Latest / Elon Musk Podcast / Elon Musk Weekly News 7-23-2023

Transcript

Elon Musk Weekly News 7-23-2023 · Elon Musk Podcast

0:00
Good afternoon, everyone, and welcome to Tesla second quarter
0:02
2023 Q&A webcast. My name is Martin Vieca, BP of
0:05
Investor Relations, and I'm joined today by Elon Musk,
0:08
Zachary Kirkhorn and a number of other executives.
0:11
Our Q2 results were announced at about 3:00 PM Central Time in
0:15
the update deck we published at the same link as this webcast.
0:18
During this call, we will discuss our business outlook and
0:20
make forwardlooking statements. These comments are based on our
0:23
predictions and expectations as of today.
0:26
Actual events or results could differ materially due to a
0:28
number of recent uncertainties including those mentioned in our
0:31
most recent filings with the SEC.
0:34
During the question and answer portion of today's call, please
0:37
limit yourself to one question and one follow up.
0:40
Please use the raise hand button to join the question queue.
0:43
But before we jump into the Q&A, Elon has some opening remarks.
0:47
Elon. Thank you, Juan.
0:49
So just the Q2 recap. In Q2, we achieved record
0:53
vehicle production and deliveries and record revenue of
0:55
about $25 billion in a single quarter and Model Y became the
0:59
best selling vehicle of any kind globally in Q1 support
1:03
surpassing the likes of likes of Corolla and Gold.
1:06
So it was the number one vehicle of any kind including vehicles
1:11
that are at a seller farming price, which is I think an
1:15
incredible achievement by the Tesla team and just a huge thank
1:18
you to our customers of for their support.
1:23
So and this came in spite of high interest rates and a lot of
1:27
macro uncertainty and nonetheless we managed to
1:31
achieve operating margin of about 10%.
1:35
We continue to target 1.8 million vehicle deliveries this
1:40
year, although we expect that Q3 production will be a little bit
1:46
down because we've got some of shutdowns to for.
1:50
A lot of factory upgrades, so just just probably a slight
1:55
decrease in production in Q3 for sort of global factory upgrades
2:03
in the long term. Autonomy we think is going to
2:06
just drive volume through the ceiling next level and and our
2:13
so future Robitaxi products, dedicated ROBITAXI products.
2:19
We think have like quasi infinite demand so and it we're
2:28
the way we're going to manufacture the robot taxi is is
2:33
also itself a revolution. So it's revolutionary design
2:37
made in a revolutionary way. It'll be by by far the highest
2:43
units per hour of of any vehicle production ever so.
2:49
Very excited about that. With respect to Autopilot and
2:53
Dojo, in order to build autonomy, we also need to train
2:57
our neural net with the data from millions of vehicles.
3:01
With the more, I mean this has been proven over and over again.
3:04
The more training data you have, the better the results.
3:08
And I mean there's times where we see basically the inner
3:13
neural net, basically it's sort of at a million training
3:17
examples. It barely works at 2,000,000.
3:21
It's it slightly works at 3,000,000 it's like wow, OK
3:25
we're seeing something for they get like 10 million training
3:27
examples. It's it's like it becomes
3:29
incredible. So you just there's just there's
3:33
just no substitute for massive amount of data and obviously
3:37
Tesla has more vehicles on the road that are collecting this
3:43
data than all the companies combined by I think maybe even
3:47
an order of magnitude. So I think we might we might
3:50
have 90% of all or a very big number.
3:55
So you know the success in a I endeavors is function of talent
4:06
sort of unique data and computer resources and and we have
4:13
outstanding capabilities or all three arenas.
4:20
And I really just don't know how anyone anyone could do what
4:22
we're doing even if they had our software and had our computer if
4:25
they did not have the the the training data.
4:28
So Speaking of which our do training computer is designed to
4:33
significantly reduce the cost of neural net training.
4:37
It is designed to it's it's so optimized for the kind of
4:41
training that we need which is a video training.
4:45
So you know would. We just see that that the
4:49
needful neural net training again to give fear quasi and the
4:54
things is is just enormous. So I think having having we we
5:01
we we expect to use both NVIDIA and Dojo to be clear but there's
5:07
there's we just see a demand for really vast main resources.
5:16
And we we think we may reach in house the old net, the old net
5:20
training capability of 100 exoplos by the end of next year.
5:26
So to to date over 300,000,000 miles have been driven using FSD
5:32
beta. That 300,000,000 mile number is
5:35
going to seem very small very quickly, right it'll it'll soon
5:40
be billions of miles then 10s of billions of miles and.
5:46
FSC will go from being better for being as good as a human to
5:50
them being vastly better than a human.
5:52
We see a clear path to full self driving being 10 times safer
5:59
than the average human driver. So and between autopilot delta
6:06
computer our inferred hardware in the car which we calls.
6:14
Sort of hardware 3-4, you know but it's really dedicated, it's
6:17
a, it's a, it's a high efficiency in first computer
6:19
that's in the car and our optimist robot translates fairly
6:24
at the cutting edge of a had moment to with regard to our
6:29
side which we continue to build our release candidates of the
6:33
cyber truck on our final production line in Austin.
6:36
I'm actually here in Austin at the Eager Factory.
6:40
This is the first truck that we're aware of that will have 4
6:43
door was over a six foot bed and 1 foot into a 20 foot garage.
6:48
So it's it's a it's for the biggest on the outside but it's
6:51
even bigger on the inside. So it's a I think that's a one
6:55
of the one of the elements of good design is it should feel
6:58
bigger on the inside than it looks on the outside.
7:00
And and this is this is no small car but we we've we really cared
7:06
about the exterior dimensions of the side truck down to the last
7:08
millimeter. So they said just before you try
7:12
to get rid of the goalie block zone, not too big, not too small
7:17
and and then really maximize the the utility of the volume and we
7:22
can't wait to start delivering it later this year.
7:26
Some some other highlights, our global supercharging network now
7:30
stands at over 50,000, roughly 50,000 connectors and over 5000
7:35
locations as I think a lot of people are aware.
7:40
The Tesla starting the the Tesla charging standard which we made
7:45
open source is now called the North American charging
7:48
standard. We're we're deeply honored that
7:51
Ford, GM, Mercedes and many other Williams have signed up to
7:56
use our connector and gain access to our charging network
8:05
and we we we strongly believe in helping.
8:09
Other other car companies to accelerate the EV revolution and
8:14
just trying to do the right thing in general.
8:16
So that's up goal there Then it's something I think I want to
8:27
emphasize like very, very strongly.
8:30
This is a very important point is that Tesla just as with the.
8:38
The North American charging standard, although we're not
8:39
licensed and that case is not licensing, we're just making it
8:42
available. But but we are very open to
8:46
licensing apples of driving software and hardware to other
8:49
car companies. And we are already in
8:52
discussions with early, just early discussions with a major
8:56
OEM about using the Tesla FSD. So we're not trying to keep this
9:03
to ourselves. We're more than happy to to
9:05
license it to others. And lastly, our new lithium
9:10
refinery and catheter facility are progressing well that in
9:15
conclusion, we continue to focus on making as many cause as we
9:19
can while maintaining healthy financials.
9:22
Our artificial intelligence development is obviously
9:25
entering a new era and we're incredibly excited about what's
9:30
to come. Our other businesses such as
9:32
Megapac supercharging service. And whatnot all started to
9:36
become a miracle contributor to overall profitability this
9:40
quarter. And then lastly, I'd just like
9:43
to profusely thank all of our employees for making it a lot of
9:49
extra effort during uncertain times.
9:51
Thank you very much for your hard work and the impact you're
9:53
making. Thank you very much, Elon.
9:56
And I think Zach did some opening remarks as well.
9:59
Yeah. Thanks, Martin.
10:01
As Elon mentioned, Q2 was another record quarter of
10:04
production and deliveries, as well as records and profit for
10:07
energy and services and other businesses.
10:09
Congratulations again to the Tesla team on the continued
10:12
progress as we navigate through a period of economic
10:16
uncertainty, rising interest rates, volatility and consumer
10:19
confidence and regulatory change.
10:21
I want to comment on our financial approach.
10:24
First, the single most important priority is to ensure we are
10:27
continuing to invest heavily in the core technologies that will
10:30
drive the longterm value of the business.
10:33
This include increasing spending on AI related technologies such
10:36
as full self driving Optimus and Dojo as well as new products
10:40
such as Cybertruck, our next generation platform and the semi
10:44
as evidenced by the continued growth in our R&D spend.
10:47
This also includes continuing our investments in capacity
10:50
expansion. Not only in our vehicle
10:52
factories, but also our supercharging network, service
10:55
internal applications and battery processes as we continue
10:59
with meaningful capital expenditures to lay this
11:01
foundation for the future. 2nd, we continue to work towards our
11:04
goals of maximizing volumes on both our vehicle and energy
11:08
business, but most importantly doing so in a way that generates
11:12
the capital to continue our pace of R&D and capital investments.
11:16
This requires a strong focus on per unit COGS reductions in each
11:19
of our key businesses. As well as working capital
11:22
improvements on raw materials, working process inventory and
11:26
customer AR, all of which progressed appropriately in Q2.
11:30
If we look specifically at our automotive business, our gross
11:33
margin showed a modest reduction and remain healthy despite
11:37
action taken to further improve vehicle affordability.
11:40
Early in the quarter, we recognize we realize per unit
11:44
cost improvements in nearly every category including
11:47
material costs and commodities, manufacturing costs and
11:50
logistics. While also continuing to rapidly
11:52
increase the build rate in our Austin and Berlin factories.
11:56
For our energy business, we improved margins and gross
11:58
profit driven by cost reductions in deal economics, particularly
12:01
with Megapac. As a reminder, storage volumes
12:05
are typically volatile sequentially based on the types
12:07
of projects and their specific revenue recognition milestones.
12:12
As we look forward to the rest of the year, I want to reiterate
12:14
Elon's comments on Q3 volumes driven by plan down times for
12:18
factory upgrades. These upgrades will also carry
12:21
some amount of factor at all cost how.
12:23
However, we are working to minimize as much as possible.
12:26
It's also important to keep in mind the uncertainty in the
12:28
macro environment, which can impact our execution positively
12:32
or negatively in the near term. Regardless, we continue to
12:35
remain dynamic with a focus on fundamental efficiency and a
12:39
longterm outlook. Congratulations again to
12:42
everybody on a great quarter. Thank you very much, Zack.
12:45
And let's go to invest your questions.
12:47
The the first question on licensing FSD we've already
12:50
answered. So let's go to the second one.
12:52
The second question is what is the status of 4680 cells and how
12:56
far are you from the specs you laid out on battery day?
13:00
When do you expect to achieve what you laid out on battery
13:03
day? Yeah.
13:05
First, I'll just start with a little bit of a production
13:07
update. So in Texas 4680, cell
13:10
production increased 80% Q2 over Q1.
13:14
And the team surpassed 10 million production cells
13:16
produced here in Texas, so Congrats to the team for that.
13:19
Their focus on yield produced our scrap bill by 40% quarter
13:22
over quarter and that resulted in a 25% reduction in cell COGS
13:28
here in Texas. We're preparing to launch our
13:30
cyber truck cell, which is 10% higher energy density than
13:33
current production. That was accomplished through
13:35
process and mechanical design optimization as we scale cyber
13:39
cell production through the end of.
13:41
The year and early next we should be in a comfortable place
13:44
on cost per cell against our battery energy density targets.
13:49
The cyber cell is at our expectations on a like for like
13:52
electrochemistry basis. We're yet to integrate silicon
13:56
or in house capital production both reviewed on battery day
13:59
which do break significant further energy density and cost
14:02
improvements, but that is a topic for another day.
14:05
Lastly, it is important to remember that most of what we
14:07
focused on a battery day with the Tesla engineered. 4680
14:10
production system and the improvements we strove to
14:13
achieve on equipment factory density, capital cost and
14:16
utility cost reduction, all of which we are realizing in our
14:19
Texas scale up to date. Thank you very much.
14:24
The next question is, can you talk more to the upcoming Tesla
14:28
Energy products and how you your thinking has evolved on the
14:31
revenue model given Tesla's A I capabilities, how do you see the
14:35
longterm mix between hardware margin and recurring softer
14:38
margins from auto bidder? As the segment accelerates.
14:44
We can't comment on future product road map, but I can
14:46
provide a cool energy Q2 update. Megapac continues to show strong
14:51
demand globally with Lathrop ramping successfully to meet our
14:54
contracted projects in 2023. As stated last quarter, Megapac
14:58
margins are in a reasonable place in line with our target
15:01
market vehicle target margins. The second final assembly line
15:05
at Lathrop is progressing on schedule, eventually doubling
15:08
Lathrop capacity ahead of our full factory ramp in 2024.
15:11
We have several exciting large projects and construction or
15:14
nearing completion, including the KES project in Hawaii, the
15:18
River Rena project in Australia, several projects in California
15:20
and one here at Gigafactory, Texas that I toured today.
15:22
Actually, we want to thank our customers, utilities and grid
15:26
operators for trusting us with these projects on the Automator
15:29
question. We continue to grow auto bidder
15:32
contracts in wholesale markets like Australia, Texas, UK and
15:34
California with over 6 GW hours under Tesla's dispatch next year
15:39
In the UK, our project performed best in the industry.
15:42
In Q2. Auto bidder does have software
15:44
margins and is an enabler for hardware sales, but it's a
15:47
relatively small contributor to revenues given how much
15:50
deployment growth on the magnet back hardware side is occurring.
15:52
It's important to remember that these large projects, these
15:56
large capital projects have lifetimes, 20 years of occurring
15:58
revenues on an analyzed basis. Relative to upfront, CapEx are
16:01
small. On the residential side, we have
16:04
some fun things happening. We recently surpassed a half
16:07
million power walls installed. Just this week we were launching
16:10
charge on solar, which allows festival power wall and vehicle
16:13
customers to charge their vehicles using their excess
16:15
solar and drive only on the sunshine that hits their roof.
16:19
Yesterday we began paying customers in Texas for
16:21
participating in our virtual power plant to provide grid
16:23
support to IRCOT. We expect these credits to lower
16:27
our median customers annual bill by 1/3.
16:29
And to increase these credits over time as ERCOT expands
16:31
market access and today we are expanding Tesla electric
16:34
enrollment to new Model 3 honors in Texas, followed by all Texas
16:38
vehicle customers over the rest of the quarter.
16:40
Unfortunately and somewhat similar to Tesla Insurance,
16:43
bringing Tesla electric and BPP capabilities to our customers
16:46
requires working through a fractured regulatory environment
16:49
on a jurisdiction by jurisdiction basis.
16:52
In the long run, the value of residential energy software and
16:55
hardware will be driven by the level of market access that
16:57
utilities. Market operators and regulators
16:59
permit for power walls eligible to provide the full stack of
17:03
energy services like peaker, peaker capacity and system
17:06
buffering such as in Australia. We can more than double the
17:09
value of ownership relative to a typical system today.
17:13
Thank you very much. The next question is, could you
17:15
quantify the benefits to COGS per unit from the IRA battery
17:20
manufacturing incentives? And secondly, battery raw
17:23
material declines year to date? Right.
17:27
I can take that. On the first part of the
17:31
question for IRA manufacturing incentives, we provided previous
17:36
guidance that we expect these to be for the course of this year
17:39
in the range of 150 million to 250 million per quarter.
17:44
We are staying within that boundary as we got it
17:46
previously. So that was the case in Q2 as
17:48
well. I will note, I think we've
17:51
mentioned this before that. This includes a 5050 sharing of
17:55
credits for qualified cells from our longterm battery partner
18:00
Panasonic. On the commodity side are we are
18:05
continuing to see improvements there.
18:07
As we've discussed previously, lithium is the most notable
18:10
improvement so far. I think I commented on this on
18:15
the last call because typically we see this coming about 1/4
18:18
before it actually is realized in our financials.
18:22
And also just as a reminder, we're not fully exposed to the
18:24
price of lithium. Our supply chain team has done a
18:27
terrific job in partnership with another bunch of other companies
18:30
to put in place some longterm agreements here.
18:35
But we do have some exposure that moves up and down.
18:38
We're also seeing benefits in aluminum and steel, which I
18:41
think is great, not as large as the lithium impacts, but they
18:46
contribute nonetheless. So if we add up the total impact
18:49
of this in Q2 relative to prior quarter, it's about the same
18:53
size and magnitude as the IRA benefits that we also received.
18:58
You know just to put this in context, you know as you look at
19:01
COGS per unit sequentially from Q1 to Q2, I think there's
19:05
there's two things to keep in mind there.
19:08
The 1st is that our SX mix and IT for deliveries increased
19:12
quite a bit from Q1 to Q2. So as you think about
19:15
fundamental cost reductions. It's important to adjust for
19:19
that. And then secondly, you know as
19:22
we continue to work on reducing our Austin and Berlin costs
19:25
which we did quite a bit of that from Q1 to Q2, You know these
19:30
factories are still a slightly above model wide production
19:33
costs elsewhere and in the quarter our mix of Austin and
19:36
Berlin related builds increased And so that's something to
19:41
consider as you model out the impact on from Q1 to Q2 in terms
19:45
of COGS per unit. And I do want to ask Claren if
19:48
there's anything else on the commodity side or just more
19:51
generally you want to add here. Yeah.
19:54
As you mentioned Zach, we we've naturally been a little bit
19:56
hedged from the losing position because they the longterm
19:59
contracts we have in place. But we have seen reduction in
20:01
pricing across the board for all commodities that and
20:04
specifically going to batteries such as nickel, cobalt and and
20:07
and graphite. And the reductions and pricing
20:10
translate into thousands of dollars when you look at it from
20:13
a PURV vehicle and plant. We're taking advantage of the
20:18
historically low commodity pricing and in certain areas to
20:20
kind of expand some of those fit price contracts through the end
20:23
of the decade since a playbook that will continue to kind of go
20:26
back to as you look to the future.
20:30
Thank you. And the next question on FSD.
20:33
Have you considered allowing FSD transferability as a lever to
20:37
allow existing customers to upgrade to a new Tesla instead
20:40
of being locked into an existing car due to the price of FSD?
20:45
Yeah, this is a question we get asked a lot.
20:48
So we're excited to announce that before Q3 we will be
20:53
allowing transfer of FSD. This is a one time amnesty.
21:00
So it needs to be take, you need to take advantage of it in Q3,
21:04
but or at least place the order in Q3 with within with within
21:08
reasonable delivery time frames, so.
21:13
Yeah, yeah, yeah. I hope this makes people happy
21:20
this one time thing. Right.
21:23
The next question, when will, when will we give more
21:27
information about the cyber truck orders, estimated delivery
21:30
schedules, pricing and specifications?
21:34
The demand is so is so far off the hook you can't even see the
21:38
hook, so that's really not an issue.
21:42
I do want to emphasize that the cyber truck has a lot of new
21:45
technology in it like a lie, it doesn't look like, it doesn't
21:51
look like you know any other vehicle because it is not like
21:53
any other vehicle. So at the and the production
21:57
ramp will move as fast as the slowest at least like the
22:01
element of the of the entire supply chain and an internal so.
22:11
You know, I I wouldn't expect you know it, I I hope it's
22:14
smooth. We're, you know, we're certainly
22:15
better production ramps than you know we've got a lot of
22:19
experience with production ramps.
22:20
But you know first order approximation is there's like
22:24
10,000 unique pots and processes in in a you know in the cyber
22:30
truck and if any one of it'll go as fast as the least lucky, you
22:35
know least well executed elements of the 10,000 so.
22:40
Always pretty difficult to predict the the ramp initially,
22:42
but I think we'll be making them in high volume next year and we
22:47
will be delivering the car this year.
22:52
Thank you. The next question is critics of
22:55
Giga casting contended that process makes vehicles harder
22:58
and more costly to repair, essentially pushing costs onto
23:01
the customer. You share some details about the
23:04
initial repair experience with Giga Cast vehicles.
23:06
That must be why everyone's copying us.
23:10
Yeah. Thank you.
23:11
On this is Larris and Martin, that that's like simply not
23:14
true. There's a misconception that
23:16
traditional bodies are easy to repair, but they're made of
23:19
multiple materials and multiple joining methods.
23:21
Spot wells and rivets have to be drilled out.
23:23
Panels and structural adhesive have to be chiseled out, tried
23:26
adhesive has to be removed, stamp things cut, blah blah,
23:28
blah. It's a crazy patch of quilt.
23:30
Yeah, And so putting that back together means time and money,
23:33
using an example of replacing a rear cast rail on the Model Y.
23:38
To do that versus like what we replaced it with from the Model
23:41
3, it's 1010 times cheaper and three times faster to do it with
23:44
the cast rail. My design team works with our
23:47
collision repair team Since we're closed loop on this with
23:50
insurance and we designed specific parts that are make it
23:53
easier and faster to repair, we have an incentive to do that
23:56
because we have our own insurance and our own body
23:58
shops. We expect that we'll continue to
24:02
do this and collision repair will continue to become cheaper
24:04
and faster over time. And we already need this
24:06
available to all body shops or a tester include Body Shop
24:09
training, yeah? Closing loop on collision repair
24:12
and factoring that into design is is a big deal.
24:14
It's crucial. I don't think anyone else can do
24:16
it with that ecosystem that we have, so yeah.
24:19
And and we are actually able to change the details of the
24:21
casting with inserts, and we actually do do that all the
24:25
time. So because the answer is
24:28
actually we're out and it should be replaced anyway.
24:31
So we can actually make design changes to the inserts and tweak
24:33
the castings. But the the, the cast that you
24:36
know, basically cast the rear, rear body or front body is
24:40
lighter, cheaper, better for noise, vibration, harshness,
24:44
much easier to manufacture and it's better in every way.
24:49
And that's why so many other car companies are copying us because
24:53
probably they don't really stick well.
24:55
They certainly put out a lot of fresh releases about it.
24:59
I think it's basically got to be how old old cars are made in the
25:02
future. Thank you.
25:06
Next question, how many optimist bots have been made and when
25:10
will they be able to start performing useful tasks?
25:12
10 million, yeah. I think, I think we're around
25:23
5:00 or 6 bots like, you know, there's a look 10, I guess.
25:28
Yeah, which, which is how many are working and what phase.
25:34
It's a it's sort of yeah like this there's more more every
25:41
month the they're saying it's large things, large interesting
25:47
things about the Octopus spot. We found that there are actually
25:53
no suppliers that can produce the actuators.
25:56
There's no there are no off the shelf actuators that work well
25:59
for a Humor Aid robot at any price.
26:02
Certainly not compelling. Yes, not a not a human robot
26:07
that can do something that you know the things that human can
26:10
do. So we've actually had to design
26:14
our own actuators that integrate the motor of the power
26:18
electronics, the controller sensors and really every one of
26:27
them is custom site so. And then of course we'll be
26:34
using the the same infrared hardware as the car.
26:40
So you know and but we are in designing these actuators,
26:46
designing them for volume production.
26:48
So they're not just lighter, tighter, more and more capable
26:52
any any other actuators that that we're aware of that exists
26:55
in the world. There's also actually
26:58
manufacturable. So we should be able to make
27:01
them in volume. The the first optimist that is
27:07
that will have all of the test that designed actuators sort of
27:11
production candidate actuators on it integrated and walking
27:16
should be on to November ish. So and then we'll we'll we'll
27:23
start wrapping up after that you know in terms of when we'll be
27:31
able to do some useful things. Like, we'll first be trying this
27:34
out in our own factories and just proving out its utility.
27:37
But I I think I think we'll be able to have it do something
27:41
useful in our factories sometime next year.
27:46
I I would be, yeah, I'm pretty, pretty confident of that.
27:52
So yeah. And that's going well.
27:58
I should say another cool thing about Optimist is that you know
28:02
there's. Just in the US alone there are
28:03
two million amputees and I was just talking to the neural link
28:09
team and by combining A neural link implant and a robotic
28:17
armour leg that for someone that has has had their arms and leg,
28:21
all arms and leg amputated, we believe we can get basically a
28:27
sideball body that is incredibly capable. $6 million man.
28:36
In real life before no word cost $6,000,000.
28:42
Sixty $1000 man since I was impressive but it's it'll
28:46
actually yeah so that that that actually could be a really I
28:51
think would be incredible to you know potentially help most
28:55
people around the world and and give them you know a robot or
29:01
like that is. As good make a long time better
29:04
than a biological 1. Thank you.
29:09
The next question is how has the order intake trended relatively
29:13
to production levels during Q2 and how is it trended in the
29:16
quarter to date period? Conceptually, how does this law
29:19
decide when is it appropriate to reduce prices or at other sales
29:24
incentives to increase demand? I guess demand is roughly track
29:29
production, so. Which is what we aim for is, is
29:36
that we look at you know something that we have that
29:38
really I think no other car maker has is that we have real
29:43
time demand and real time production like so seven days a
29:47
week, you know, I get an e-mail order generated e-mail that
29:52
shows output from both factories and orders globally.
29:57
So it's like a real time thing on the pulse of earth.
30:01
Basically and and we would you know we just we're just course
30:07
according to what the word of the of the public is.
30:12
You know buying a new car is a is a big decision for vast
30:16
majority of people. So you know at anytime there's
30:19
economic uncertainty people generally pause on your car
30:23
buying at least to see to see what happens and you know and
30:29
then obviously another challenge is the.
30:31
In the interest rate environment, as the interest
30:33
rates rise, the affordability of anything bought with that
30:38
decreases, so effectively increasing the price of the car.
30:43
So when interest rates rise dramatically, we actually have
30:47
to reduce the price of the car because the the interest
30:50
payments increase the price of the car.
30:53
So and and this is at least at least up until recently it was
30:58
the, I believe the sharpest. Interest rate rise in history,
31:03
Do we have to do something about that?
31:07
And if somebody's going to crystal ball for the global
31:09
economy, I really appreciate it if I got borrow that crystal
31:12
ball. DMS.
31:14
Yeah, exactly. Damn.
31:15
Yeah, it should be known on Twitter.
31:19
So I mean one, one day it seems like the world economy is
31:24
pulling apart and the next day everything's fine.
31:27
I don't know what that was going on.
31:30
I'd be totally frank, I wish I did.
31:34
So I mean that's why that's why I say like yeah, I always, you
31:38
know you know on Twitter I posted like you know, just
31:42
really advising because I care a lot about the the, the, the sort
31:48
of the small shareholders especially ones that have stuck,
31:51
stuck with us through through thick and thin.
31:53
I love you guys and so the. We we can't control these macro
32:00
shocks you know or the the manic depressive nature of the stock
32:04
market. So that's why I I recommend
32:06
against margin loans in times that are turbulent.
32:10
You know if I if times are are not that turbulent actually a
32:12
margin loan can be a smart move within reason but we're in I
32:18
would call it turbulent times. I like I have very high
32:23
confidence, confidence in the longterm value of Tesla.
32:28
Like I, I see it. I really, you know, see a path
32:31
to a 10X. Well, we're not telling these
32:33
days about you. The 5X increase in the value of
32:39
the company, maybe a 10X and the the where things go along the
32:46
way that the trials and tribulations and the mood of the
32:49
mood of the markets, well, I cannot predict.
32:52
And so you know, I mean. The the old adage of buy and
32:59
hold, this is right, you know, for investment advice I say like
33:06
identify companies products you love see if they you know, does
33:11
it seem like they'll continue to make good products or great
33:14
products. Buy that stock and holds it,
33:18
That's it. But you all went and the reason
33:24
companies exist is to make goods and services.
33:27
Ideally, great goods and services, they don't exist for
33:31
any other reason they shouldn't. So that's why you should buy
33:34
stock of a company that makes me like some has a great future
33:38
pipeline. It's common sense actually.
33:42
And and then and then generally if you see if if you provided
33:49
you're confident about what that company's products or services
33:52
are when the market panics by. And when the market is, you
33:59
know, overly exuberant, yeah, you can sell.
34:01
I'm not recommending yourself to as a bitch, but but yeah, buy
34:09
low, sell high. You know, Warren Buffett
34:16
actually I think has saying I'm paraphrasing him.
34:18
But you know, a publicly traded company is like, it's like
34:22
imagine you're living in your house and some.
34:25
A crazy manic depressive guy comes to, stands at the outside
34:29
your house and yells property prices at you, you know, so it's
34:35
and it's a different price every day.
34:37
The house is still the same house, so who's talking about?
34:44
Yeah, that's warm. Thank you.
34:48
Let's go to the next question. With the emphasis of price cuts
34:51
to drive volume growth eating into automotive gross margin,
34:54
can investors expect to see automotive gross margin
34:57
stabilize or even rise to the efficiencies outpacing the cuts?
35:01
And if so, when? Where's that cross the ball?
35:07
Again, if I may, It's like look the short term variances in in.
35:17
The most margin profitability really are minor relative to the
35:22
long term picture. Autonomy will make all of these
35:28
numbers look silly. I'd recommend looking at Arc
35:31
Invest. I think their analysis is very
35:34
good, the best that they you know I mean and generally fin,
35:41
Twit or like the finance, smart finance people on Twitter.
35:46
Follow their accounts. They're great.
35:50
So that that's what that's that's an idea where you'll get
35:56
the best, best info. So you know, I strongly believe
36:02
Tesla is an epic long term investment and don't swear when
36:07
you know things go up and down. In fact, the market panics by if
36:13
your market's a little too exuberant sell time.
36:15
But. But just generally like like I
36:19
feel I confident you know we will deliver a long term but
36:22
can't control short term. So and these the autonomy is
36:27
really where it's at exactly. I fully agree with you.
36:32
I mean, I think the only thing in the short term that matters
36:37
is, is what I said in my opening remarks, which is, you know, are
36:41
we generating enough money to continue to invest?
36:44
In, you know, the portfolio of products and technologies that
36:48
the technical teams are investing in right now.
36:51
And this is intense. It's intense in terms of
36:55
investment. It's intense in terms of
36:57
potential. Frankly, I think it's ridiculous
36:59
that we have positive free cash flow in a capital intensive
37:03
business while investing massive amounts of money in new
37:07
technology. That is super hard.
37:10
And critical integration, it's not even just like new products,
37:12
but also. Yeah, we actually make our shit.
37:17
We don't. Yeah.
37:19
And and. So others.
37:25
And. Very cool.
37:28
And so at least from my perspective, what matters is
37:31
continuing to generate the cash to invest and you know that
37:35
means continuing to be hyper focused on your term cost
37:39
reduction. Does everything we do in your
37:41
term cost reduction provides capital to reinvest hyper
37:45
focused on working Capital Management which we've made
37:50
quite a bit of progress there on the raw materials and whip as a
37:55
side of that and very focused on accounts receivables as well to
37:59
ensure that we can continue to reinvest, reinvest the cash.
38:04
You know that this is what we're focused and you know so there's
38:08
you know a set of this that. We control, you know we have a
38:11
pipeline of cost reductions, but we are getting tailwinds in the
38:15
commodity space right now as Carn mentioned, that's helpful
38:20
variability around average selling prices.
38:22
You know goes back to Elon's point and we don't control
38:26
interest rates, we don't control macro consumer sentiment, but we
38:31
have an obligation to be responsive to that to ensure
38:34
that we're matching supply and demand and keeping things
38:37
balanced. And so this is how we're
38:39
managing the next handful of quarters and soon enough these
38:44
quarters will be behind us. They won't be part of the
38:48
present value of future cash flows of the business.
38:51
And so we wanna make sure we keep that view and make sure
38:54
that the long term of the business is exactly the way that
38:57
we want it to be it. Will suit, yeah.
39:01
All right. Thank you very much.
39:02
And now let's go to analyst questions.
39:06
The first question comes from Dan Levi from Barclays.
39:09
And then feel free to unmute. Yourself.
39:14
Great. Good evening.
39:17
Thank you. Bob, wanted to start first with
39:20
the with a question about here at Person AI and Dojo.
39:24
It's pretty clear it sounds like you're accelerating your focus.
39:27
Can you maybe provide us with a sense of what the process is of
39:32
refining a product, Is it more machines and and maybe you could
39:35
give us a sense of you know when the the payout starts to when
39:41
you start to see the payout and what the resource outlay is.
39:44
You know what should we expect on the OpEx front as a result of
39:47
this? Sorry, are you saying how much
39:51
are we gonna spend on Dojo or yeah?
39:54
Warranty of Dojo. Yes.
39:58
Well, we're not going to be open loop on our Dojo expenditures.
40:01
So but I mean I think we, we will be spending you know for
40:07
the north of a billion for the next year on on you know flip
40:10
through the end of next year, it's whatever billion dollars in
40:13
Dojo. And yeah, so we've got a truly
40:21
staggering amount of of video data to do training on.
40:25
And this is another thing I like.
40:26
In order to copy us. You also need to spend billions
40:29
of dollars on training compute. I mean it's like and it's also
40:35
hard to you know you need the data, you need the training
40:38
computer. It's like they think well things
40:41
needed to actually achieve it at scale toward generalized
40:44
solution for autonomy is it's this is this is one of the
40:47
hardest problems ever. You know, you see a lot of AI
40:50
companies doing, you know, Llms and and and whatnot and I'll say
40:54
if they're so great, why can't they make yourself a great
40:57
driving car? Because it's harder, that's why.
41:05
So but I do think those, let's say, I think there's, you know,
41:08
great, great AI companies out there, but but just
41:12
fundamentally the the. About the the staggering amount
41:16
of data we've got to process is going to process somehow and
41:19
custom silicon is the best way to do that.
41:24
So that that's what Dojo is is designed to do is is you know
41:30
optimize for for video training it's it's not optimized for all
41:35
of them it's optimized for video training with with with video
41:40
training you have a a. A much higher ratio of compute
41:45
to memory bandwidth. So this you know whereas LMS
41:51
tend to be memory bandwidth choked.
41:55
So that's that's it. I mean like so we're also we
42:00
have some, we're using a lot of the video hardware continued to
42:04
you. We'll actually take the video
42:08
hardware as fast as NVIDIA was delivered to us.
42:11
Tremendous respect for Jensen. Kind of video on they've done an
42:16
incredible job and and frankly I don't know if they could deliver
42:23
us enough Gpu's we we might not be Dojo but they can't.
42:29
So because they got so many customers they've been kind
42:33
enough to you know but nonetheless prioritize something
42:37
about our GPU orders but. Yeah, the, the, the sheer
42:45
magnitude of video training because like I said, we're not
42:47
trying to just get as good as human.
42:49
We want to get to, you know, 10 times better than human, maybe
42:52
100 times better than human. Right now I believe there's
42:56
something on the order of 1,000,000 automotive deaths per
42:58
year. And and if you say permanent
43:02
serious injuries, I think it's probably closer to 10 million
43:04
per year. And you know, so it matters if
43:09
you're, if you're, you know. Twice as good as as humanity 10
43:15
times you know like 10 times better than here would would
43:18
still mean 100,000 deaths on and and a million severe permitted
43:23
injuries. So it's like OK well we'd
43:25
probably be 100 times better So there's there's really you know
43:29
it's a March of 9th and and we want to achieve as perfect
43:34
safety as possible and to ask truly mind boggling amounts of
43:39
video and and and computer needed for that.
43:42
So and then I just, yeah, I do think there's this other
43:45
applications for for Dojo, but we just definitely needed for
43:49
video training. Right.
43:51
Just to just to add to what Elon mentioned, so you know the
43:55
numbers that he mentioned are you know between R&D spend and
43:59
capital spend and you know this is moving quickly you know and
44:05
so we provide a three-year outlook on our capital expense
44:09
we are considering. These paid expenses in that
44:13
outlook and as that moves up and down, we'll continue to update
44:17
our guidance in the queue. Yeah, I wanna say the big, the
44:21
fundamental rate limiter on the progress of both self driving is
44:25
trading. But that's if if we had more
44:29
training compute we would get it done faster.
44:32
So that's that's it. And it's just difficult to
44:36
predict how quickly we can execute on it.
44:40
Create. Thank you.
44:42
Just as as a follow up, I recognize you know there's
44:46
there's incredible macro uncertainty right now, but
44:50
you're sticking with your near term your volume target 50%
44:54
kegger. As we just think about sort of
44:57
in the in the year ahead, you know cyber truck is going to be
45:01
some contribution. You know there's going to be
45:04
some help for further EV penetration growth, but to to
45:07
what extent are you willing to? Sacrifice on pricing to keep
45:12
that 50% volume kegger intact. Or, you know, are you?
45:18
Thinking differently about margins versus your prior
45:21
commentary of willing to sacrifice on margins to get more
45:23
share. It's not like getting what
45:26
share, it's just that you can think of every car that we that
45:30
we sell or produce that that that has a full autonomy
45:35
capability as. Actually, something that in the
45:40
future may be worth as much as five times what it is today
45:46
because, you know, average passion is big.
45:50
Pastor Bickle is doing like maybe 10 hours of driving a
45:53
week. You know, one if if it's sort of
45:55
1 if let's say it's 1 1/2 hours there, on average that's 10
45:58
hours a week. Ish if you've got an autonomous.
46:07
If that that vehicle is able to operate autonomously and and and
46:13
and use be used in some so we either either dedicated
46:16
autonomous or partially autonomous like like Airbnb like
46:19
maybe sometimes you allow your card to be used by others
46:24
sometimes you want to use it exclusively just like you know
46:26
Airbnb you know drink Airbnb with a room in your hands you
46:33
know that the the value is just tremendous so.
46:36
I think it's sort of it would be.
46:39
I think it makes it does make sense to sacrifice margins in
46:42
favor of making more vehicles because we think in the not too
46:46
busy future they will have have a dramatic valuation increase.
46:50
I think the test of fleet value increase at the point of which
46:54
we can upload full self, you know full self driving and it's
46:56
approved by regulators will be the single biggest step change
47:02
in asset value maybe in history. Thank you.
47:09
Let's go to the next analyst. The question comes from Emmanuel
47:12
Rosner from Deutsche Bank. Thank you very much.
47:17
Two questions for me as well. First, following up on on the
47:20
autonomy, So you know before you start launching these dedicated
47:24
Robo Taxi vehicles on unexisting vehicles, you're improving FSD.
47:29
You know incrementally what is your latest targeted timing to?
47:34
Essentially release a non beta version or an ISOF version that
47:38
would trigger much higher take rates and would Tesla benefit
47:42
from lowering the price of FSD? Well, obviously as.
47:52
People have sort of made fun of me, and perhaps the quite fairly
48:01
have made fun of me, my predictions about.
48:04
Achieving full sub driving have been optimistic in the past.
48:11
The the reason they've been optimistic is what it tends to
48:14
look like is the we'll make rapid progress with the new
48:18
version of of FSD but but then it will curve over
48:24
logarithmically so. So first logarithmic curve looks
48:29
like you know just fairly straight upward line diagonal
48:33
up. And so if you extract like that
48:36
it then you you have a great thing.
48:37
But then it's actually logarithmic.
48:39
It curves over. And then there's been a series
48:41
of log that stacked logarithmic curves.
48:44
Now I don't. I don't.
48:49
I'm the the, the, the boy who cried FSB.
48:53
But I I I think I think we'll be better than human by the end of
48:58
this year. That's not to say we're treated
49:04
by regulators and and I'm saying that then that would be in the
49:09
US because we got to focus on one market 1st.
49:12
But I think we'll be better than here and by the end of this
49:14
year. I've been wrong in the past, I
49:16
may be wrong this time and that the from the price of FSD.
49:22
So the way thing is the price of FSD is actually very low.
49:26
It's not high. When you go back to what I say
49:29
earlier, the value of the car increases dramatically if it is
49:34
actually autonomous. You know $58,000 is is actually
49:39
a low price, not a high price. And that we will offer and you
49:47
know and we do sort of offer FSD as a sort of monthly
49:51
subscription, although like most people don't know that.
49:55
So I recommend like maybe trying it out as a monthly subscription
49:58
so you don't have to go with the $15,000 thing.
50:01
But I think yeah, yeah, the the obviously if the car is built
50:08
several times it's original price, $15,000 is actually a low
50:14
price for it. Thank you.
50:18
And the next question comes from William Stein from Trust.
50:22
William, go ahead and unmute. Great.
50:26
Thank you very much for taking my question.
50:31
I'd like to ask about the stick on this A I topic.
50:38
We've read you know with great interest the developments in
50:42
Dojo today and you've spoken about FSD but you've also Elon
50:46
you started this X dot a I company and you know for
50:51
investors that think that. There might be quite a bit of
50:53
value in the AI features and products of Tesla.
51:00
It might be concerning to see you, you know, pursuing another
51:03
endeavor where AI is the focus. So can you talk about how X dot
51:07
AI might overlap, might perhaps compete with Tesla or in other
51:14
ways perhaps it it enhances the value of of what Tesla does?
51:18
Thanks very much. Yeah, I I think we'll actually
51:22
enhance the value of Tesla there.
51:27
There were just some some of the world's best AI engineers and
51:32
scientists that were willing to join a startup, but they were
51:35
not willing to join a large, relatively established company
51:39
like like Tesla. So I was like that that's
51:42
actually how it got started. I was interviewing a few people
51:44
and they're like, no, we we want to do a startup.
51:46
I was like and that's all I I I couldn't convince them to join.
51:51
Tesla so, so it's like OK well you know better to start up that
51:58
that I run that's then then they go work somewhere else that's
52:02
kind of the the genesis of of XAI&XAI is is focused on a sort
52:07
of a GI. Yeah so it's but I'd like so I
52:15
think there will be some. Value that XAI brings to Tesla.
52:24
You know also for some of the best for the very best people in
52:27
the world, they really just want to look at interesting problems.
52:30
If you take say you know amateur science group you know really
52:34
what convinced the Charlie Coleman to leave Apple where it
52:37
was very happy and well compensated and both at the.
52:48
Where we think it's the best material science group in the
52:50
world was that he got to work at both Tesla and SpaceX.
52:53
He he wasn't willing to leave Apple if it was just how slow,
52:57
but he was like to know it was Tesla and SpaceX.
53:01
So sometimes you get the best out in the world.
53:02
That's the kind of thing you know you need to do.
53:05
And that actually has been very beneficial to Tesla so.
53:10
If you know if I can squeeze one more mundane question in, I
53:14
wonder if you think you can hit. The 1.8 million unit number with
53:19
current pricing or you anticipate needing to continue
53:23
to lower prices because it seems like they've are stabilized, the
53:26
trends are stabilized in the last maybe month and a half.
53:29
Should we expect a sort of continued decreases or or more
53:34
stabilization for the rest of the year?
53:42
You know, we have. We have a really.
53:45
We started the referral program, which I think will be quite
53:47
effective. You know, as as Zach was saying
53:51
earlier, we don't control the macroeconomic conditions.
53:55
So if interest rates continue to rise that reduces the
53:58
affordability of cars, you know, and and for a lot of people
54:03
they're really trying to balance that's, you know, barely
54:06
breaking even every month. In fact, if you look at the rise
54:09
in credit card debt, they are in fact not breaking even every
54:11
month. Credit card debt is this looking
54:16
scary? So yeah, we're like, which is
54:21
don't control the micro conditions.
54:23
If micro conditions stable, I think prices will be stable.
54:26
If they're not stable then we would have to lower prices,
54:31
yeah. Thank you.
54:34
Let's go to Colin Rush from Oppenheimer.
54:38
That's so much, guys. You know is you're building out
54:41
Dojo and implementing what truly is going to be a highly complex
54:46
set of software. Can you speak to the maturity of
54:48
the operating system and how much software you're expecting
54:52
to use in that? System This is a custom software
54:58
stack so but it is design such that you can run at the high at
55:06
a high level high torch. And JAX, so, but then yeah we
55:14
have to customize it to actually run on a custom silicon.
55:18
So this the software stack is a combination of open source
55:23
software and then and that tells the software all the way to the
55:26
bear silicon, which is the case for the first computer in the
55:29
car. OK.
55:34
Thanks so much. That's super awful.
55:35
And then can you speak to how you're managing some of the
55:37
geopolitical risks? Relative to your capacity
55:40
expansion, you know, obviously as you guys continue to grow at
55:43
this rate, you're going to be putting some folks out of
55:45
business and you know there's going to be some impacts around
55:48
regional economies. So just want to understand how
55:50
you're thinking about that in terms of some of your CapEx
55:53
plans and and how you're managing some of those
55:54
relationships with with different countries and regions?
56:02
Well, we, yeah, this is a period of unusual geopolitical risk.
56:09
So I think we're the best we can do is you know have factories
56:14
and many parts of the world such that if things get difficult one
56:19
part of the world we you know, we can still keep keep things
56:22
going in the rest of the world. Thank you.
56:28
The next question comes from Mark Delaney from Goldman Sachs.
56:37
Thank you very much for taking that question.
56:39
Tesla has been making progress reducing costs and did so again
56:42
last quarter. Can you give an update on when
56:44
you think automotive clouds per vehicle could be under the
56:46
historical $36,000 per vehicle level and what are the key puts
56:49
and takes together? This is I think it was asked
56:55
since in the past this is very difficult to forecast.
56:58
You know there's a series of costs that we manage in series
57:01
of costs in which we don't control.
57:04
And so you know, particularly on the commodity side where labor,
57:08
Costco, etc, it's just hard to say.
57:13
Yeah, We saw very inflationary, like strong inflationary
57:16
pressures for a while last year and and now which obviously
57:23
makes it very difficult to reduce COGS.
57:26
You know, now we're seeing what seemed to be deflationary
57:29
pressures. Certainly deflationary deflation
57:33
is a pressure. But we're seeing, you know,
57:37
commodity prices dropped, dropping as as was mentioned
57:40
there, you know, as car mentioned a moment ago.
57:44
And I think what do you think basically the trend should be
57:50
deflationary commodity low definitely.
57:53
Is that and then? There's also the the unit
57:55
economics improves as volumes grow.
57:57
That's the other thing we're seeing as we're becoming a
57:59
bigger and bigger part of a lot of suppliers, economies of scale
58:02
come into play. There's equipment depreciation
58:04
that comes into play, equipment Douglas Commission five to seven
58:07
years ago that used to be a part of the peace price that's
58:10
completely amortized. So we'll see situations or peace
58:13
price come down because that equipment contribution has gone
58:17
away. And then just we continue to
58:19
have this mentality of continuous improvement in terms
58:22
of Labor, reducing labor, improving automation and and
58:26
just continue to get better at what we do.
58:27
So we have seen I think every quarter we have seen an
58:30
approved. Of course, the commodities
58:32
spiked up and down. Just in general, the trend is
58:36
towards being more efficient. Yeah, when and totally agreeing
58:40
there. Unless the prices weren't
58:41
absolutely insane there for a while.
58:43
Yeah, and they're recovering now.
58:45
But also don't know what used to.
58:47
Be, Yeah. And you know we're still early
58:49
in the ramps. Well, not early in the ramp, but
58:52
early in the cost down curve. I lost an umbrella, yeah.
58:56
So it takes time to work the cost out at first.
58:58
It's a focused on ramp. Ramp.
59:00
It brings cost. And well be and then your and
59:03
quality cost. Yeah.
59:04
And then once that stabilizes, we can divert bandwidth to cost
59:07
reduction. And so Austin Berlin saw quite a
59:11
decent amount of cross reduction and a fundamental basis from Q1
59:14
to Q2. We'll continue to do that work.
59:16
That will be helpful as though we're just gonna keep shipping
59:20
away at it. Yeah, packaging is a big, big
59:24
element to it just takes 2. Logistics is normalizing, which
59:27
is great. That Cuba legalization,
59:32
something that team has been very focused on, So every bit of
59:37
it. Yeah.
59:38
It's hard. Logistics is under appreciated.
59:42
Yeah. So saying goes like that was one
59:46
of tactics was one of logistics, yeah.
59:49
And we've made tremendous improvements in cost and all
59:51
fronts you know express costs. We are down or down pre
59:57
pandemic. Expect cost levels now and our
1:00:01
goal is to go further down. Yeah.
1:00:06
So when we look at our progress from Q1 to Q2 on cost, know the
1:00:09
way that we look at internally and normalize for the impacts of
1:00:13
mix shift with Austin and Berlin being a higher percentage of our
1:00:16
mix, normalize for S&X being a higher percentage of our mix in
1:00:20
Q2 versus Q1, the sequential cross reduction, it might be the
1:00:24
largest we've had in a while. Anything.
1:00:27
It's great work on behalf of the test, the team, and we just
1:00:31
gotta keep it up. Yeah, it's a good game of
1:00:33
pennies, but Game of Thrones is good pennies.
1:00:38
Mark, do you have a follow up question?
1:00:40
I think you're muted. Yep.
1:00:43
Yeah. Thank you very much for all the.
1:00:45
Details on that. Maybe you could put a finer
1:00:47
point on the downtime impact that you spoke about in your
1:00:51
prepared comments in terms of production impact.
1:00:54
And then also to what extent there's a margin impact from
1:00:57
those that you're playing in the square?
1:01:00
Thank you. Yeah, the, the downtime, you
1:01:09
know, we don't know exactly the number of cars impacted because
1:01:13
you know, kind of the way that we go into downtime windows for
1:01:16
upgrades. Is you know, we set aside a
1:01:19
period of time but then the team is challenged it as quickly as
1:01:22
possible so that we can get the factories up and running again
1:01:24
and minimize that. So it's not, it's not profound
1:01:29
reduction, quite small. Too much if the weeds here.
1:01:33
I mean like we're asking for level of precision that is not
1:01:36
possible to answer, so let's move on.
1:01:40
Yep. I think this is unfortunately
1:01:42
all the time we have for today, so we'll speak to you all in the
1:01:47
next three months. Thank you very much.
1:01:48
Thank you.