1
00:00:03,440 --> 00:00:08,400
Ejaaz:
Elon is playing Game of Thrones yet again, this time releasing a new AI chip

2
00:00:08,400 --> 00:00:12,500
Ejaaz:
that is going to power his RoboTaxi, his new Tesla Optimus Primes,

3
00:00:12,540 --> 00:00:15,740
Ejaaz:
and all of his devices and cars that he launches right now.

4
00:00:15,900 --> 00:00:20,840
Ejaaz:
And the reason why this is so important is this is going to contend directly with Nvidia.

5
00:00:21,020 --> 00:00:23,240
Ejaaz:
He's coming for Jensen Huang's throat.

6
00:00:23,480 --> 00:00:27,660
Ejaaz:
But in other news, he also released this new product called the Megablock,

7
00:00:27,820 --> 00:00:33,820
Ejaaz:
which consists of Megapacks, which can power everything from homes to massive towns.

8
00:00:34,080 --> 00:00:39,120
Ejaaz:
But the reason why this is so important is it's gonna unlock 50% of the energy

9
00:00:39,120 --> 00:00:43,200
Ejaaz:
and power that the US and the world currently does not have access to.

10
00:00:43,300 --> 00:00:45,360
Ejaaz:
And that is so, so important.

11
00:00:45,520 --> 00:00:50,880
Ejaaz:
But starting off with these new AI chips called Tesla's AI-5 chip,

12
00:00:51,060 --> 00:00:55,600
Ejaaz:
which is supposedly meant to be 40 times better than their current AI-4 chips.

13
00:00:55,760 --> 00:00:57,000
Ejaaz:
But don't take my word for it.

14
00:00:57,200 --> 00:01:02,140
Ejaaz:
Take it from the man himself. And we're finalizing the design of AI5,

15
00:01:02,440 --> 00:01:05,640
Ejaaz:
which will be an immense jump from AI4.

16
00:01:06,320 --> 00:01:12,900
Ejaaz:
By some metrics, the improvement in AI5 will be 40 times better than AI4.

17
00:01:13,640 --> 00:01:16,300
Ejaaz:
Wow. So 40%, 40 times.

18
00:01:18,060 --> 00:01:24,220
Ejaaz:
And this is because we work so closely at a very fine-grained level on the AI

19
00:01:24,220 --> 00:01:26,120
Ejaaz:
software and the AI hardware.

20
00:01:26,120 --> 00:01:30,480
Ejaaz:
Work josh i know that you watched the entire presentation for the ai5 chips

21
00:01:30,480 --> 00:01:34,440
Ejaaz:
and for the transformer stuff i want to hear your take before we go any further.

22
00:01:34,440 --> 00:01:37,300
Josh:
Uh yeah we'll start one thing at a time we'll do the ai chips

23
00:01:37,300 --> 00:01:40,880
Josh:
first because those are freaking awesome so ai chips

24
00:01:40,880 --> 00:01:43,860
Josh:
have been big a big part of tesla since the beginning of the

25
00:01:43,860 --> 00:01:46,500
Josh:
company when you require autopilot in cars it means you

26
00:01:46,500 --> 00:01:49,160
Josh:
need a lot of processing a lot of inference power on these chips to

27
00:01:49,160 --> 00:01:52,540
Josh:
run locally on the vehicles ai5 is the natural extension of

28
00:01:52,540 --> 00:01:55,320
Josh:
what they've been working on so if you've owned a tesla you're familiar with

29
00:01:55,320 --> 00:01:58,420
Josh:
the tesla they started with ai1 they went up to 2.5 3

30
00:01:58,420 --> 00:02:01,260
Josh:
3.5 now we're at four which is currently shipping on

31
00:02:01,260 --> 00:02:04,440
Josh:
all vehicles hardware five is a slight change

32
00:02:04,440 --> 00:02:07,480
Josh:
in the strategy because previously tesla had these two domains

33
00:02:07,480 --> 00:02:10,620
Josh:
of experts in the world of ai and chip fabrication they

34
00:02:10,620 --> 00:02:14,120
Josh:
had their dojo team which was building this like commercial

35
00:02:14,120 --> 00:02:17,240
Josh:
infrastructure arm and then they had the ai chip

36
00:02:17,240 --> 00:02:20,340
Josh:
infrastructure team that was building the chips that actually went in the vehicles what they

37
00:02:20,340 --> 00:02:23,220
Josh:
did recently is they merged the two together into a single

38
00:02:23,220 --> 00:02:26,260
Josh:
ai fabrication team and they are working solely on

39
00:02:26,260 --> 00:02:32,460
Josh:
ai5 and ai6 now what makes ai5 so special well aside from it being much faster

40
00:02:32,460 --> 00:02:36,840
Josh:
than ai4 which is currently running on vehicles like elon just described ai5

41
00:02:36,840 --> 00:02:40,680
Josh:
also does this interesting thing where they're starting to actually use it for

42
00:02:40,680 --> 00:02:45,240
Josh:
real-time inference in data centers as well so an interesting thing that happens with

43
00:02:45,840 --> 00:02:48,440
Josh:
AI5 that I don't think a lot of people are going to recognize,

44
00:02:48,440 --> 00:02:53,000
Josh:
and I want to kind of draw a tie to Apple, is that they now own the vertical

45
00:02:53,000 --> 00:02:54,660
Josh:
integration of this hardware chip.

46
00:02:54,780 --> 00:02:59,020
Josh:
So, Ejaz, if you remember previously, Apple used to use Intel chips for their MacBooks.

47
00:02:59,120 --> 00:03:02,940
Josh:
When you had a MacBook Pro, you used an Intel chip. And it worked really well

48
00:03:02,940 --> 00:03:05,420
Josh:
until Apple rolled out the M-series chips.

49
00:03:05,600 --> 00:03:06,780
Josh:
And when they put those chips in

50
00:03:06,780 --> 00:03:11,140
Josh:
a MacBook, suddenly I no longer needed my 65-pound PC to render a video.

51
00:03:11,280 --> 00:03:14,200
Josh:
All I needed was my tiny little macbook and my battery life

52
00:03:14,200 --> 00:03:17,340
Josh:
doubled the processing power quadrupled everything about

53
00:03:17,340 --> 00:03:20,620
Josh:
the device got so much better because it was vertically integrated into

54
00:03:20,620 --> 00:03:23,800
Josh:
the system they own the entire stack from the hardware to software and everything

55
00:03:23,800 --> 00:03:27,060
Josh:
in between this is tesla's attempt at the m1 chip but

56
00:03:27,060 --> 00:03:30,020
Josh:
for robotics so tesla currently has two arms of

57
00:03:30,020 --> 00:03:32,820
Josh:
real compute real world compute and that's

58
00:03:32,820 --> 00:03:36,280
Josh:
their autonomy in the cars and autonomy in the robots ai5 and

59
00:03:36,280 --> 00:03:39,500
Josh:
shortly soon after ai6 will be the actual chip that runs in

60
00:03:39,500 --> 00:03:42,460
Josh:
both of those things it's a universal chip that exists to

61
00:03:42,460 --> 00:03:46,660
Josh:
run local inference on any sort of autonomous hardware device and it's going

62
00:03:46,660 --> 00:03:50,560
Josh:
to be really freaking fast really impressive and it'll it'll be small enough

63
00:03:50,560 --> 00:03:53,420
Josh:
that it could fit into all of these smaller devices so i think that's why this

64
00:03:53,420 --> 00:03:58,320
Josh:
is a big deal is this is very much apple's m-series chip moment uh but for tesla

65
00:03:58,320 --> 00:04:01,200
Josh:
which i think is really cool and super super exciting yeah

66
00:04:01,200 --> 00:04:05,860
Ejaaz:
Um i was reading up on uh a few of like the important stats about this new chip

67
00:04:05,860 --> 00:04:11,500
Ejaaz:
and I read that it can basically service up to 250 billion parameter models,

68
00:04:11,680 --> 00:04:15,300
Ejaaz:
which isn't the largest models we've ever quoted on this show, Josh.

69
00:04:15,460 --> 00:04:18,100
Ejaaz:
We've been talking about like trillion parameter models coming from like the

70
00:04:18,100 --> 00:04:20,620
Ejaaz:
likes of, I don't know, Kimi K2 out of China and stuff like that.

71
00:04:20,740 --> 00:04:25,020
Ejaaz:
But I think the important part here is the efficiency that comes from this chips.

72
00:04:25,120 --> 00:04:30,140
Ejaaz:
I believe this is like the cheapest silicon per power watt chip that would exist

73
00:04:30,140 --> 00:04:33,120
Ejaaz:
theoretically once they start mass producing it.

74
00:04:33,260 --> 00:04:42,660
Ejaaz:
I think it's slated for 2026 out of TSMC, which is close to another big AI hardware giant, Josh, right?

75
00:04:42,960 --> 00:04:48,140
Ejaaz:
And that is NVIDIA. And it got me thinking about NVIDIA as a whole,

76
00:04:48,320 --> 00:04:51,560
Ejaaz:
which basically has had the linchpin on any

77
00:04:51,560 --> 00:04:54,380
Ejaaz:
kind of AI GPU or chip that

78
00:04:54,380 --> 00:04:57,740
Ejaaz:
is needed from some of the biggest players right open ai anthropic

79
00:04:57,740 --> 00:05:00,700
Ejaaz:
um google they all use um and

80
00:05:00,700 --> 00:05:03,400
Ejaaz:
rely on nvidia this tiny company which is producing all these

81
00:05:03,400 --> 00:05:06,140
Ejaaz:
kinds of things not tiny company but is manufacturing all these

82
00:05:06,140 --> 00:05:09,480
Ejaaz:
kinds of things on a tiny island called taiwan uh

83
00:05:09,480 --> 00:05:12,420
Ejaaz:
via tsmc and so there's never really been

84
00:05:12,420 --> 00:05:15,420
Ejaaz:
some kind of major competitor there have been a few

85
00:05:15,420 --> 00:05:18,740
Ejaaz:
attempts at this out of china because they haven't had access to

86
00:05:18,740 --> 00:05:21,540
Ejaaz:
nvidia stuff um but they've never really gotten

87
00:05:21,540 --> 00:05:26,040
Ejaaz:
anywhere and now we're seeing a real challenger to the throne which hasn't materialized

88
00:05:26,040 --> 00:05:31,260
Ejaaz:
just yet but has plans to do that and i was checking out this video um which

89
00:05:31,260 --> 00:05:34,940
Ejaaz:
uh really kind of like summarizes what elon is trying to pull off here and maybe

90
00:05:34,940 --> 00:05:40,400
Ejaaz:
we can play it super quickly what constellation of all these little components fits.

91
00:05:41,090 --> 00:05:45,290
Ejaaz:
And then he figures out, oh, this Optimus Teams is this, and RoboTex is this, and this, and this.

92
00:05:45,430 --> 00:05:48,990
Ejaaz:
Oh, we actually don't need this NVIDIA thing. We can build this into one chip,

93
00:05:49,110 --> 00:05:53,290
Ejaaz:
and then it matches the Samsung constellation, and then we tweak it a little bit more.

94
00:05:53,470 --> 00:05:53,670
Josh:
Boom.

95
00:05:53,970 --> 00:05:58,330
Ejaaz:
And that's what I mean with reality engineering. So what he's referencing here

96
00:05:58,330 --> 00:06:03,370
Ejaaz:
is the current setup for Tesla and all the devices and machines that they build,

97
00:06:03,870 --> 00:06:08,990
Ejaaz:
is it all requires different chips with slightly different architectures, right?

98
00:06:08,990 --> 00:06:12,350
Ejaaz:
And as you mentioned earlier, Josh, since kind of like August,

99
00:06:12,610 --> 00:06:16,870
Ejaaz:
at least that's been public knowledge, he being Elon has started merging all

100
00:06:16,870 --> 00:06:19,990
Ejaaz:
of his teams that are building these different kind of chip architectures into

101
00:06:19,990 --> 00:06:23,250
Ejaaz:
one team to focus on creating one mega chip.

102
00:06:23,370 --> 00:06:27,330
Ejaaz:
And call me a low IQ person, but the thing that immediately pops into my head

103
00:06:27,330 --> 00:06:31,750
Ejaaz:
is the Iron Man kind of like sparkling glass that's in the center of his chest.

104
00:06:32,070 --> 00:06:36,350
Ejaaz:
And I'm thinking of this all powerful core that basically can power anything

105
00:06:36,350 --> 00:06:40,930
Ejaaz:
from his robotaxis to his Optimus Prime robots, which are being teased and stuff

106
00:06:40,930 --> 00:06:43,270
Ejaaz:
like that. And I think that that is super powerful.

107
00:06:43,670 --> 00:06:48,010
Ejaaz:
Building that in-house is going to be a huge investment and probably a very

108
00:06:48,010 --> 00:06:52,050
Ejaaz:
risky endeavor, but probably there's no one better for the job than Elon.

109
00:06:52,270 --> 00:06:56,210
Ejaaz:
I mean, Josh, do you think similarly or am I kind of like being too much in

110
00:06:56,210 --> 00:06:57,290
Ejaaz:
my fantasy brain right now?

111
00:06:57,630 --> 00:06:59,830
Josh:
No, no, that's right. That's right. They're going to do it. They're going to

112
00:06:59,830 --> 00:07:02,290
Josh:
do it. It's going to be remarkable. AI5 is going to be

113
00:07:02,730 --> 00:07:05,510
Josh:
pretty great um in terms of raw compute about eight

114
00:07:05,510 --> 00:07:08,570
Josh:
times more powerful than ai4 ai6 i believe

115
00:07:08,570 --> 00:07:11,410
Josh:
is um what has been teased as being the real game changer

116
00:07:11,410 --> 00:07:14,530
Josh:
that's where we're really going to see the world start to shift that's probably

117
00:07:14,530 --> 00:07:17,610
Josh:
going to be the one at mass scale with optimus that's going to be the one where you

118
00:07:17,610 --> 00:07:20,470
Josh:
have full robotaxes all around the world i do think

119
00:07:20,470 --> 00:07:23,910
Josh:
that i might disagree in one point in the sense that i don't

120
00:07:23,910 --> 00:07:26,730
Josh:
know if it's an nvidia killer or if it's going to take any sort of market

121
00:07:26,730 --> 00:07:29,650
Josh:
share from nvidia outside of just the sales that tesla gives

122
00:07:29,650 --> 00:07:32,350
Josh:
to nvidia because if you're if you're a company all you really

123
00:07:32,350 --> 00:07:35,210
Josh:
want is gpus and all you really want is compute

124
00:07:35,210 --> 00:07:38,710
Josh:
and intelligence so if tesla's able to create an

125
00:07:38,710 --> 00:07:41,570
Josh:
infinite amount or if they're able to scale this vertically and create

126
00:07:41,570 --> 00:07:44,350
Josh:
more than they're able to buy from nvidia they'll probably just

127
00:07:44,350 --> 00:07:47,270
Josh:
keep them for themselves so maybe nvidia loses the purchase order from

128
00:07:47,270 --> 00:07:50,410
Josh:
tesla but i don't see any reason for tesla to actually sell them to anybody

129
00:07:50,410 --> 00:07:54,230
Josh:
else because it's such a benefit to have that much compute power under one roof

130
00:07:54,230 --> 00:07:57,510
Josh:
and for it to be maximally efficient so not only are they going to have a ton

131
00:07:57,510 --> 00:08:00,610
Josh:
more chips but the chips are going to be a lot more energy efficient power if

132
00:08:00,610 --> 00:08:04,370
Josh:
like all of the the things will just be much much better so

133
00:08:04,370 --> 00:08:07,390
Ejaaz:
So i see your point of view i think why i

134
00:08:07,390 --> 00:08:10,550
Ejaaz:
would argue against that is elon seems to

135
00:08:10,550 --> 00:08:17,030
Ejaaz:
exhibit a behavior where if something isn't built to his kind of perfect specification

136
00:08:17,030 --> 00:08:21,890
Ejaaz:
he gets super annoyed and just builds it in-house right if he can't like outsource

137
00:08:21,890 --> 00:08:25,870
Ejaaz:
that in any way we saw that last week when we covered uh starlink's acquisition

138
00:08:25,870 --> 00:08:29,750
Ejaaz:
of a company which basically gives them access to crazy spectrum broadband,

139
00:08:29,750 --> 00:08:32,250
Ejaaz:
which will now allow all his satellites,

140
00:08:32,410 --> 00:08:34,890
Ejaaz:
which he's eventually gonna be releasing in space,

141
00:08:35,190 --> 00:08:40,590
Ejaaz:
to beam down lasers to our cell phones and get access to premium 5G and internet

142
00:08:40,590 --> 00:08:42,490
Ejaaz:
access wherever the hell you are, right?

143
00:08:42,690 --> 00:08:44,970
Ejaaz:
If you haven't seen that episode, definitely go and check that out.

144
00:08:45,110 --> 00:08:49,390
Ejaaz:
So I have a feeling that he might get annoyed enough to just go whole hog and

145
00:08:49,390 --> 00:08:52,790
Ejaaz:
go for Nvidia's throat, but I agree, like, if there's no necessary business

146
00:08:52,790 --> 00:08:53,650
Ejaaz:
means, he probably won't.

147
00:08:54,140 --> 00:08:58,220
Josh:
Yeah. If I'm Tesla and I have infinite compute power that I can generate myself

148
00:08:58,220 --> 00:09:00,020
Josh:
by making it, I'm taking all of it.

149
00:09:00,200 --> 00:09:03,440
Josh:
I want all the chips. I'm not giving that to anybody else. That's my new competitive advantage.

150
00:09:03,840 --> 00:09:06,480
Josh:
And similar to Apple, like now you see it in all of their mobile,

151
00:09:07,080 --> 00:09:11,280
Josh:
like the new iPhone, a single threaded processor from the new iPhone is equivalent

152
00:09:11,280 --> 00:09:16,420
Josh:
to the M4 Max in a phone, which is just unbelievable, like efficiency and compute

153
00:09:16,420 --> 00:09:17,760
Josh:
power in such a small device.

154
00:09:17,900 --> 00:09:22,020
Josh:
So for Tesla to give that up seems unlikely, but oh my God, what a huge unlock

155
00:09:22,020 --> 00:09:25,200
Josh:
it's going to be when they actually make it. And this isn't even the only Tesla news this week.

156
00:09:25,440 --> 00:09:29,320
Josh:
I mean, you teased the Megablock, right? The new energy thing that they have going on.

157
00:09:29,540 --> 00:09:32,780
Josh:
Do you want to just kind of explain what's going on with this new technology?

158
00:09:32,860 --> 00:09:35,200
Josh:
Because this, to me, this is even more exciting than the chips. This is pretty cool.

159
00:09:35,960 --> 00:09:38,620
Ejaaz:
Well, Josh, I kind of want you to explain it for me. I mean,

160
00:09:38,680 --> 00:09:41,740
Ejaaz:
look at these images that are like staring at me in the face, right? Okay.

161
00:09:41,880 --> 00:09:46,500
Ejaaz:
Okay. So let me give you my left curve take, right?

162
00:09:46,800 --> 00:09:54,280
Ejaaz:
Cool. These look like some 1970s transformer type uh gizmos here it looks like

163
00:09:54,280 --> 00:09:58,400
Ejaaz:
an old school data center it looks boring as hell josh what am i what am i staring

164
00:09:58,400 --> 00:10:02,200
Ejaaz:
at what's a megawatt per hour and why is it important.

165
00:10:02,200 --> 00:10:04,920
Josh:
Yeah okay so we can maybe define a few things that

166
00:10:04,920 --> 00:10:07,940
Josh:
are going to be important for this segment which is a kilowatt which is shown

167
00:10:07,940 --> 00:10:10,560
Josh:
kw and that's the rate it's kind of like you can think

168
00:10:10,560 --> 00:10:13,300
Josh:
of each as like miles per hour but for power and then

169
00:10:13,300 --> 00:10:16,460
Josh:
we have kilowatt hours which are kwh and that's

170
00:10:16,460 --> 00:10:19,060
Josh:
the amount of energy which is kind of like gallons of fuel for a

171
00:10:19,060 --> 00:10:21,980
Josh:
car and then we have megawatts and megawatt hours which

172
00:10:21,980 --> 00:10:25,660
Josh:
are just a thousand times multiple on those so kilowatts megawatts kilowatt

173
00:10:25,660 --> 00:10:28,500
Josh:
hours megawatt hours and the easy anchors are kind of like

174
00:10:28,500 --> 00:10:32,140
Josh:
a phone battery has 10 watt hours which is 0.01 kilowatt

175
00:10:32,140 --> 00:10:34,960
Josh:
hours a battery on an electric car like a tesla

176
00:10:34,960 --> 00:10:38,020
Josh:
has 75 kilowatt hours and so on and so forth so the average

177
00:10:38,020 --> 00:10:41,080
Josh:
home uses about 855 of these kilowatt hours

178
00:10:41,080 --> 00:10:44,020
Josh:
per month that's about 28 and a half kilowatt hours per day about

179
00:10:44,020 --> 00:10:47,040
Josh:
1.2 kilowatts of continuous draw so that's

180
00:10:47,040 --> 00:10:50,740
Josh:
kind of how we can frame all of this um megablock did

181
00:10:50,740 --> 00:10:53,880
Josh:
a really cool thing where they modularized these battery

182
00:10:53,880 --> 00:10:59,220
Josh:
packs so within a megablock is a megapack a megapack has a bunch of energy stored

183
00:10:59,220 --> 00:11:05,020
Josh:
inside and it is able to be deployed at scale to subsidize the grid so what

184
00:11:05,020 --> 00:11:08,700
Josh:
the megablock is is really just you just you could think of like a small nuclear

185
00:11:08,700 --> 00:11:11,920
Josh:
reactor almost except it doesn't actually generate its own energy it It just

186
00:11:11,920 --> 00:11:12,820
Josh:
distributes the energy.

187
00:11:13,160 --> 00:11:16,260
Josh:
And a lot of the problem with the grid that we have today is that

188
00:11:16,690 --> 00:11:19,790
Josh:
We're generating a lot of energy. We're just not using a lot of it.

189
00:11:19,910 --> 00:11:22,790
Josh:
And there's actually a chart that I want to show in one of these videos,

190
00:11:23,070 --> 00:11:26,390
Josh:
if you don't mind pulling it up, which shows the amount of energy that we use

191
00:11:26,390 --> 00:11:28,270
Josh:
relative to the amount that we produce.

192
00:11:28,470 --> 00:11:32,030
Josh:
So you could see, EJS, in this chart, the amount of energy that we're actually

193
00:11:32,030 --> 00:11:35,650
Josh:
using relative to the amount that we have, it's like 30 to 50%.

194
00:11:35,650 --> 00:11:38,070
Josh:
We're using such a low amount of energy.

195
00:11:38,170 --> 00:11:40,470
Josh:
And that's just because we don't have good efficiency in the grid.

196
00:11:41,230 --> 00:11:44,710
Josh:
The way it works, and what the person in this presentation is describing right

197
00:11:44,710 --> 00:11:47,410
Josh:
now, is you could see how it shifts up a little bit.

198
00:11:47,550 --> 00:11:51,190
Josh:
The grid has to accommodate for the highest point of energy consumption,

199
00:11:51,390 --> 00:11:52,630
Josh:
no matter where you are in the world.

200
00:11:52,810 --> 00:11:56,810
Josh:
So if there's a huge blip in energy demand, they need to meet that energy demand.

201
00:11:56,890 --> 00:11:59,850
Josh:
So there's no blackouts, which means they have to pour a lot of resources into one place.

202
00:12:00,130 --> 00:12:02,890
Josh:
And it's not necessarily the most efficient place. It's just the place that

203
00:12:02,890 --> 00:12:05,010
Josh:
has a lot of demand draw at one specific time.

204
00:12:05,150 --> 00:12:09,030
Josh:
What these mega packs do and what these mega blocks do is they subsidize that demand.

205
00:12:09,210 --> 00:12:12,290
Josh:
So I'm getting all my mega blocks and mega packs confused but

206
00:12:12,290 --> 00:12:15,150
Josh:
basically what it is is you can drop one

207
00:12:15,150 --> 00:12:17,950
Josh:
of these things into an area and it could subsidize the grid

208
00:12:17,950 --> 00:12:20,630
Josh:
in a way that will smooth out that bumpy line that

209
00:12:20,630 --> 00:12:23,270
Josh:
we just saw so one of the big things and the big problems that we

210
00:12:23,270 --> 00:12:26,590
Josh:
have with current ai data training centers for example is

211
00:12:26,590 --> 00:12:29,570
Josh:
getting the amount of energy that is required to these places is

212
00:12:29,570 --> 00:12:33,050
Josh:
nearly impossible it's so hard and it's not because the energy doesn't exist

213
00:12:33,050 --> 00:12:36,410
Josh:
it's because we don't have an efficient way to get it to the place so these

214
00:12:36,410 --> 00:12:41,130
Josh:
mega blocks exist as a way to store that energy and then release it onto the

215
00:12:41,130 --> 00:12:44,910
Josh:
grid in a way that smooths out those curves and allows you to optimize for more

216
00:12:44,910 --> 00:12:47,230
Josh:
energy consumption on a regular basis.

217
00:12:47,430 --> 00:12:53,270
Ejaaz:
That's super cool. I was thinking of other ways that I can kind of like understand

218
00:12:53,270 --> 00:12:57,050
Ejaaz:
the magnitude of power that's going to be kind of like fueled through one of these.

219
00:12:57,230 --> 00:13:00,230
Ejaaz:
And you mentioned that these mega blocks are basically going to be composed

220
00:13:00,230 --> 00:13:01,950
Ejaaz:
of these things called mega packs, right?

221
00:13:02,330 --> 00:13:06,230
Ejaaz:
And then I was like, well, okay, how powerful are of these individual mega packs.

222
00:13:06,390 --> 00:13:13,570
Ejaaz:
And I was looking it up and one mega pack alone is enough to power 170 average

223
00:13:13,570 --> 00:13:15,850
Ejaaz:
American homes for about a day.

224
00:13:16,500 --> 00:13:20,420
Ejaaz:
Which is insane, right? And then I was like, okay, well, what is that compared

225
00:13:20,420 --> 00:13:23,600
Ejaaz:
to the amount of power that's in a single Tesla car, which, you know,

226
00:13:23,700 --> 00:13:26,500
Ejaaz:
you own one of these, Josh? And it's 67 of them.

227
00:13:26,760 --> 00:13:31,280
Ejaaz:
So I'm trying to think about like how this kind of like stacks upon each other

228
00:13:31,280 --> 00:13:36,160
Ejaaz:
as like Elon like keeps kind of like rolling all these like energy trade-offs

229
00:13:36,160 --> 00:13:38,380
Ejaaz:
into one super efficient thing.

230
00:13:38,620 --> 00:13:42,180
Ejaaz:
And I kind of like envision a world where you can kind of create a city in the

231
00:13:42,180 --> 00:13:45,880
Ejaaz:
middle of a desert or somewhere where it's very resource scarce and have a fully

232
00:13:45,880 --> 00:13:50,280
Ejaaz:
functioning society of people out there that can just kind of like survive on

233
00:13:50,280 --> 00:13:51,220
Ejaaz:
their own with these Megapacks.

234
00:13:51,360 --> 00:13:55,040
Ejaaz:
It also gets me thinking about eventually when we end up with,

235
00:13:55,040 --> 00:13:59,240
Ejaaz:
you know, proficient space travel and we end up landing on different planets

236
00:13:59,240 --> 00:14:02,540
Ejaaz:
and stuff, we could probably be able to power all of this from some of these

237
00:14:02,540 --> 00:14:04,760
Ejaaz:
Megapacks, which is super, super cool in my opinion.

238
00:14:05,460 --> 00:14:09,220
Josh:
Yeah, and it's not to discount the fact that we still need to power these Megablocks.

239
00:14:09,400 --> 00:14:13,240
Josh:
So the Megablocks are just batteries. They don't actually generate any net new energy.

240
00:14:13,240 --> 00:14:16,140
Josh:
But in the case that you can supply energy to these

241
00:14:16,140 --> 00:14:18,740
Josh:
blocks well they will take care of all the rest and like you

242
00:14:18,740 --> 00:14:21,460
Josh:
said ejes where there's a place where there's not a lot of infrastructure available

243
00:14:21,460 --> 00:14:25,440
Josh:
or a lot of infrastructure gets wiped out you could just plug these into a electricity

244
00:14:25,440 --> 00:14:29,300
Josh:
generating source and it will take care of the rest and it's a really cool compact

245
00:14:29,300 --> 00:14:33,500
Josh:
thing i mean it's four megapacks they turn them into one block and then that

246
00:14:33,500 --> 00:14:36,280
Josh:
one block provides about 20 megawatt hours per block

247
00:14:36,860 --> 00:14:40,040
Josh:
It's a huge amount. And the speed at which they're able to actually deploy these,

248
00:14:40,120 --> 00:14:41,440
Josh:
I think, is one of the most interesting things.

249
00:14:41,600 --> 00:14:45,120
Josh:
Where in the presentation, they said they can get one of these deployed in 20 days.

250
00:14:45,380 --> 00:14:50,000
Josh:
And that means with enough of them, you can get up to a gigawatt of energy in 20 days.

251
00:14:50,140 --> 00:14:53,240
Josh:
And if you'll remember from our past episodes, a few of the data centers like

252
00:14:53,240 --> 00:14:57,660
Josh:
OpenAI's Giants data center or XAI's Colossus data center, they're actually

253
00:14:57,660 --> 00:15:01,500
Josh:
targeting for one gigawatt or two gigawatt of energy,

254
00:15:01,680 --> 00:15:05,480
Josh:
which is a huge amount that cannot be deployed very, very quickly with these megablocks.

255
00:15:05,480 --> 00:15:09,320
Josh:
So it's a really powerful way of not only subsidizing the grid,

256
00:15:09,480 --> 00:15:12,560
Josh:
but unlocking energy that we previously would not have access to.

257
00:15:12,760 --> 00:15:15,440
Josh:
I think a lot of the times we say, and I've been guilty of this,

258
00:15:15,520 --> 00:15:16,980
Josh:
like we don't have enough energy.

259
00:15:17,140 --> 00:15:21,240
Josh:
And the reality is we don't, but we could have a lot more if we just distribute

260
00:15:21,240 --> 00:15:25,600
Josh:
it properly because we are losing half of it, almost half of it to just inefficiencies

261
00:15:25,600 --> 00:15:27,760
Josh:
in the grid. And this will hopefully smooth that out over time.

262
00:15:28,540 --> 00:15:32,760
Ejaaz:
Being a show where we discuss AI a lot, particularly AI models,

263
00:15:32,780 --> 00:15:39,880
Ejaaz:
and we've mentioned how much energy and compute it requires to pre-train some of these models, right?

264
00:15:40,540 --> 00:15:43,900
Ejaaz:
I'm wondering whether there are more efficient ways that we can train models

265
00:15:43,900 --> 00:15:48,380
Ejaaz:
using these mega blocks or these mega packs, whichever one you want to consider.

266
00:15:48,780 --> 00:15:54,140
Ejaaz:
I saw someone kind of like have a good take on X the other day where he basically said,

267
00:15:54,260 --> 00:15:58,920
Ejaaz:
at nighttime, typically, it's harder to train some of these or pre-train some

268
00:15:58,920 --> 00:16:02,460
Ejaaz:
of these AI models because it's not as efficient at nighttime for many different

269
00:16:02,460 --> 00:16:04,160
Ejaaz:
reasons. for example, it might be cooler.

270
00:16:04,660 --> 00:16:08,760
Ejaaz:
And if you could supply some of these mega packs or these mega blocks and have

271
00:16:08,760 --> 00:16:13,080
Ejaaz:
that continually train certain models, I think we'll start to see some of these

272
00:16:13,080 --> 00:16:15,840
Ejaaz:
things in Colossus 2 was the point I was going to make.

273
00:16:15,960 --> 00:16:18,860
Ejaaz:
And I'm curious whether there is any kind of like overlap with

274
00:16:19,240 --> 00:16:24,100
Ejaaz:
elon's new data center that he's building out where is it now is it in texas.

275
00:16:24,100 --> 00:16:26,040
Josh:
I believe it's tennessee tennessee

276
00:16:26,040 --> 00:16:30,840
Ejaaz:
Okay cool yeah i'm curious whether he might start implementing a bunch of these

277
00:16:30,840 --> 00:16:34,900
Ejaaz:
different components that he's building with tesla into some of his ai driven efforts.

278
00:16:34,900 --> 00:16:37,680
Josh:
Yeah i'm sure they will be because if you remember i think

279
00:16:37,680 --> 00:16:40,780
Josh:
i forget what percentage of it was but almost all the electricity that

280
00:16:40,780 --> 00:16:44,100
Josh:
went into the cloxus one structure in the first factory

281
00:16:44,100 --> 00:16:46,860
Josh:
was from mega packs and a

282
00:16:46,860 --> 00:16:49,720
Josh:
mega block is just a juiced up steroid version

283
00:16:49,720 --> 00:16:52,420
Josh:
of that um and that's mostly because i mean in

284
00:16:52,420 --> 00:16:55,980
Josh:
their case the the issue was that grid power isn't always completely

285
00:16:55,980 --> 00:16:59,860
Josh:
smooth and when you're powering up and down a hundred thousand gpus at once

286
00:16:59,860 --> 00:17:03,880
Josh:
the energy swings are very large very quick and they needed a way to smooth

287
00:17:03,880 --> 00:17:07,160
Josh:
that out so the mega pack solved that now with a mega block it's just like that

288
00:17:07,160 --> 00:17:10,500
Josh:
on steroids it is a tremendous amount of energy that you can smooth out into

289
00:17:10,500 --> 00:17:15,080
Josh:
these data centers that doesn't mess up a single training rod and there's this crazy sat,

290
00:17:15,400 --> 00:17:19,220
Josh:
I mean, I mentioned they can be rolled out in 20 days, but 200 of these sites

291
00:17:19,220 --> 00:17:22,400
Josh:
would enable a 20% increase in total U.S.

292
00:17:22,580 --> 00:17:26,140
Josh:
Electricity generation without building a single new thermal generator.

293
00:17:26,300 --> 00:17:28,340
Josh:
It's just by better utilizing what we already have.

294
00:17:28,820 --> 00:17:32,620
Josh:
So if these things start to get deployed at scale, it can seriously make a huge

295
00:17:32,620 --> 00:17:35,540
Josh:
impact in the amount of energy that we have and the cost per kilowatt more efficient,

296
00:17:35,760 --> 00:17:37,260
Josh:
more importantly, that we always talk about,

297
00:17:37,660 --> 00:17:41,440
Josh:
without actually needing to spin up more solar farms

298
00:17:41,440 --> 00:17:44,300
Josh:
or gas turbines or whatever type of energy source we

299
00:17:44,300 --> 00:17:47,360
Josh:
need because we have it it's just a matter of just deploying

300
00:17:47,360 --> 00:17:50,140
Josh:
it more efficiently and that's exactly what mega pack does and that's why

301
00:17:50,140 --> 00:17:52,860
Josh:
it's really it's so important because if this gets rolled out

302
00:17:52,860 --> 00:17:56,940
Josh:
at scale we could see the cost per kilowatt drop very very

303
00:17:56,940 --> 00:17:59,640
Josh:
much and also they can be rolled as data centers very

304
00:17:59,640 --> 00:18:02,500
Josh:
very quickly so in the case that someone needs a lot of power well they don't

305
00:18:02,500 --> 00:18:05,740
Josh:
have to go through hell and back to get some energy there they could just roll

306
00:18:05,740 --> 00:18:09,180
Josh:
one of these bad boys out and and juice it all up okay so you just we kind of

307
00:18:09,180 --> 00:18:12,660
Josh:
talked about why this is so important but i understand that they did some pretty

308
00:18:12,660 --> 00:18:17,760
Josh:
amazing things in order to achieve this that you can hopefully walk us through well

309
00:18:17,760 --> 00:18:22,880
Ejaaz:
Okay listen i'm a massive fan of pokemon and the main reason why i was is because

310
00:18:22,880 --> 00:18:27,080
Ejaaz:
these pokemon cards have some of the coolest stats ever and it seems like we

311
00:18:27,080 --> 00:18:32,180
Ejaaz:
have it for the mega pack 3 so introducing the pokemon card version of the Megapack 3,

312
00:18:32,620 --> 00:18:36,500
Ejaaz:
5 megawatts per hour of usable AC energy.

313
00:18:36,780 --> 00:18:39,860
Ejaaz:
Its weight is a casual 86,000 pounds.

314
00:18:40,360 --> 00:18:47,140
Ejaaz:
This I found super interesting. It operates between minus 40 degrees Celsius to 60 degrees Celsius.

315
00:18:47,360 --> 00:18:50,220
Ejaaz:
So basically, it's impervious to any temperature fluctuation.

316
00:18:50,360 --> 00:18:54,840
Ejaaz:
You can have it pretty much manufactured anyway. That's a humongous range.

317
00:18:55,120 --> 00:18:59,960
Ejaaz:
And 75% of the mass of a Megapack 3 is battery Now,

318
00:19:00,120 --> 00:19:05,600
Ejaaz:
having come hot off the Apple episode last week, Josh, where that stat basically

319
00:19:05,600 --> 00:19:09,320
Ejaaz:
rings true for the new iPhone Air that's going to be delivered to my building

320
00:19:09,320 --> 00:19:11,120
Ejaaz:
in about a couple of days time,

321
00:19:11,360 --> 00:19:14,420
Ejaaz:
I kind of see a trend developing here, which is pretty cool.

322
00:19:14,600 --> 00:19:18,120
Ejaaz:
A single module weighs as much as a Cybertruck.

323
00:19:18,570 --> 00:19:21,450
Ejaaz:
These things are much larger than I thought. I kind of thought this might be

324
00:19:21,450 --> 00:19:27,210
Ejaaz:
like a jet pack type thing that astronauts might wear, but maybe that's in V3.

325
00:19:28,090 --> 00:19:33,530
Ejaaz:
And it will partly use battery cells from Tesla's new seven gigawatt an hour

326
00:19:33,530 --> 00:19:38,030
Ejaaz:
LFP battery facility in Nevada coming late online in 2025.

327
00:19:38,250 --> 00:19:42,490
Ejaaz:
That's cool because we've spoken about this new battery facility quite a bunch

328
00:19:42,490 --> 00:19:45,770
Ejaaz:
of times on this show, but we've never really kind of like understood maybe

329
00:19:45,770 --> 00:19:48,210
Ejaaz:
the strategy of where he's feeding a bunch of these cells.

330
00:19:48,570 --> 00:19:51,830
Ejaaz:
This is probably one of the major outlets that will get there as well.

331
00:19:52,410 --> 00:19:57,970
Ejaaz:
Super cool. And I kind of understand now why Elon is buying a billion dollars

332
00:19:57,970 --> 00:20:02,570
Ejaaz:
worth of additional shares this morning that news broke. Pretty hilarious.

333
00:20:02,870 --> 00:20:06,150
Josh:
Stock price is doing pretty good as well. I think people are excited.

334
00:20:06,470 --> 00:20:10,290
Josh:
This is one of those cases where if any company in the world were to achieve

335
00:20:10,290 --> 00:20:12,650
Josh:
any of these things, it would be massive news.

336
00:20:13,010 --> 00:20:15,690
Josh:
And there's one company that's at the forefront doing all of these things.

337
00:20:16,190 --> 00:20:19,270
Josh:
I mean, the AI5 chip is an incredible feat of engineering.

338
00:20:19,430 --> 00:20:22,750
Josh:
It's not done yet. They're finishing the design process. But assuming they do

339
00:20:22,750 --> 00:20:25,050
Josh:
deliver like they always do, maybe it'll probably be late.

340
00:20:25,150 --> 00:20:27,250
Josh:
But assuming they do, it's going to be a remarkable piece of tech.

341
00:20:27,350 --> 00:20:30,650
Josh:
And it's going to run a ton of the very first humanoid robots that we see walking around.

342
00:20:30,850 --> 00:20:34,570
Josh:
And now they're also rolling out this energy plan to subsidize the grid that's

343
00:20:34,570 --> 00:20:38,690
Josh:
been faltering and hasn't been upkept correctly.

344
00:20:38,950 --> 00:20:43,450
Josh:
And it's amazing to see the progress of some of these innovations and not only

345
00:20:43,450 --> 00:20:45,270
Josh:
how quickly they're rolling it out, but how effective they are.

346
00:20:45,270 --> 00:20:48,110
Josh:
I mean every single one of these new product rollouts is the best

347
00:20:48,110 --> 00:20:51,150
Josh:
in class and i don't see that changing anytime soon so

348
00:20:51,150 --> 00:20:54,990
Josh:
i it's a good time to be bullish it's a good time to be bullish on tesla on

349
00:20:54,990 --> 00:20:58,990
Josh:
the entire ecosystem around them on the xai team who also benefits from this

350
00:20:58,990 --> 00:21:02,510
Josh:
i mean there's a lot of all you start to see all the vertical integration how

351
00:21:02,510 --> 00:21:05,870
Josh:
it all feeds into each other and they all help each other in this this really

352
00:21:05,870 --> 00:21:08,110
Josh:
nice ecosystem and it gets me fired up i

353
00:21:10,210 --> 00:21:12,410
Josh:
These things, when they work at scale, will change the world,

354
00:21:12,750 --> 00:21:14,730
Josh:
particularly the megablock. I mean, this is huge.

355
00:21:15,130 --> 00:21:21,530
Ejaaz:
For a long time, I think the idea of electricity or rather like solar-powered

356
00:21:21,530 --> 00:21:24,230
Ejaaz:
electricity powering everything kind of seemed like a gimmick.

357
00:21:24,530 --> 00:21:28,390
Ejaaz:
And the major reason for that wasn't because solar wasn't powerful enough.

358
00:21:28,390 --> 00:21:30,450
Ejaaz:
It's because the distribution of it sucked.

359
00:21:30,630 --> 00:21:34,850
Ejaaz:
And now we have a vehicle that can clearly distribute this energy way more efficiently,

360
00:21:35,050 --> 00:21:37,290
Ejaaz:
50% of energy to be precise.

361
00:21:37,290 --> 00:21:41,950
Ejaaz:
So I'm super excited about this. And hopefully, if you will listen to this,

362
00:21:42,170 --> 00:21:46,570
Ejaaz:
this was educational for you and you learned something new about Elon Musk,

363
00:21:46,730 --> 00:21:52,230
Ejaaz:
who seems to be releasing paradigm-shifting technologies every couple of weeks now, Josh.

364
00:21:52,550 --> 00:21:57,070
Ejaaz:
I'm very bullish, Tesla. It is probably the stock that I own in the largest

365
00:21:57,070 --> 00:22:00,070
Ejaaz:
quantity outside of some of my crypto stuff. Oh, nice. Okay.

366
00:22:00,430 --> 00:22:05,250
Ejaaz:
Oh, yeah. Dude, how could I not be bullish? We talk about it about 50% of episodes. Oh, good.

367
00:22:05,290 --> 00:22:07,110
Josh:
Well, you never know who's actually on board, right?

368
00:22:07,230 --> 00:22:12,030
Ejaaz:
Like, I'm glad to hear this. No, I've got my skin, teeth, and bones in the game.

369
00:22:12,330 --> 00:22:16,410
Ejaaz:
So I'm super pumped about this and I'm excited to see where it goes.

370
00:22:16,550 --> 00:22:21,230
Ejaaz:
But again, if you think that there is a better topic that we could be discussing,

371
00:22:21,370 --> 00:22:25,850
Ejaaz:
if there's something that Elon has released or is talking about that we've missed

372
00:22:25,850 --> 00:22:29,990
Ejaaz:
and that you think we're not covering because our Tesla bags are consuming us

373
00:22:29,990 --> 00:22:32,150
Ejaaz:
and there's a competitor that needs to be described,

374
00:22:32,690 --> 00:22:35,050
Ejaaz:
we chat to us, our DMs are open, our comments are open.

375
00:22:35,290 --> 00:22:38,550
Ejaaz:
We also like hearing feedback, particularly the feedback that

376
00:22:38,550 --> 00:22:43,790
Ejaaz:
is spicy that disagrees with us that wants to hear different takes um josh and

377
00:22:43,790 --> 00:22:47,650
Ejaaz:
i uh we like to fight but mainly we agree with a lot of the stuff that we discuss

378
00:22:47,650 --> 00:22:51,730
Ejaaz:
we kind of discuss topics that we enjoy and if there's anything else that you

379
00:22:51,730 --> 00:22:55,170
Ejaaz:
would like to hear about please reach out like subscribe share with your friends

380
00:22:55,170 --> 00:22:58,310
Ejaaz:
and we will see you on the next one which.

381
00:22:58,310 --> 00:23:01,430
Josh:
Will probably be meta connect big hardware episode coming this week so don't

382
00:23:01,430 --> 00:23:05,350
Josh:
don't miss it it's gonna be a big one thank you guys for watching we'll see you guys in the next one

383
00:23:06,320 --> 00:23:13,550
Music:
Music