You actually largely cover semiconductors, and you're sort of focused on this work around AI that Tesla is doing. There was something that jumped out at me in the 10-Q that the CFO also mentioned, which was in the quarter gone, and Tesla recognized more than $300 million worth of FSD revenues. Now take that and tell me the story it gives you. Yeah. The story it gives me is that they released FSD on the cybertruck. So for cars that they've been selling for a little while, they can certainly recognize the revenue that they've already been paid for for that technology that they've already delivered and continue to deliver. So that's great. And, you know, look, I think FSD is an amazing technology, but, you know, if you've read my reviews or if you look at any of the many reviews on YouTube or other platforms, you can see that there

are known failure modes for this technology. It does not work in the sense that you can actually run it without a driver, and that continues to be a problem no matter what the 10-Q says. This is a real problem for the company. Have you seen any improvements, William, since last you were on to talk about some of those failings within FSD? Yes, we've reviewed three different versions of this. I think the I forget, which was the last time I was on your show to discuss it. But certainly this version, 12.5, which is the latest one that's broadly available to consumers, this is still one that doesn't work. Have there been improvements? Certainly, they mostly relate to the feel of the car in that it feels more humanlike, less herky jerky. But along with that, I think there have been some documented cases of rolling through stop signs and other, let's say, minor driving infractions that are more

human-like, but maybe not exactly what one is shooting for for a driverless car in terms of trying to maximize the safety. Okay. Not only William does he need to improve on the technology, therefore, it seems like you're still saying, despite the improvements. And what's so notable is they need the regulatory approval. This is where you drill in the extraordinary nature of one Elon Musk, because he's not only the CEO of Tesla, but he's also potentially going to have a political role and FSD regulation coming. Ed Yes. William What I wrote about last night was Musk did not cite the election directly and he did not name Donald Trump. But what he did say was that if there's going to be a Department of Government efficiency, he would like to work on nationwide or federal level rules for the deployment of autonomous technology on US roads, and which is as close to commenting on the election as I think you could get in the context of an

earnings call. What did you make of those comments? I think he's just making some fairly logical comments about what would be the simplest, most logical way to pardon me to regulate this technology. And at the federal level means you have one hurdle to clear instead of 50 individual ones. I don't interpret that as some, you know, super partisan view of the world. It's just something that would make his life easier and the life of Tesla easier. That's that's all I take. That is I think it's worth looking at as Uber and Lyft again. Right. Because the news for me that jumped out was this confidence that next year Tesla will offer some form of ride hailing in Texas and California. You've critiqued FSD. My question to you is based on those headlines that came out on Ride Hailing, are you better understanding the jump from FSD and consumer cars to this proprietary ride hailing service in the future?

So, a couple of things. First of all, I have a colleague that covers these ride-hailing stocks. I don't. But I think what should be relatively clear is these issues: a ride-hailing service and full autonomy are not the same thing. They are fairly separate. You know, this company could turn on this ride-hailing service, you know, assuming they get regulatory approval for it, without having FSD at all. And FSD is something that could come later and really has. Yeah. In a way that they're certainly complementary. And having both of them together could result in significant earnings growth for the company. But they're not perfectly linked ideas. They are separate. With William, the amount that they need to spend. I mean, whether or not they turn on that switch or not. And of course, that ultimately means whether you have a driver at the wheel or not. I'm interested as to whether you think

the upgrade. What is it to an 11 billion CapEx spend? Is what needs to be invested right now by the business. Oh, well, this is not a huge difference. Right? In the prior quarter, they guided this number to 10 billion. Now he's saying 11. Wasn't a whole lot different from our model. We have a similar CapEx number next year, I think, in a way. You know, this might be a sort of off-the-run remark, but I think the biggest winner out of this topic is Nvidia. You know, Musk emphasized how much they're spending on each 108, 100. Those are hopper generation chips from a video equivalent use and what they anticipate doing over the next year. This is a big investment in NVIDIA. These are this training capability does not come from their internal D1 chip or their dojo computer. It's going to be spent at least in the near term, at least for the next year. More spending with the NVIDIA.

Truist's Stein: Tesla FSD Is Amazing Tech But Has Failures

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.