Welcome to Electrified, it's your host Dylan Ummus, quick shout out to my newest patron, Ron R. Thank you for choosing to support the channel. Recurrent Auto just published a study talking about the impacts of DC fast charging on battery degradation. They studied these effects on over 12,500 Tesla vehicles in the US. They compared cars that fast charge at least 90% of the time and also cars that fast charge less than 10% of the time, so on either end of the spectrum. The results show no statistically significant difference in range degradation.
Looking at the Model 3, a data set of 6,300 vehicles, whether you fast charged frequently the blue line or rarely the green line, you ended up essentially in the same place after 2000 days, right around 10 to 12% battery degradation. And for the Model Y, a sample size of 4400 vehicles, you end up either between 8 or 9% of the battery degradation after 1000 days. If you're new to the EV world, some of their advice was avoid fast charging with your battery super cold or super hot without pre-conditioning. So if you're going to a supercharger, plug it in the nav and then your battery will do that pre-conditioning on its own. So this study does a good job of showing us that it's not really fast charging or slow charging that does the damage, but we already knew it's really the ends that matter, so if you run your battery down to zero all the time or you constantly charge to 100, that might be where you can increase the battery degradation.
In North Carolina, Duke Energy Utility is rolling out an interesting pilot program for electric vehicle owners. Duke Energy has partnered with GM, Ford and BMW. This pilot will allow residential customers in North Carolina to use up to 800 kilowatt hours per month to charge an EV at home for a fixed monthly fee of $20 or $25 depending on location. And they're saying in these regions the average EV driver uses less than 15 kilowatt hours on any given day so that monthly limit should be plenty enough. This one year pilot program launches November 1st this year. It sounds like the two main goals of this program would be one predictable rates for these EV owners and then two ensuring that these EV owners are not charging during peak hours if they can avoid it and there will be some features like scheduling your charging sessions as well.
Today's a big day as it has become official that hardware 4 Tesla vehicles are now getting FSD beta for the first time. I thought it was a joke for a few hours but it turns out many people are reporting it is indeed real. And let me just say upfront not everybody with hardware 4 is going to get this in the first wave so you may need to deploy some patience. And if you're thinking to yourself well what's going on here Elon just said hardware 4 software will lag hardware 3 by at least another 6 months. Yes that can still be true he never said that hardware 4 would never have FSD beta just that it's going to lag so maybe these hardware 4 owners get an older version of the FSD beta definitely not V12 yet and most likely not even 11.4.7 yet. But as I'm recording there are people all over the internet reporting different anecdotes of hardware 4 getting a version of FSD beta. And this is such a welcome site for all of those new hardware 4 owners that as of yesterday we're probably thinking they're now going to have to wait 6 months to a year to have FSD beta at all because they decided to buy a brand new Tesla. And it looks like that's not going to be the case. Tesla scope is saying it's packaged in the software 2023.26.11. So let's say hypothetically hardware 3 owners get version 12 later this year maybe hardware 4 owners don't see version 12 until summer 2024 there's that 6 month lag. But at least in the interim both hardware 3 and hardware 4 vehicle owners will have some variant of FSD.
During the 10 year anniversary of the first superchargers installed in Norway, Tesla Europe is now offering free supercharging in all of Europe today to all EVs. We got the Tesla China weekly insurance data and 17000 for Tesla. Plugging that data in for week 8 of quarter 3 if you wanted to compare to the same weekend quarter 2 that number was 12800. So looking at the first 8 weeks of quarter 3 we now sit at 89.2 thousand compared to the same first 8 weeks in quarter 2 where we were at 79.8 thousand so ahead of pace. Going all the way back to the end of October last year there were only 3 other weeks where the reading was higher than that 17000. So yeah it's just one week but it's the fourth best week in almost one year.
Especially ahead of the election it's always very interesting to watch how different news sources cover the same story. Spending is a big topic and with pal warning us about the possibility of higher rates of the 131 sources covering the story 17 lean left 17 lean right and 64 were center. And personally I've always found it very interesting to watch how different sources talk about anything Elon related and it's very easy to do so when you can just check the left headlines the center headlines and then right leaning source headlines.
The ability to do all of this is thanks to today's sponsor ground news one that many of you guys have really been enjoying. Every news story will give you a visual representation of the political bias the ownership as well as the factuality of different sources. As I've said before the blind spot feed has really been valuable to see which news stories are being disproportionately covered by one side of the aisle. And with the ground news vantage plan you can get access to everything they have to offer including the my news bias feature a dashboard that shows you how your reading habits change over time and what your top sources are. You can use my link below to sign up for free but you can also use it to get 30% off that vantage plan subscription. I think it'll help you digest the news more efficiently and with less bias because it has for me and I think that's a mission worth supporting in this day and age. Thanks to ground news for sponsoring this video.
I know this is tough to see but business Korea's formatting is subpar either way they said Samsung is said to be the favorite to win Tesla's order for hardware 5. We talked about this before but they're just reiterating what we've heard they're planning to use a for nanometer process for hardware 5 and they say Samsung is already supplying Tesla with a 14 nanometer process FSD chip they're actually referring to hardware 3 which was what replaced hardware 2 that was on Nvidia's drive PX2 platform and I'm pretty sure but not positive that hardware 4 is on a 7 nanometer process so hardware 5 looking to go even smaller which is more efficient and faster. And for anyone new Tesla is designing their hardware chips but then they're actually manufactured by companies like Samsung or TSMC people ask why doesn't Tesla just make or manufacture their own chips well just look into what it takes to build a fab or a fabrication plant and you'll understand why they don't do that.
And just touching on what Ross said about Tesla doesn't have anything like Nvidia's H100 chips I'm not ready to dive into it I am looking into it but for now I just want to remind people it's oftentimes more about the software that's running on the hardware especially when it comes to very specific applications like Tesla's full self driving training. So hypothetically if Tesla's dojo actually wasn't on par from a hardware perspective with the H100 from Nvidia because it's specific to what Tesla is optimizing to do it may be more effective in actually training Tesla's data set for what it's trying to optimize for. Given the magnitude of what this training ramp is going to mean for Tesla, Robotaxes autonomy margins all of it we cannot gloss over this H100 situation.
So here we are at the end of August Tesla was on track to have about 10,000 units of A100 GPUs online and training. Now if we assume that Tesla's recent addition as of yesterday of the 10,000 unit H100 cluster was not included in this graph when Tesla released the graph in the first place that means Tesla just added the equivalent of 100,000 A100 GPUs because Nvidia claims the H100s are 9 times more capable for AI training and I just rounded up to 10 for the sake of example. So if that really is the case that would bring Tesla's Nvidia hardware alone to the equivalent of about 110,000 A100 GPUs and then wherever dojo is that on the ramp process bringing that online.
And I still saw a lot of people over the weekend saying well the V12 drive wasn't that impressive I don't know what the big deal is. It really has nothing to do with the drive itself and everything to do with the underlying architecture that Tesla has moved to. So from here going forward Tesla should in theory be able to iterate and train at a much faster pace than it has to date. One because it removes all of that manual code and all of those heuristics and then two because it has much more compute power actually online with the H100 cluster and dojo ramping as well. And not to mention the fleet continually growing and testers globally operating gathering the best data which is part of why I'm now guessing that we see a robo taxi in operation somewhere in the United States quarter one of 2025. If not sooner.
Esther said project Highland has been spotted in Europe for the first time but I don't think that's accurate. You may recall earlier this month one was spotted in Spain. Despite that technicality this appears to be the first time in Germany.
Just in case any of you aussies are in New South Wales, Victoria, Southeast Queensland or South Australia. You may be able to get up to $3,800 off a Tesla power wall but only through September 15th. Part of the incentive is for joining origins VPP program but this will be below if you want to learn more.
Redwood materials just raised $1 billion in a series D and they plan to continue building capacity expanding the domestic battery supply chain and allowing customers to purchase battery materials made in the US for the first time a big milestone. Redwood has now raised nearly $2 billion of equity capital along with that $2 billion loan from the DOE.
In case you heard about Toyota's production system malfunction this week where there was a glitch in it prevented them from ordering components. Well that's apparently been solved they'll restart operations at assembly plants across Japan today.
Cruise is adding another city to its list this time around it's Seattle Washington where they will deploy a small fleet of piloted vehicles. This phase is just for testing and gathering and Seattle will join Charlotte, North Carolina, Atlanta, Georgia, Miami, Florida, Nashville, Tennessee and Phoenix, Arizona. Cruise is also doing testing in three cities in Texas, Austin, Dallas and Houston in addition to San Francisco. Cruise now estimates it's covering more than 1 million driverless miles per month and cumulatively has crossed 4 million driverless miles.
But think about it you have cruise with about 400 vehicles in total going to select cities gathering data manually versus Tesla with 4 million vehicles in its fleet gathering data globally actually in more countries than in cities. And look I want cruise and Waymo to succeed the more options for customers the better but to anybody that thinks cruise and Waymo are playing the same game as Tesla in my humble opinion they're just uneducated on what's really going on.
BYD's management has a target of selling 3 million EVs this year so maybe about half of those are full BEV which would put them around 1.5 million full battery electric vehicles for 2023. BYD's gross margins are holding steady between 15 and 20% this year but remember BYD does a lot more than just make vehicles. However don't forget on the vehicle front BYD has added to luxury EV brands Yang Wang and Feng Changbao which has vehicles that sell over the 1 million yuan mark which is about $140,000 or more than double some of its earlier higher end vehicles. So just keep in mind right now outside of the super high end luxury stuff BYD has almost all of the price bands covered.
It looks like Elon and Zuckerberg may be in a room together on September 13th for an AI meeting. This will be the first of Senator Chuck Schumer's AI insight forums and the strategy is to give Congress more influence over the future of AI. Exactly what everybody wants to see.
This publication charged, interviewed more than a dozen executives, engineers and analysts from automakers, DC fast charging network operators and charging hardware firms and other businesses.
Overall this year's developments reflect deep dissatisfaction among automakers other than Tesla with the state of US fast charging. And they said it's hard to overstate the disgust and anger at Electrify America among virtually every person we interviewed. The network has come to be viewed as the most minimal effort VW group could have exerted to comply with their 10 year $2 billion settlement from dieselgate. It's no surprise EA has steadfastly refused to discuss its reliability statistics.
One Tesla automakers have had it with EA. Initial hopes that EA would provide a new large scale nationwide network of fast charging stations have now curdled into a desire to see EA out of the game altogether with lots of bad blood directed at the VW group.
特斯拉汽车制造商对 EA 忍无可忍了。最初希望 EA 能提供一个全国性的大规模快速充电站网络,但现在他们完全希望 EA 退出这个游戏,对大众集团满怀恶意。
Tesla aside, all networks are perceived to be more focused on getting new stations in the ground and photo ops with politicians and government money than funding operations and maintenance. One person they spoke to said I have $18 million to build new EV chargers and $0 to fix the broken ones previously deployed. One engineer and one executive even suggested VW deliberately did a subpar job. Remember dieselgate? Fool me once shame on you fool me twice. Would it really surprise any of us if internally VW was out to sabotage the entire EV movement so they can just rest on their ice laurels? Sadly it really would not surprise me.
Interesting development in China, Sinopek, the country's top oil refiner is now predicting peak gas. It's now expecting gasoline demand in China to peak this year two years earlier than its previous outlook. For some context, in China, August data is telling us that plug-in vehicles have a 38% of new passenger vehicle sales. And they said well over 5% of the passenger vehicle fleet is now either BEV or plug-in hybrid. There are obviously many many many factors that go into peak gas but it does seem like we have another few years in the United States.
GM has partnered with Google to boost its AI for its on-star virtual assistant. Google Cloud's dialogue flow conversational AI tech is currently powering the on-star VA currently providing answers to more than a million owner questions monthly in North America. The software and services VP said generative AI has the potential to revolutionize the buying ownership and interaction experience inside the vehicle and beyond, enabling more opportunities to deliver new features and services. For now though, it's still mostly navigation or answering some questions about GM's products.
Ford Pro is rolling out some new commercial chargers both AC and DC, two options for each. The goal here is really to integrate seamlessly with Ford's existing telematics and charging software.
Ford Pro正在推出一些新的商用充电器,包括交流和直流两种选择,每种都有两个选项。这里的目标是与福特现有的远程监控系统和充电软件无缝集成。
Don't forget to check out ground news below if you're interested and if you opt for a paid option, you should be able to afford it pretty easily after today.
Hope you guys have a wonderful day, you can find me on X and Instagram linked below. Please like the video if you did and a huge thank you to all of my Patreon supporters.