Latest / Elon Musk Podcast / Tesla Dojo: A Computational Marvel

Transcript

Tesla Dojo: A Computational Marvel · Elon Musk Podcast

0:01
Hi everybody and welcome back to the Elon Musk podcast where we
0:05
discuss all things technology, space, entrepreneurship and
0:09
more. And today we're talking about
0:11
something that's definitely caught the attention of Wall
0:13
Street Tesla shareholders and a I enthusiasts alike.
0:17
Tesla stock is climbing higher is something to do with the very
0:21
own supercomputer at Tesla Dojo. Recently, Morgan Stanley
0:26
upgraded Tesla stock from equal weight to overweight and hiked
0:30
the price target from $200.00 to $400.00.
0:34
But this optimistic outlook largely stems from their view of
0:38
Dojo's immense potential, not just in advancing Tesla's A I
0:42
capabilities, but also as a service that could be offered on
0:46
its own. Now, Dojo's goal is to
0:48
effectively process millions of terabytes of video data captured
0:53
from real life driving situation from Tesla's plus 4 million
0:57
cars. Now this goal led to a
0:59
considerably different architecture than conventional
1:02
supercomputer designs and going live in July 2023, Dojo serves
1:07
as the training ground for Tesla's machine learning models
1:10
aimed at bolstering its full self driving system or FSD.
1:14
And unlike conventional supercomputers, Dojo is
1:17
architected to handle millions of terabytes of real life video
1:20
data from over 4 million Tesla vehicles on the road right now.
1:26
Now this unique task called for an architecture that broke away
1:29
from traditional supercomputer designs, focusing on efficiency
1:33
and scalability. Now, before Dojo, Tesla was
1:36
already a formidable name in computing.
1:38
The company had several parallel computing clusters, one of which
1:41
was, by some metrics, among the top supercomputers globally.
1:45
However, there were debates about its actual performance due
1:48
to unclear metrics. These test the GPU clusters were
1:52
pivotal in training neural networks to better understand
1:55
real world driving conditions, and with MOJO, Tesla aims to
1:59
supersede its own lofty standards.
2:02
Mentioned first in 2019 during Tesla's Autonomy Investors Day,
2:07
MOJO faced some blaze due to power and thermal challenges.
2:11
Nevertheless, it officially came into the limelight in 2021 and
2:15
moved into production use in 2023, equipped with 10,000
2:20
NVIDIA H100GP U's to enhance his performance.
2:24
Now Dojo's reception among analysts is mixed, yet mostly
2:27
positive. And while some say it's
2:29
impressive but won't change the supercomputing game, others call
2:33
it an inhouse developed game changer that can massively speed
2:36
up the development of autonomous vehicles.
2:39
Now the endorsement from Morgan Stanley is particularly
2:41
significant, suggesting that Dojo could be a catalyst for an
2:45
unprecedented increase in Tesla's market value.
2:48
The idea is that Tesla's focus on custom designed a I chips for
2:52
video training could afford a formidable advantage in the
2:55
autonomous vehicle sector and beyond.
2:59
At Tesla's A I day in 2021, the introduced Dojo, but at the time
3:04
they were just getting started. They had a first chip and some
3:07
training tiles, but we're still in the process of constructing a
3:11
full Dojo cabinet in a cluster. Fast forward to a I day 22.
3:15
They made considerable headway, including developing a full
3:19
system tray, and they're aiming to ramp it up to a 100 exaflop
3:24
capability by the end of 2024. This is particularly significant
3:29
as Tesla's ambitions becoming increasingly computer
3:32
constrained, especially in selfdriving endeavors.
3:35
Moreover, Tesla's feature doesn't just rest on car sales.
3:39
Supercharging network with partnerships like the one
3:42
they've received established with Hilton Hotels presents
3:45
another potential revenue stream.
3:47
Adam Jonas believes that the advantages and the advancements
3:51
and chip technology will add to this by possibly extending the
3:54
self driving licensing and operating system licensing.
3:59
Now let's talk numbers for those who love them.
4:01
Morgan Stanley's price target shift from 2:50 to 400 isn't
4:05
trivial. The analyst broke down the value
4:07
and factored in Dojo's impact on the overall valuation.
4:11
This was a pivotal enough shift to lift Tesla stock by nearly
4:15
10% following the notes release. And get this, aside from Tesla,
4:19
Morgan Stanley also sees the possibility of non Tesla fleet
4:23
licensing revenue, which means the reach of Tesla's
4:26
technological advancements could go beyond its own lineup of
4:30
electric vehicles. And while the core business of
4:33
Tesla has always been tied to electric vehicles, this new
4:36
focus on Dojo is a revenue generator isn't really a wild
4:40
shift. In fact, we've mentioned this
4:42
before. The Dojo could potentially be
4:44
offered as a service, argumenting its utility and
4:48
contributing to Tesla's overall value.
4:51
So the concept isn't new. But what is notable is how
4:54
mainstream financial analysts are now seeing Dojo not just as
4:58
a little side project for Tesla for their own vehicles, but as a
5:01
key aspect of Tesla's future growth and the value.
5:05
Now. One more thing before we go
5:07
There are other market forces at play.
5:09
Currently, there's a looming potential of a massive UAW
5:13
strike that can negatively affect Tesla's competitors.
5:16
This labor strike in competing companies could inadvertently
5:20
give Tesla a huge edge, further fueling optimism around Tesla
5:25
stock and future prospects. That wraps up the show today.
5:29
Hope you found it informative. If you enjoyed the show, don't
5:32
forget to hit the subscribe or the follow button on your
5:34
preferred podcast platform. It's absolutely free, takes a
5:38
second, and helps us out tremendously, and this way
5:41
you'll never miss out on any Elon Musk news.
5:44
O thanks for tuning in and please take care of yourselves
5:47
and each other and I'll see you in the next one.