Welcome to Electrified, it's your host, Dylan Loomis. Quick shout out to my newest patron, Kerry. Thank you for choosing to support the channel. CERNBasher on X had a great framing of Tesla's regulatory credit sales, and no matter what happens with these in the future, they are very real right now. Since inception, Tesla has spent a total of $5.1 billion on compute, both spending on AI infrastructure and computer equipment, hardware and software. Meanwhile, over just the last three years, Tesla has received a reg credit revenue of $5.5 billion, which is how you draw the conclusion Tesla's competition has paid for Tesla's entire compute spend. And at the core, it's really that compute capacity, both the hardware and the software that is eventually going to allow Tesla to turn its current fleet of nearly 7 million vehicles into ongoing profit generators for all of its vehicles that will eventually be added to the Tesla RoboTaxi network that Tesla will take a portion of those profits from.
And as you'll see toward the end of the video, there's a reasonable chance these reg credits stay part of Tesla's business for years to come. If you're not already following CERN, I would encourage you to do so. I'll have his X profile linked below. Sawyer wasn't touched with the Tesla owners of Kentucky, and the word is that soon, Kentucky will finally allow Tesla to sell vehicles direct to customers in the state. The post said, after several, back and forth talks between Tesla and the Kentucky Motor Vehicle Commission were excited to announce Tesla will soon be offering direct sales in Kentucky. We don't know when this will go into effect, but when it does, no more need for any of these prospective Tesla owners to have to drive a few hours out of state to pick up their vehicle thanks to outdated franchise laws.
The poll for Tesla to invest in XAI has closed, and Elon responded to it saying, looks like the public is in favor, will discuss with the Tesla board. But if you've been following along, you may remember late last year, Elon said the same thing, will discuss with the Tesla board when Jason DeBolt and Dave Lee were talking about a partnership between Tesla and XAI. Just wanted to say there are no guarantees this leads to anything, and it definitely doesn't mean it's going to be in the next few weeks. And when you factor in what the Tesla board just got done dealing with for the majority of the first half of this year with Elon's compensation plan case in Delaware, they may want to proceed with a bit more caution, maybe a bit more pushback on Elon's ideas, and at the very least, going a lot further to actually fully inform all of the investors for any vote that would happen.
Personally I'm a lot more excited about a Tesla and XAI partnership or investment from a 3-6 year time period than I am for the next 3 years. The main reason for that is the valuations for some of these AI startups are certainly debatable and that boils down to the fact that they're spending a ton of money on compute but there still hasn't been a lot of proving it when it comes to being able to earn money. The word is open AI may be doing a few billion in revenue per year but what type of profit are they doing from that still a question. Elon has said before that the CapEx required for compute alone is going to be a few billion dollars a year just to be in the game. The hardware, the software, the cooling, the data centers, the energy use, it's not a cheap business.
This Mike P on X who historically has been very critical of Tesla's FSD really changed his tune with 12.5. He said that was by a long shot the most incredible FSD experience I've ever had. The rest was not family friendly but I really just wanted to highlight what Elon said, still some issues with this release but you can see the magic already. We're pushing it out only to those who have requested advanced releases for regular users 12.5.3 is probably suitable for wide release. Replying to Chuck Cook, Elon said we have yet to unlock the full potential of hardware 4 that may enable up to a further 8x increase in parameters. I know many hardware 3 owners out there are starting to get antsy feeling like they're going to get left in the dust.
Personally I would not be there just yet. I would wait at least a few weeks to see how the Tesla AI team can optimize things and as we've heard the plan is still to get 12.5 working for hardware 3. And trust me the last thing I'd ever want to do is to throw salt in anybody's wounds right now but I will be honest in that this is the one hardware item that I personally think it's sometimes worth waiting for if you're considering buying a Tesla. Full disclosure, getting hardware 4 was a significant part of the reason I waited even longer to eventually get a Tesla. And I know not everybody has the luxury of waiting but I did want to get that out there for the future as we head toward AI5 chatter just something to keep in the back of your mind if you have the ability to wait. One last encouragement if you do have hardware 3, don't forget it's in Tesla's best interest to make sure that all of those hardware 3 vehicles will eventually be capable of becoming a Robotaxi. That's a significant part of the fleet so I'm very confident Tesla will do everything in its power to make sure that can happen for hardware 3 as well. It's just looking like it may take some extra time to do so.
This right here should be arguments 1 through 3 for the Tesla needs to advertise camp. Alex said I can't believe some people don't believe in FSD to which Elon said most people have no idea it exists. If this is what Elon really believes then why not spend a few million dollars at a minimum to make some ads showing the benefits of FSD. Elon said technically actually SmartSummon is separate code from 12.5 but it'll be bundled with 12.5.x next month. Something tells me if this works well and reliably it's going to become a very powerful marketing tool so many videos will end up going viral. The reactions of people alone in the parking lots when they look into a Tesla and realize there's nobody in the driver's seat they should be priceless. I'm not sure how Tesla will handle it from an accounting standpoint but actually SmartSummon and Banish are some of the last features for the full FSD suite. Which means releasing those features may allow Tesla to recognize more of that FSD revenue upfront rather than deferring it.
We've been talking about deferred revenue on the energy side but don't forget on the auto side mainly it's from Tesla's FSD and at the end of quarter 2 that balance sits at 3.66 billion dollars. And of that balance Tesla expects to recognize 940 million dollars of revenue in the next 12 months. James Downa said 12.5 has started going out to model 3 vehicles as well but as far as I know it's still only AI4. Which again if you missed yesterday's video was hardware 4. To James Downa Robert was asking what type of parameter counts will hardware 3 be able to handle. James said Tesla probably is not near the theoretical limit of performance it takes crazy amounts of software development to get there. Hardware 3 is too recent and most performance enhancement is not linked to parameter count anyway. Small models can have big performance hardware 4 can get to higher performance with less development so it can get there sooner but hardware 3 has a long way to go before it becomes impractical to keep improving performance on it. Simply put Tesla may need to spend some extra engineering time and resources on optimizing 12.5 for hardware 3 but as I said I think they're still heavily incentivized to make that investment. But as we like to say here patience will likely be required.
To those of you stuck on 12.3.6 I really do feel for you I do now have 12.5 and have been able to test it for about 4 or 5 hours so far. As I was expecting it's really good but it's not perfect however there were significant and noticeable improvements for me from 12.3.6. I'm just going to show you two quick clips from my first drive. Oh looks like this road is actually closed I don't know if it's going to realize that there's cones over there. I don't see a sign though. I also really like testing this out but I had 3.6 long enough to know what it was about. And there are cones and it probably would have driven right through those cones.
Oh boy I wonder if people are going to say you disengaged too early. I kind of want to try that again because huh yeah I think I'm going to try that again. But oh okay so yeah it's ignoring the navigation which would basically take it to the right. It has it doing a u-turn there which obviously doesn't need to this is a shorter out so I got to be ready for this one to see if it's going to recognize these cones this time or not so this is going to be attempt number two. This time it took to turn a little bit wider and did get around them I didn't do anything. So I was ready to go.
Took the turn wider now let's see what happens though because it might go straight through this lane you're not this is not even a lane you're not supposed to be doing this. So what do you guys think the second time around did it just get lucky that it happened to take that turn a bit wider or did it actually learn from the first disengagement just a few minutes before. And just to show you in case you missed it the first time this right here is not actually a lane this is a no drive zone this lane is right turn only. But with 12.3 this was a problem every single time here's 12.5.
So you can see if we go straight this lane turns into a left turn lane only for the highway. So it's it should get over to the right to go straight but there's cars to my right so this should be interesting. I'm guessing it's just going to again go straight through this left turn lane which is illegal. Or now it's trying to turn right and it does at the last minute but now it would always get in this right turn lane this is a right turn only lane and it needs to be in the left lane and it's not but it is actually getting over this time before it would just drive through those white lines without a signal.
So that was better than before not perfect but I'll take it. I kind of want to go try that again so that's exactly what I'm going to do but he's staying behind us and this time it's getting over earlier so that's good. So that's very promising now this turns into a right turn only so it should be getting left right now and it's not and again it's in this right turn only lane but it's realizing it now signaling and getting over so that is definitely a significant improvement from 12.3.6. Overall though it is indeed noticeably smoother than 12.3.6 and I was only out today with traffic for about an hour or so but I had zero disengagements there are still times when I'll press on the accelerator to get it up to speed and then it usually does hold the speed there.
I'll be honest with what was happening with 12.4 I thought there was a world where maybe Tesla got stuck learning how to train these nets for a few weeks or months but with 12.5 that fear for me is mostly gone and very encouraged with this release and I know I know it's a very small sample size but I'm considering taking a small road trip this weekend just to put it through its paces. The ability to accelerate your Cybertruck delivery for 30,000 referral credit points is now back. JD Power just put out its 2024 appeal study and you know how I feel about JD Power but just hitting the highlights. It stands for Automotive Performance Execution and Layout. It measures how well each new vehicle provides positive experiences with design, performance, safety, usability, comfort, perceived quality and beyond. This is for new vehicle owners after 90 days of ownership. This study has been around for decades and they usually get between 80 and 100,000 responses. On a 1000 point scale for the premium category Porsche, Jaguar and Land Rover led the way, the premium average was 870 points. Tesla is still not officially rank eligible but their score would have been right in line with the average at 870. Vivian was 900 and Polestar was 839, the lowest on the list. In the summary though listen to what they said for full EVs, recent launches from traditional manufacturers have surpassed perennial leader Tesla when it comes to owners level of emotional attachment and excitement with their new vehicle. Their takeaways across all fuel types, emotional satisfaction is highest among owners of non-Tesla EVs with a score of 877. Tesla remains strong for satisfaction among its loyal buyers but is lackluster with conquest buyers. Newer EVs are achieving higher appeal scores due to improved battery range and better interior materials.
I'm speculating here but we know these traditional automakers are selling full EVs in very low quantities. It's likely still the very loyal either Ford, GM or Stellantis traditionalists who will like any vehicle that automaker puts out. Just like Tesla has its core group of raving fans, believe it or not, other companies do as well. It's just now with a volume Tesla is doing of course, some of those customers aren't necessarily going to be hardcore Tesla fans, it's more of your average consumer. I'm not at all trying to make excuses for Tesla, they're not a perfect company and they're not for everyone, just something to keep in mind. Yeah and I think this is like the complications of modeling this out are one of the reasons why it's probably gone unnoticed. So we've got this really big increase in deployments, quarter over quarter, which translates to the huge increase in revenue but it's not just that, it's this $1.4 billion increase in revenue quarter over quarter plus the $1.8 billion that Bradford highlighted that's going to be kind of future revenue. So I think what a lot of analysts might be doing is simply taking the total energy revenue, let's say this was $3 billion, dividing that by the deployments and saying okay, Tesla selling for $321 per kilowatt hour. Meanwhile, we know that their average pricing has been more like $450, $4, $64, $70 per kilowatt hour. So the fact that you've got this kind of big hole in your ASP is actually very supportive of the fact that you have this big component of the ASP, which is tied to the UPL. If you'd like to learn more about Tesla energy, I highly encourage this video from Matt Smith, Bradford Ferguson and Tesla Larry on the Rebellion Air channel, also on YouTube. And if you don't know what Tesla UPLs are, I would go back and watch yesterday's video. Tesla is continuing to expand its footprint in and around Austin this time they're planning to lease a 183,000 square foot building that is accessible via train. The site is near an area that Tesla has used to ship its finished models by rail. One of these times, I'm fully expecting one of these buildings or leases to be the first Robotaxi hub where they come to charge, get cleaned, etc. The article said it's unclear how Tesla will use this location. By the way, I'm not predicting I think this will be that Robotaxi hub, but it's coming at some point.
Previously, Volvo said it would not sell a single car globally after 2030 that would not be fully electric. Now they're saying it's going to take time to bridge different parts of the world for full electrification. A Volvo dealer said Volvo has gotten way out over their skis with this EV-only strategy. This dealer must be listening to electrified. In the first half of this year, deliveries of Volvo's electrified vehicles were down to 1,900 down 74%. The dealer said Volvo now realizes it cannot be successful without plug-in hybrids and mild hybrids. Stellantis still has some of the highest days of supply numbers in the industry. If you look at the US, this is a spot where our days of sales are still over 90.
Tesla is at 18. Stellantis, which has been hanging its hat on profitability, had its net income fall 48% in the first half of this year. Its operating margin also dipped below its goal of 10%. Margins declined most significantly in North America and now they're going to reduce production in North America. If you missed it, GM is now delaying production of its electric pickups at its Orion plant until mid-2026, pushing him back roughly half a year. GM is also delaying Buick's first EV for the North American market. That vehicle was likely to come from China and thanks to the new tariffs, that's likely why it's at least delayed they did not give a new time frame so I won't be surprised if it doesn't come at all.
This week, the CEO of Stellantis said the auto industry is in turmoil. To remember what Elon said on X, he believes that obliteration is coming for the auto industry. We can't forget that true inflation levels and higher interest rates make things very difficult for all automakers. Ford stock got crushed today, I'm really only interested in its Model E division which once again lost $1.1 billion for the quarter. Its EBIT margin is still negative 99% and as you can see, they're really not making that much progress yet. Tesla is slowly becoming less and less reliant on the auto industry but for now they still have significant exposure to that industry which is indeed very cyclical.
本周,Stellantis 的 CEO 表示汽车行业正处于动荡之中。让人想起Elon在X上说的话,他认为汽车行业将面临毁灭。不能忘记的是,真实的通胀水平和更高的利率让所有汽车制造商的状况非常艰难。今天福特的股票大幅下跌,我真正感兴趣的只有其Model E部门,但该部门本季度再次亏损11亿美元。其息税前利润率(EBIT)仍为负99%,如你所见,他们还没有取得太大进展。特斯拉正逐渐减少对汽车行业的依赖,但目前他们仍在很大程度上暴露在这个确实非常周期性的行业中。
GM's 2025 Cadillac Optic is supposed to be eligible for the full tax credit, it's supposed to come later this year. The CEO of X-Pang that just tested 12.3.6 here in the States said that while an end-to-end approach is best to reach level 2 or level 3 driving, it may lack what it takes to reach level 4. He said recently I talked about end-to-end with several level 4 bosses and they all believed that end-to-end is the best route for level 2 and level 3 driving but it's definitely not the preferred choice for level 4. I really think that end-to-end and large models will eventually lead to level 4. I tried to get some sort of beat or translation on what he could mean by large models being paired with an end-to-end network but I came up empty. Just a quote to store away for the months ahead.
A judge has ruled that Rivian will go to trial over allegations that ex-Tesla employees stole trade secrets for Rivian. Rivian's motion to dismiss was denied. This case dates back to 2020 and we should have an official trial date set this week. Tesla stock closed the day at $220.25 up 1.97% while the Nasdaq was down 0.93%. Today Ford was down 18.3%. It was a lower volume day for Tesla trading about 17 million shares below the average volume in 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.