请确保你已经订阅并开启了通知。你也可以在X上找到我，我们明天再见，等待8月29日星期二的特斯拉日报节目。谢谢。