Hey everybody Rob Nower here and today we've got some exciting new details on exactly how impactful Tesla's new H100 cluster is going to be for their capabilities. We've also got a couple updates on Project Highland, some news on Tesla battery research, and a few other items as well.
Alright starting off with the stock, not quite as exciting of a day for Tesla today, down 1 tenth percent closing at $256.90 while the NASDAQ was up half a percent today. As we mentioned when looking at the calendar, a couple economic reports that were published today, the revised estimate for GDP for the second quarter was published. That was revised down. So the advance release, the first release estimated GDP at 2.4 percent, GDP now coming in at 2.1 percent with this revised release. We should get the final number, I think that'll come in about a month or so. So not the final release, but interesting to see that revised down. Obviously, the lower that number is, represents worst things for the economy, but potentially impacts expectations for interest rates as well. And kind of paired alongside that with the GDP release was the Q2, PCE or personal consumption expenditures. Those were revised. So this is different than the PCE report that will get for July tomorrow. That'll come out before market open. That's the monthly indicator. This is the quarterly indicator. These get revised too, of course, and this, in this case, second quarter of PCE was revised up from 1.6 percent to 1.7 percent. So not an extreme revision a little bit higher though, but still just wanted to make sure that that's clear that that's not the PCE report, which will be for July that we get tomorrow, which is obviously more recent than the data that we're talking about here for Q2. So just want to make that clear.
And then I want to talk more about the H100 and video cluster that Tesla is turning on right now. Just a little bit about this over the last couple of days. So of course, Tesla is turning on their 10,000 H100 cluster. We talked before about how maybe with a conservative estimate, maybe that's something like two times an A100 equivalent, which is kind of how Tesla has framed their GPU capacity in the past for training capabilities. That was conservative. Some other estimates out there when as high as 9x, something like that. Elon has now clarified this, which allows us to get a much better understanding of how big of an advancement this is exactly for Tesla in their training capabilities. So Elon saying that in their testing, H100s are about three times faster than a 100s. So it's just something to keep in mind whenever you're looking at the benchmark data, obviously those kinds of things are going to depend on exactly what use cases are being tested for those benchmarks. And that's going to vary from real world as we're seeing here with Elon's tweet. Over the last three x, that's huge. Again, a little bit higher than my conservative estimate. So I'm glad to see this. And I want to put this in context for what we had talked about probably a month, month and a half ago or so now, which was the graph that Tesla shared about the increases in compute compute capacity that they are planning to have over the next 18 months or so.
So remember, we broke out the pixel math on this, each chunk of these 100,000 A100 GPU equivalent blocks was 69 pixels. At the beginning of this year, Tesla started at seven pixels of that filled in. So you've got roughly 10,000 A100 equivalents to start this year for Tesla. Now that's grown actually significantly, even though it hardly looks like it on this graph to the equivalent of 12 pixels right around the time that Tesla had published this, which wasn't equivalent of 17,000 A100 GPUs. Now that can come from a little bit of dojo, but probably mostly A100s at that point in time, Tesla then said that they were starting production of dojo, which was a part of what will drive Tesla's compute capacity up significantly over the next 18 months.
But as we have heard Elon say, part of that's going to be driven by Nvidia. And now it's more clear with the H100 cluster that they have talked about, probably what's driving the most significant part of the ramp up, you know, in this next three to six months.
So again, putting this in context with the H100 cluster that we now have, as Tesla said, that's about 10,000 H100s. And if those are about three times faster than the A100s, which means we're looking at that equivalent of about 30,000 A100s, if we kind of reference back to this graph.
So adding that to the 17,000 that we had roughly at the time that this was published, you know, again, there's maybe a little bit of dojo increase in there during that period of time. But going from 17,000 now ups you the equivalent of about 47,000 A100s with the addition of the H100s. So if we go back to then the beginning of this year, you're talking about now almost 50,000 compared to where Tesla was at about 10,000 to start the year.
So year to date already increasing their training capacity by five times what it was. That means that at the beginning of the year, something that would have taken Tesla three months to train is now only going to take them about two and a half weeks. Or said another way, something that would take a week to train. Now only going to take them about a day and a half.
Now, as we talked about with FSD version 12, this is going to be a big deal because that's really what's going to drive the progress, what's going to drive sort of the march of nines, at least if Tesla's architecture and plans come to fruition the way they want them to video training data is the constraint as Tesla has put it.
So the constraint to doing that is the computer hardware. And that's exactly what we're talking about here. Anytime you can increase your constraint from, you know, one X to five X like Tesla has done in the last eight months, that's going to hopefully lead to much, much faster progress.
So we're in an extremely exciting period of time right now. And again, a lot of these things aren't going to be immediately visible. But that is just a fundamental change in where Tesla's capabilities are. And hopefully the rate of progress that we then see in the actual consumer product. So obviously we'll have to wait and see if that comes eventually. But it's very exciting to see what what could be unlocked from this capability.
So I don't want to go on and on about this, but it's very exciting. And obviously this is just the beginning stages. Right now this actually already puts us kind of where Tesla had originally plotted out to be in sort of October in that period of time. So it looks like there may be even a little bit of head of schedule shockingly on this massive increase that they had planned.
Now, I don't know. Obviously they said go live for the H100 cluster right now. I don't know if that's going to take them any time to kind of ramp up and actually integrate to the training process. But if that is sort of a flip of the switch type of a situation, then, you know, we'd be looking at Tesla being a head of schedule on this ramp, not even considering maybe the impacts of dojo yet. Although this gives us more clarity that again, a lot of this is probably driven by the H100s.
So start of production for dojo that still could take, you know, a pretty significant amount of time to start to contribute to this growth trend. And maybe that's really what's, you know, driving the increases for next year. So again, exciting time and we'll continue to keep an eye on that. And I'm very thankful that Tesla shared these updates with us.
All right, moving on just a couple of quick updates or rumors, I guess, on Project Thailand. So Tesla Chan on X noting that according to local China, China news, Model three, highly on trial test production is over. So you can take that obviously different different ways, but presumably based on some other rumors that we have heard in the past, the trial production was, you know, the period of time before they would put it into full scale production, which again, according to some of those rumors seem to be targeted for September. There's more rumors, of course, of maybe some sort of an announcement on Friday.
So trial production testing trial test production ending before then could be a good sign that maybe that remains on track. So we'll see. I'm not putting all my hopes on any announcement on Friday, but just something to keep an eye on as we head towards that date.
And then just wanted to honestly just show this render. I think it's one of the best renders that we have seen of what Project Thailand could look like. Obviously, this is just a render, not a final photo by any means. But it looks great if, you know, if this is something, if it looks something like this, I think most people will be quite happy with that.
And then a drive Tesla candidate, a drive Tesla candidate to day reporting that these two sets of wheels are going to be the new wheels on the updated Model 3. So obviously these would be matte black versus, you know, right now we've got more of the aluminum wheels on the Model 3, at least in the United States.
And similar style here on the larger wheels, a little bit different in terms of, you know, what the arrow, uncapped arrow wheels look like. I think they've got a photo of those here to compare to. Yeah. So you can see pretty similar there, except for the coloring, and then a little bit more of a change on the arrow wheels. Presumably these ones would also have a cap, but I don't know, kind of tough to tell with what we're seeing here. So again, all of these I would put in the rumor category, but we seem to be starting to see quite a bit more information on Highland. All right.
Next we've got an update on Tesla's battery research. So as we've talked about many times in the past, Tesla partners with the Dollhousey Research Group in Canada, I guess the Don Research Group at Dollhousey University in Canada. So Electric Autonomy, who has had some information on various different topics about, you know, EVs and Tesla's conversations with Canada in the past, they visited the Don Research Group and had a really interesting article that I'll link in the description today. I would recommend reading it. We're not going to go through everything because it's pretty lengthy, but some pretty interesting insights that they shared. So in their conversations with one of the members of the team, he talked about a couple of samples, I guess four different samples of different batteries that they were testing at that time and said, you know, of one of these pouches that they were looking at, that's probably the best lithium ion cell that you can make today. So they've got a few different chemistries here, one of which is lithium iron phosphate, some nickel manganese cobalt, et cetera.
So the one that he was referring to there specifically, he highlighted a new ingredient, which would be silicon carbide. So losing my place on where that's mentioned, but yeah, there, I mentioned to highlight it, didn't quite get it highlighted. So silicon carbide, saying that that is helping drive the energy density. So of course, we've known about that for a long time. Tesla does plan to integrate silicon into 4680s. Silicon carbide here would potentially be a little bit different. But they talked about how, you know, again, this is his words, potentially the best battery in the world with him saying that the lithium iron phosphate then might take second place. Obviously, there's different characteristics, different use cases for all different types of batteries with the silicon battery. Obviously, what you got to kind of worry about with that is degradation. So he noted that it doesn't cycle as many times, obviously as a lithium iron phosphate battery would. But again, for Tesla's use cases, you don't need necessarily quite as many cycles, at least in automotive, as they talk about here with the nickel manganese cobalt battery, which you know, has sort of famously been talked about as quote unquote million mile battery in the past. And we've covered that probably three years ago at this point, maybe. But talking about the cycle life said that that is now up to 19,500 cycles and counting. If each cycle, you know, is full at 300 kilometers, you're talking about 6 million kilometers, which would be, you know, just shy of 4 million miles. So pretty crazy to kind of just see the potential life, lifespan of the batteries that are, you know, potentially being used in EVs today. So again, I'll put the link to this in the description again, a lot of other interesting details in that nice to kind of see some of that being shared.
And then the last few things for today, just a quick update on Tesla superchargers in Australia, the driven.io is reporting that Tesla has opened up nearly half of its superchargers in Australia. There's only 63 superchargers right now in Australia, but they've opened up about 30 of these. I think they were testing this with with five of them previously. So a pretty big expansion to non Tesla owners, supercharger access there in Australia.
And then the Wall Street Journal today with a report that the Department of Justice and the SEC are investigating Tesla over a secret, secret glass house project. This is something we talked about Wall Street Journal had a report on this probably a month ago or so, and that followed a Bloomberg report on this, I don't know, six months a year ago or so.
So we've talked about this a number of times. I don't want to go into the details because there's really no additional details that are shared this time around other than the fact that the DOJ and SEC are looking into it. The gist of it was that there was some special glass that was ordered for Tesla. And there's some communication that maybe that was for a house or something similar like that for Elon and just questions around the appropriation of those funds and what they were being used for.
So again, as we talked about when this initially came out, there was, you know, conflicting information on whether or not it was even for a house. And there's no new information in this at this point in time. The only new information again is that Wall Street Journal is saying that the DOJ and the SEC are looking into it. It doesn't mean that anything is going to come to it come of it. But obviously that always gets a lot of attention. So we'll see. I'm sure we'll have more reporting on that.
And then finally for today, just a quick update on Rivian. So Rivian has announced the range estimates for their new dual motor variant of the R1T still waiting on the R1S figures. But interesting to see these for the R1T. They're pretty good. So with their Max Pack and 21 inch wheels, the EPA range on the R1T is going to be up to 410 miles. Obviously that declines a little bit if you go with the smaller wheels. And then if you opt down to the large pack versus the Max Pack, that range drops a bit, of course, with a smaller pack to around 350 miles, again, depending on the exact configuration.
So it looks like pretty good numbers there for Rivian. Of course, the dual motor is using a new in-house developed motor from Rivian. So I think they're excited to get that out and that configuration and of course, the EPA estimates here. All right.
That'll wrap it up for today then. So as always, thank you for listening. Make sure you're subscribed and signed up for notifications. You'll also find me on X at Tesla podcast. And we'll see you tomorrow for the Thursday, August 31 episode of Tesla Daily. Thank you.