1
00:00:00,020 --> 00:00:04,120
Josh:
One of the most famous investors in the world, Peter Thiel, he wrote a book called Zero to One.

2
00:00:04,240 --> 00:00:07,500
Josh:
Some of you might have heard it, some of you may have not, but it's all about

3
00:00:07,500 --> 00:00:11,720
Josh:
monopolies and how much of an advantage having a monopoly has in the world.

4
00:00:11,800 --> 00:00:17,000
Josh:
Now, what we've seen recently is a company named NVIDIA reaching a $5 trillion

5
00:00:17,000 --> 00:00:22,220
Josh:
monopoly. It's the biggest monopoly to ever exist in business, and it's huge.

6
00:00:22,400 --> 00:00:26,320
Josh:
It's made tons of people, infinite amounts of money, but something is happening.

7
00:00:26,540 --> 00:00:31,020
Josh:
Something is wrong. It appears as if that monopoly is starting to slip out of their hands.

8
00:00:31,220 --> 00:00:35,080
Josh:
And in this episode, we're going to talk about why and what that means for the entire market.

9
00:00:35,220 --> 00:00:41,800
Josh:
I mean, when you're a $5 trillion asset, the size of your company makes a meaningful impact on the market.

10
00:00:41,940 --> 00:00:46,240
Josh:
So when a company like NVIDIA loses hundreds of billions of dollars over a short

11
00:00:46,240 --> 00:00:49,120
Josh:
period of time like the last month, we got to start paying attention to this

12
00:00:49,120 --> 00:00:50,780
Josh:
because it's dragging everything else down with it.

13
00:00:50,860 --> 00:00:53,620
Josh:
So up here on the screen, we have some charts that I want us to walk through

14
00:00:53,620 --> 00:00:57,200
Josh:
EJS. So if you could just kind of show us the difference between NVIDIA and

15
00:00:57,200 --> 00:01:01,340
Josh:
who we believe to be the competition that is responsible for knocking these

16
00:01:01,340 --> 00:01:03,200
Josh:
hundreds of billions of dollars off their market cap.

17
00:01:03,580 --> 00:01:07,140
Ejaaz:
Yeah, so this really scary looking red chart that you have in front of you is

18
00:01:07,140 --> 00:01:11,680
Ejaaz:
the last month's performance for NVIDIA stock. And it is down a staggering rate.

19
00:01:11,900 --> 00:01:18,340
Ejaaz:
13, almost 14%, which equates to over $500 billion of market cap loss.

20
00:01:18,760 --> 00:01:24,220
Ejaaz:
That's a billion with a B, which is just an insane amount of money for the largest

21
00:01:24,220 --> 00:01:27,380
Ejaaz:
company or stock company in the world to lose.

22
00:01:27,720 --> 00:01:31,480
Ejaaz:
So the obvious question that comes in is why and where's that market,

23
00:01:31,780 --> 00:01:33,680
Ejaaz:
where's that money kind of flowing towards?

24
00:01:33,880 --> 00:01:37,960
Ejaaz:
Well, I want to show you another chart, Josh, which is Google's chart.

25
00:01:37,960 --> 00:01:39,600
Ejaaz:
And do you notice a similarity?

26
00:01:39,900 --> 00:01:44,780
Ejaaz:
Over the last month, it is up almost the same amount in percentage market cap

27
00:01:44,780 --> 00:01:49,220
Ejaaz:
for a very peculiar reason, or maybe it's not so peculiar.

28
00:01:49,540 --> 00:01:52,160
Ejaaz:
Josh, have you heard of these? You've heard of these things called TPUs, right?

29
00:01:52,560 --> 00:01:54,000
Josh:
We're talking about this. Oh, there's no things called TPUs.

30
00:01:54,220 --> 00:01:57,680
Ejaaz:
Yeah. There's no things called TPUs. You know, Josh and I like to kind of go

31
00:01:57,680 --> 00:01:58,840
Ejaaz:
back and forth and discuss this a lot.

32
00:01:58,900 --> 00:02:02,980
Ejaaz:
In fact, we actually put out a bull episode on Google, which a bunch of you

33
00:02:02,980 --> 00:02:06,420
Ejaaz:
watched a few weeks back. and I don't want to be running victory laps here,

34
00:02:06,500 --> 00:02:09,460
Ejaaz:
but it turned out that Josh and I might have been onto something.

35
00:02:09,580 --> 00:02:14,400
Ejaaz:
But I want to dumb down what's going on here, showed by this really hilarious

36
00:02:14,400 --> 00:02:16,960
Ejaaz:
graphic or comic from the semi-analysis team.

37
00:02:17,260 --> 00:02:24,760
Ejaaz:
And it basically goes, Google came out with this new rock, new shiny rock called TPU version seven.

38
00:02:24,900 --> 00:02:29,820
Ejaaz:
It's basically their version of NVIDIA's GPUs, but it's built by themselves in-house.

39
00:02:29,960 --> 00:02:34,360
Ejaaz:
And it's actually really, really good. It gives you an average of 30% to 50%

40
00:02:34,360 --> 00:02:38,360
Ejaaz:
cost savings for the exact same performance or equivalent of an NVIDIA GPU.

41
00:02:38,740 --> 00:02:43,480
Ejaaz:
And it, in some cases, performs even better, up to one and a half to two times better, right?

42
00:02:43,900 --> 00:02:46,940
Ejaaz:
And so you've got NVIDIA in the leather jacket here on the right,

43
00:02:47,080 --> 00:02:52,760
Ejaaz:
which says, actually, my ROC, my GPU is faster, right? And Google's like, is that true?

44
00:02:52,980 --> 00:02:56,800
Ejaaz:
And then everyone ends up using Google's TPUs. And the point being made here

45
00:02:56,800 --> 00:03:03,680
Ejaaz:
is for the longest time, Josh, NVIDIA held the monopoly on the AI training and

46
00:03:03,680 --> 00:03:05,300
Ejaaz:
inference market via their GPUs.

47
00:03:05,400 --> 00:03:08,320
Ejaaz:
It's all anyone and everyone could use to train their models.

48
00:03:08,460 --> 00:03:12,720
Ejaaz:
It was the only option that they had. And now Google presents a real threat

49
00:03:12,720 --> 00:03:16,000
Ejaaz:
to NVIDIA's market dominance by presenting these TPUs.

50
00:03:16,080 --> 00:03:20,620
Ejaaz:
Now, initially, they use these TPUs to train their own model in-house.

51
00:03:20,820 --> 00:03:24,080
Ejaaz:
In fact, Google's never purchased NVIDIA GPUs to train their own models,

52
00:03:24,140 --> 00:03:27,560
Ejaaz:
and yet they have the best models, which tells us that the TPUs is something

53
00:03:27,560 --> 00:03:29,400
Ejaaz:
to really contend with NVIDIA's GPUs.

54
00:03:29,560 --> 00:03:33,680
Ejaaz:
But most recently, Josh, they've started selling these to other companies,

55
00:03:33,840 --> 00:03:38,060
Ejaaz:
supposedly, to train their own models. And so we're reaching a point now where

56
00:03:38,060 --> 00:03:40,480
Ejaaz:
Google and NVIDIA is a direct comparison.

57
00:03:40,600 --> 00:03:43,660
Ejaaz:
And we're seeing that in the market share dynamics that are happening now.

58
00:03:43,760 --> 00:03:48,400
Ejaaz:
You know, you've got NVIDIA losing up to $500 billion and Google gaining the

59
00:03:48,400 --> 00:03:51,120
Ejaaz:
same amount over the same month. It's just pretty insane to see.

60
00:03:51,600 --> 00:03:55,960
Josh:
Yeah, I can't stress this enough, how insane that delta between the two stock charges.

61
00:03:56,120 --> 00:04:00,760
Josh:
That's 26% combined in one month. So the market is really pricing in the fact

62
00:04:00,760 --> 00:04:02,840
Josh:
that this monopoly is starting to crumble.

63
00:04:03,000 --> 00:04:05,680
Josh:
Now, I think we have reasoning why that's not necessarily the case.

64
00:04:05,680 --> 00:04:06,700
Josh:
That will come later in the episode.

65
00:04:06,900 --> 00:04:09,780
Josh:
But for now, there is some real forces at play. I mean, EJs,

66
00:04:09,780 --> 00:04:11,540
Josh:
you were talking about them selling TPUs.

67
00:04:11,980 --> 00:04:15,920
Josh:
This morning, I saw Morgan Stanley make this announcement. They said about every

68
00:04:15,920 --> 00:04:20,120
Josh:
half a million TPUs Google sells can add about 13 billion in revenue.

69
00:04:20,380 --> 00:04:23,640
Josh:
And Google is planning to sell 12 million of those over the next two years.

70
00:04:23,760 --> 00:04:27,640
Josh:
So it's a significant amount of revenue that Google can expect to come down the pipe.

71
00:04:27,920 --> 00:04:31,600
Josh:
And it's the first time that we're really starting to see a legitimate competitor

72
00:04:31,600 --> 00:04:37,420
Josh:
to the NVIDIA GPU cluster. Now, that's not to say the GPU is done for.

73
00:04:37,540 --> 00:04:39,440
Josh:
There's a lot of competitive advantages to a GPU.

74
00:04:39,640 --> 00:04:42,600
Josh:
I suspect they're not going anywhere, but there is now...

75
00:04:42,950 --> 00:04:46,390
Josh:
Another market force at play. And when we see a market force kind of cutting

76
00:04:46,390 --> 00:04:50,510
Josh:
in, it starts to price cascade and the monopoly slowly starts to fade.

77
00:04:50,650 --> 00:04:55,350
Josh:
I do want to give a brief history lesson, E.J. on the history of Google and

78
00:04:55,350 --> 00:04:59,210
Josh:
their AI program, because what a lot of people don't understand is Google really

79
00:04:59,210 --> 00:05:02,850
Josh:
is the godfather of AI from the beginning of time to now.

80
00:05:03,030 --> 00:05:06,610
Josh:
And they've just had this problem where they haven't been able to actually build

81
00:05:06,610 --> 00:05:09,530
Josh:
products that scale or sell products to users.

82
00:05:09,590 --> 00:05:12,370
Josh:
But they've been doing this since all the way back in 2011.

83
00:05:12,690 --> 00:05:15,870
Josh:
And what we're seeing here is the original paper that a lot of people would

84
00:05:15,870 --> 00:05:20,350
Josh:
conceive to be the first time that a neural network proved that it was work.

85
00:05:20,630 --> 00:05:25,570
Josh:
And they trained a massive unsupervised model on 16,000 CPU cores.

86
00:05:25,630 --> 00:05:29,050
Josh:
This was before GPUs existed on random YouTube frames.

87
00:05:29,230 --> 00:05:33,590
Josh:
They didn't use any labels. They use no supervision. And then one neuron spontaneously

88
00:05:33,590 --> 00:05:35,430
Josh:
learned the concept of a cat.

89
00:05:35,610 --> 00:05:40,310
Josh:
So this seems so stupid. It's like, oh my God, it can recognize a cat.

90
00:05:40,310 --> 00:05:44,190
Josh:
But this was the first time in history a machine was able to identify something

91
00:05:44,190 --> 00:05:45,730
Josh:
without explicitly written instructions.

92
00:05:46,010 --> 00:05:49,290
Josh:
And that moment inside Google set off a light bulb that eventually led to them

93
00:05:49,290 --> 00:05:53,070
Josh:
creating Google Brain, which was enough of a breakthrough for them to start

94
00:05:53,070 --> 00:05:54,930
Josh:
creating AI inside of their in-house system.

95
00:05:55,090 --> 00:05:58,530
Josh:
So EJS, if you remember Google Translate, which has been around seemingly forever,

96
00:05:58,850 --> 00:06:00,910
Josh:
Google Translate is a result of AI.

97
00:06:01,190 --> 00:06:05,390
Josh:
That was an early test implementation of Google Translate. And what that actually

98
00:06:05,390 --> 00:06:08,090
Josh:
enabled is they could suggest to sponsors...

99
00:06:08,740 --> 00:06:12,140
Josh:
Or advertisers on the platform, which companies are more likely to click,

100
00:06:12,260 --> 00:06:14,900
Josh:
which users are more likely to click. And that created the whole AdSense model.

101
00:06:15,140 --> 00:06:18,700
Josh:
It created, oftentimes when you type into Google search, it'll autocomplete for you.

102
00:06:18,980 --> 00:06:22,580
Josh:
These were all very early versions of AI before we even realized what AI was,

103
00:06:22,700 --> 00:06:25,500
Josh:
which led to the invention of the TPU almost nine years ago.

104
00:06:25,640 --> 00:06:28,940
Josh:
And the TPU is this vertically integrated chip that we see today taking over

105
00:06:28,940 --> 00:06:31,900
Josh:
basically the entire world, one company at a time now.

106
00:06:32,080 --> 00:06:34,920
Josh:
So they've been doing this. I mean, a lot of people don't realize the TPU has

107
00:06:34,920 --> 00:06:36,860
Josh:
been around for nine years now that they've been iterating.

108
00:06:36,980 --> 00:06:40,100
Josh:
We're currently up to version seven, which is the Ironwood TPU.

109
00:06:40,280 --> 00:06:43,560
Josh:
And it's just this incredible testament to the fact that Google actually has

110
00:06:43,560 --> 00:06:45,820
Josh:
been doing this for over a decade now, almost 15 years.

111
00:06:46,220 --> 00:06:50,360
Josh:
And we're finally starting to see the fruits of their labor grow and be exposed

112
00:06:50,360 --> 00:06:52,180
Josh:
in public markets. And my God, it's explosive.

113
00:06:52,400 --> 00:06:56,880
Ejaaz:
It is insane that, you know, they've been working on this for over a decade, right?

114
00:06:57,060 --> 00:07:00,200
Ejaaz:
And like that compounded value is really starting to show now because like,

115
00:07:00,440 --> 00:07:02,700
Ejaaz:
I'm guessing like everyone back in the day was just kind of like,

116
00:07:02,820 --> 00:07:04,260
Ejaaz:
what is this machine learning thing?

117
00:07:04,440 --> 00:07:07,280
Ejaaz:
Like, I can't imagine any kind of like a chatbot being beneficial to us

118
00:07:07,280 --> 00:07:10,680
Ejaaz:
and then fast forward to 2022 chat gpc goes viral

119
00:07:10,680 --> 00:07:13,400
Ejaaz:
and suddenly everyone's kind of raving about gpus and google's kind of

120
00:07:13,400 --> 00:07:17,400
Ejaaz:
like quietly smirking and smiling not buying any of nvidia's gpus being like

121
00:07:17,400 --> 00:07:20,740
Ejaaz:
hey we invested in this a decade ago and it's finally paying off which is just

122
00:07:20,740 --> 00:07:24,760
Ejaaz:
kind of uh insane to to think about and like josh like they came up with the

123
00:07:24,760 --> 00:07:26,160
Ejaaz:
transformer as well right which

124
00:07:26,160 --> 00:07:31,180
Ejaaz:
is like 2017 architecture exactly for this alarm so super cool to see.

125
00:07:31,360 --> 00:07:36,660
Ejaaz:
Now, I want to kind of like step aside and kind of frame the narrative for what

126
00:07:36,660 --> 00:07:37,720
Ejaaz:
we're about to discuss right now.

127
00:07:38,300 --> 00:07:41,920
Ejaaz:
We're talking about Google versus NVIDIA, and there's many different ways that

128
00:07:41,920 --> 00:07:43,180
Ejaaz:
we can kind of compare the two, right?

129
00:07:43,240 --> 00:07:47,900
Ejaaz:
The most obvious one is through TPUs versus GPUs that you mentioned.

130
00:07:48,300 --> 00:07:51,740
Ejaaz:
And one of the biggest questions that I think listeners have on their mind,

131
00:07:51,860 --> 00:07:55,860
Ejaaz:
Josh, is like, okay, well, if Google's going to compete with NVIDIA, where's the proof?

132
00:07:56,060 --> 00:07:58,000
Ejaaz:
Like, who are they selling this stuff to?

133
00:07:58,440 --> 00:08:00,880
Ejaaz:
Like, surely they can't be selling to any major competitors,

134
00:08:01,120 --> 00:08:02,660
Ejaaz:
right? Surely they can't be selling to major companies.

135
00:08:02,780 --> 00:08:07,220
Ejaaz:
So can they actually really compete? And I just have one article I want to share

136
00:08:07,220 --> 00:08:08,500
Ejaaz:
with you, one little tweet.

137
00:08:08,700 --> 00:08:13,840
Ejaaz:
You might have heard of Meta or Zuckerberg, who is rumored to be spending tens

138
00:08:13,840 --> 00:08:20,460
Ejaaz:
of billions of dollars in 2026 on Google's CPUs to train their llama or big

139
00:08:20,460 --> 00:08:22,720
Ejaaz:
llama models coming up in the future.

140
00:08:22,820 --> 00:08:27,180
Ejaaz:
Now, of course, you're not too unfamiliar with Meta's kind of progress recently,

141
00:08:27,400 --> 00:08:28,520
Ejaaz:
or rather their spending budget.

142
00:08:28,680 --> 00:08:33,960
Ejaaz:
They've spent, I think, to the effect of $25 billion this year just to hire talent,

143
00:08:34,120 --> 00:08:36,940
Ejaaz:
to train the model we haven't even seen the model uh to

144
00:08:36,940 --> 00:08:40,500
Ejaaz:
begin with so the fact that they're planning on using google's infrastructure

145
00:08:40,500 --> 00:08:43,380
Ejaaz:
uh supposedly to train their models is is

146
00:08:43,380 --> 00:08:46,800
Ejaaz:
no easy feat and so you might be wondering well like why why would they choose

147
00:08:46,800 --> 00:08:50,980
Ejaaz:
tpus over nvidia's gpus that's kind of like the standard kind of framework to

148
00:08:50,980 --> 00:08:54,800
Ejaaz:
to go down well i just want to show you the scaling chart which basically shows

149
00:08:54,800 --> 00:09:00,960
Ejaaz:
that uh the tpu which is google's um infrastructure their gpu versus the GB300,

150
00:09:01,260 --> 00:09:03,300
Ejaaz:
which is NVIDIA's latest GPU,

151
00:09:04,000 --> 00:09:06,460
Ejaaz:
There's a significant cost difference, right?

152
00:09:06,640 --> 00:09:12,560
Ejaaz:
If you were to use Google's TPUs, the IronWord TPU v7, you would save 30 to

153
00:09:12,560 --> 00:09:15,940
Ejaaz:
50%, depending on the amount of TPUs that you would use,

154
00:09:16,340 --> 00:09:21,220
Ejaaz:
in training your model. Now, when you consider that Meta's CapEx spend for the

155
00:09:21,220 --> 00:09:27,060
Ejaaz:
next year, I believe is going to be something along the lines of $67 to $85 billion.

156
00:09:27,760 --> 00:09:33,120
Ejaaz:
That is a lot of cost saving if you are able to use Google's TPU.

157
00:09:33,300 --> 00:09:37,880
Ejaaz:
So from an economical sense, it makes a hell of a load of sense, right?

158
00:09:38,220 --> 00:09:41,580
Ejaaz:
And then the other thing I was thinking about, Josh, is why would Meta kind

159
00:09:41,580 --> 00:09:45,580
Ejaaz:
of use Google's TPUs for their own systems, right? Aren't they directly competing?

160
00:09:45,880 --> 00:09:49,680
Ejaaz:
Well, there's a secret kind of detail that I learned about this.

161
00:09:49,780 --> 00:09:55,020
Ejaaz:
The way that Google's TPUs are designed makes it really performant for something

162
00:09:55,020 --> 00:09:56,940
Ejaaz:
called recommender systems.

163
00:09:57,320 --> 00:10:01,320
Ejaaz:
Now, a recommender system is the system that is used behind ads,

164
00:10:01,920 --> 00:10:03,600
Ejaaz:
behind social media algorithms.

165
00:10:04,180 --> 00:10:07,480
Ejaaz:
Meta is arguably one of the biggest companies which uses these things.

166
00:10:07,600 --> 00:10:12,700
Ejaaz:
So if they can have a hyper-performant TPU, train the AI model to use on their

167
00:10:12,700 --> 00:10:15,640
Ejaaz:
own social media platform that will make it inevitably better

168
00:10:16,270 --> 00:10:21,050
Ejaaz:
30 to 50 percent less cost, it seems like a complete no-brainer.

169
00:10:21,230 --> 00:10:25,710
Ejaaz:
And if you add this deal to the Anthropic deal that are purchasing 1 million

170
00:10:25,710 --> 00:10:28,870
Ejaaz:
TPUs from Google, as well as another deal that we're about to talk about,

171
00:10:29,070 --> 00:10:31,830
Ejaaz:
that's just insane. Don't you think that's pretty insane, right?

172
00:10:32,370 --> 00:10:35,230
Josh:
Yeah. And going back to that chart that you just brought up earlier,

173
00:10:35,450 --> 00:10:39,290
Josh:
which shows the cost difference, like if you're a big AI company spending billions on training models,

174
00:10:39,610 --> 00:10:42,510
Josh:
Google is now offering a system that can cost only 25 to 50

175
00:10:42,510 --> 00:10:45,810
Josh:
cents for every dollar you'd spend on NVIDIA's best hardware

176
00:10:45,810 --> 00:10:48,790
Josh:
and that is a huge deal because i mean

177
00:10:48,790 --> 00:10:51,670
Josh:
when it comes down to it cost per compute cost per unit of compute

178
00:10:51,670 --> 00:10:54,670
Josh:
and training is so large when you're at the scale of these companies

179
00:10:54,670 --> 00:11:00,030
Josh:
spending tens of hundreds of billions of dollars that 25 50 of the cost is massive

180
00:11:00,030 --> 00:11:03,070
Josh:
and granted like you said they're not good for everything but if you're a company

181
00:11:03,070 --> 00:11:06,430
Josh:
like meta who's building suggestion algorithms that's particularly good for

182
00:11:06,430 --> 00:11:10,270
Josh:
this is a no-brainer and it seems to me like now the only threshold will be

183
00:11:10,270 --> 00:11:14,370
Josh:
how quick can google actually create these manufacture them spin them up,

184
00:11:14,470 --> 00:11:17,630
Josh:
put them in server racks and get them online so people can start using them.

185
00:11:17,730 --> 00:11:20,730
Josh:
Because this unlocks a whole new use case for AI that we haven't seen in the

186
00:11:20,730 --> 00:11:22,470
Josh:
past that we'll see now because of

187
00:11:22,470 --> 00:11:26,250
Josh:
the lower cost and also the increased efficiency of these Ironwood TPUs.

188
00:11:26,650 --> 00:11:29,630
Josh:
And Google's innovating quick, man. I mean, each one of these TPUs is coming

189
00:11:29,630 --> 00:11:33,330
Josh:
out every single year and each one is significantly better than the last.

190
00:11:33,490 --> 00:11:40,870
Ejaaz:
Every 500k TPUs that Google sell adds 10% to their 2027 Google Cloud Rev and

191
00:11:40,870 --> 00:11:43,450
Ejaaz:
3% to their at 2027 earnings per share.

192
00:11:43,890 --> 00:11:47,270
Ejaaz:
That is insane. So they don't need to sell. Yeah, $13 billion.

193
00:11:47,470 --> 00:11:53,890
Ejaaz:
They don't need to sell near as much GPUs as NVIDIA sells. They just need to sell a couple hundred K.

194
00:11:54,230 --> 00:11:57,370
Ejaaz:
And if this is just one deal that they're cementing with Meta,

195
00:11:57,690 --> 00:12:01,210
Ejaaz:
can you imagine how much revenue they're just going to churn from this?

196
00:12:01,270 --> 00:12:05,930
Ejaaz:
It was rumored that the Anthropic deal, where they're selling around,

197
00:12:05,930 --> 00:12:10,630
Ejaaz:
I think it's a couple hundred TPUs to them, is going to earn them $50 billion.

198
00:12:11,680 --> 00:12:15,440
Ejaaz:
Next year just to train Anthropics kind of like next forward model.

199
00:12:15,580 --> 00:12:17,740
Ejaaz:
So just kind of insane to see.

200
00:12:17,960 --> 00:12:22,000
Ejaaz:
There is a counter thesis to this deal, which is, you know, I'm going to put

201
00:12:22,000 --> 00:12:27,320
Ejaaz:
my tinfoil hat on here, Josh, which is Meta's kind of doing this so that they

202
00:12:27,320 --> 00:12:29,480
Ejaaz:
can negotiate better terms with NVIDIA

203
00:12:29,980 --> 00:12:33,880
Ejaaz:
or AMD to kind of get like better chip deals saying, hey, look,

204
00:12:33,980 --> 00:12:36,980
Ejaaz:
we'll go with Google unless you guys give us a cheaper kind of route.

205
00:12:37,120 --> 00:12:39,100
Ejaaz:
I think this is kind of like conspiracy theory.

206
00:12:39,320 --> 00:12:42,140
Ejaaz:
I mean, the metrics around Google's GPUs kind of prove themselves.

207
00:12:42,480 --> 00:12:46,120
Ejaaz:
But it's just something to keep in mind. I don't want to get too much into my bold thesis here.

208
00:12:46,220 --> 00:12:49,000
Josh:
I was going through a lot of the deals that we're surfacing today.

209
00:12:49,460 --> 00:12:52,520
Josh:
One of them being with Foxconn and Google, where now Foxconn is responsible

210
00:12:52,520 --> 00:12:54,420
Josh:
for building 1,000 server racks a week.

211
00:12:54,640 --> 00:12:56,760
Josh:
Next year, they're doing 2,000 server racks a week.

212
00:12:57,060 --> 00:12:59,840
Josh:
There was an announcement earlier today where Google is now partnering with

213
00:12:59,840 --> 00:13:02,480
Josh:
AWS, the cloud server, to provide more infrastructure.

214
00:13:02,720 --> 00:13:06,820
Josh:
So we're starting to see, again, more of these deals that are happening around

215
00:13:06,820 --> 00:13:09,940
Josh:
the Google TPU world, which is super fascinating.

216
00:13:10,140 --> 00:13:13,260
Josh:
And then this leads to the final point, which is the head of AI infrastructure

217
00:13:13,260 --> 00:13:15,860
Josh:
in a meeting from a few months ago saying that,

218
00:13:16,340 --> 00:13:19,360
Josh:
Google must double AI compute every six months to meet its demand.

219
00:13:19,520 --> 00:13:22,860
Josh:
So there is no shortage of demand. There is no shortage of infrastructure.

220
00:13:23,100 --> 00:13:26,380
Josh:
There is no shortage of support to get these TPUs out to the world.

221
00:13:26,540 --> 00:13:29,640
Josh:
And what we're going to start to see is how big of an impact this really does

222
00:13:29,640 --> 00:13:32,780
Josh:
have on a company like NVIDIA now that there is someone else in the market.

223
00:13:32,880 --> 00:13:37,540
Josh:
There is a second seller for a company like Meta who wants to build massive AI systems.

224
00:13:37,720 --> 00:13:42,880
Ejaaz:
You could argue one of the most obvious bull signals to purchasing or investing

225
00:13:42,880 --> 00:13:47,860
Ejaaz:
in Google stock was the fact that Berkshire Hathaway bought a three and a half

226
00:13:47,860 --> 00:13:51,560
Ejaaz:
billion dollar stake in Google literally a couple of weeks ago.

227
00:13:51,700 --> 00:13:56,120
Ejaaz:
And then, funnily enough, the leaked information around them selling TPUs to

228
00:13:56,120 --> 00:14:00,080
Ejaaz:
Meta and them striking this deal with NATO kind of like surfaced, right?

229
00:14:00,340 --> 00:14:03,340
Ejaaz:
So there's a lot of momentum behind Google right now.

230
00:14:03,420 --> 00:14:07,720
Ejaaz:
There's a lot of big, valuable investors and kind of infrastructure providers

231
00:14:07,720 --> 00:14:12,300
Ejaaz:
getting behind the Google train right The momentum is palpable, to say the least, right?

232
00:14:12,480 --> 00:14:16,860
Ejaaz:
And this NATO deal is another example of it, right? Like, we're going from the

233
00:14:16,860 --> 00:14:19,800
Ejaaz:
hyperscaler kind of consumer level to the government level as well.

234
00:14:19,860 --> 00:14:23,680
Ejaaz:
So all types of organizations are treating this with a very high importance

235
00:14:23,680 --> 00:14:26,600
Ejaaz:
that Google is going to play an inevitably big role here.

236
00:14:26,700 --> 00:14:30,300
Ejaaz:
So then that begs the question, well, what's NVIDIA going to do about this?

237
00:14:30,400 --> 00:14:33,780
Ejaaz:
Are they just going to continue losing hundreds of billions of dollars in their

238
00:14:33,780 --> 00:14:36,000
Ejaaz:
market cap? Or are they going to strike back?

239
00:14:36,160 --> 00:14:39,780
Ejaaz:
And there's two frames of thought about this, Josh. Number one is,

240
00:14:40,000 --> 00:14:46,060
Ejaaz:
so NVIDIA's next generation of GPUs is going to be around the Rubin architecture.

241
00:14:46,060 --> 00:14:47,760
Ejaaz:
It's called Rubin, right?

242
00:14:48,200 --> 00:14:54,620
Ejaaz:
They introduced a new spec, Josh, after Google's TPUs, their latest TPUs got

243
00:14:54,620 --> 00:15:00,200
Ejaaz:
released, which upped a lot of the watts or compute performance for the Rubin architecture.

244
00:15:00,640 --> 00:15:04,320
Ejaaz:
Now, some might say this is just coincidental, but some might say this is a

245
00:15:04,320 --> 00:15:09,820
Ejaaz:
general reaction to the fact that Google just has a higher performance TPU versus theirs.

246
00:15:10,080 --> 00:15:13,480
Ejaaz:
And so they needed to kind of like up the metrics of their next generation if

247
00:15:13,480 --> 00:15:18,220
Ejaaz:
they wanted to compete and appear attractive to their competitors or to their customers themselves.

248
00:15:19,290 --> 00:15:22,910
Ejaaz:
But then there's also the argument where it's just kind of like NVIDIA and Google

249
00:15:22,910 --> 00:15:26,770
Ejaaz:
are kind of playing in kind of like different ballpark and they already know

250
00:15:26,770 --> 00:15:27,870
Ejaaz:
this. They're playing different games.

251
00:15:28,150 --> 00:15:32,610
Ejaaz:
The argument here in this tweet being that Google's TPUs are great,

252
00:15:32,750 --> 00:15:37,130
Ejaaz:
but they're only for very specific niche use cases. If you have, you mentioned earlier,

253
00:15:37,910 --> 00:15:41,710
Ejaaz:
the recommendation or search algorithm, then, you know, these ASICs are going to be really good.

254
00:15:41,870 --> 00:15:45,630
Ejaaz:
The benefit of NVIDIA's GPUs is that they're highly generalizable.

255
00:15:45,790 --> 00:15:49,590
Ejaaz:
So if you wanted to train a model in a different way or test out a new method

256
00:15:49,590 --> 00:15:54,310
Ejaaz:
to kind of like inference or train your model, GPUs are by far the best architecture

257
00:15:54,310 --> 00:15:56,050
Ejaaz:
or the best infrastructure to use.

258
00:15:56,310 --> 00:16:00,370
Ejaaz:
So you could argue that Google, that, sorry, NVIDIA is sitting pretty comfy.

259
00:16:00,810 --> 00:16:05,990
Ejaaz:
And Jensen went on a show or an interview this week, basically saying that he's

260
00:16:05,990 --> 00:16:08,350
Ejaaz:
not worried about Google. Obviously, you expect him to say that,

261
00:16:08,670 --> 00:16:12,430
Ejaaz:
but mainly for the fact that this is a positive sum game.

262
00:16:12,750 --> 00:16:16,410
Ejaaz:
You know, Jevon's paradox, if you create more GPUs, it's not going to be a fact

263
00:16:16,410 --> 00:16:19,450
Ejaaz:
that you have oversupply. There's just going to be increased demand for compute.

264
00:16:19,690 --> 00:16:22,690
Ejaaz:
And I think Jensen knows this, and that's why he's just kind of running forward.

265
00:16:22,810 --> 00:16:27,470
Ejaaz:
The fact of the matter is, there isn't enough NVIDIA GPUs to supply the customers,

266
00:16:27,590 --> 00:16:28,550
Ejaaz:
even if you wanted to, right?

267
00:16:28,610 --> 00:16:32,890
Ejaaz:
He needs to ramp up infrastructure production. That's why he's been visiting

268
00:16:32,890 --> 00:16:36,370
Ejaaz:
TSMC for the last couple of weeks. So I think he knows this and I don't think

269
00:16:36,370 --> 00:16:40,630
Ejaaz:
he's too worried, but he's definitely sweating a little. I don't know if you think the same.

270
00:16:40,990 --> 00:16:44,830
Josh:
Yeah, no, I mean, I'm sure it sucks. It's like you were just running the show

271
00:16:44,830 --> 00:16:47,390
Josh:
and now suddenly there's someone else who has a good product.

272
00:16:47,730 --> 00:16:50,370
Josh:
It's not to say that it's going to harm the company too much.

273
00:16:50,430 --> 00:16:53,250
Josh:
And I think for anyone who's listening, if you take away one thing from this

274
00:16:53,250 --> 00:16:56,890
Josh:
episode, it's that both of these companies are going to succeed wildly.

275
00:16:56,930 --> 00:17:01,170
Josh:
And there is going to be a shortage of supply for compute for a very long time.

276
00:17:01,170 --> 00:17:05,810
Josh:
If you believe that AI is as impressive and as important as a technology as

277
00:17:05,810 --> 00:17:09,490
Josh:
it really is, then you also have to believe that all of the compute around us

278
00:17:09,490 --> 00:17:12,250
Josh:
must be replaced by it and must have it embedded inside of it.

279
00:17:12,310 --> 00:17:15,570
Josh:
In order to do so, you need to shift the entire technological infrastructure

280
00:17:15,570 --> 00:17:19,270
Josh:
of all the hardware that exists over to infrastructure that supports AI.

281
00:17:19,850 --> 00:17:24,010
Josh:
Everything. And we are just a fraction of a percentage through that transition.

282
00:17:24,250 --> 00:17:27,970
Josh:
So as a result, there could be many more Googles, many more NVIDIAs,

283
00:17:28,050 --> 00:17:29,550
Josh:
and there would still be a shortage.

284
00:17:29,710 --> 00:17:33,790
Josh:
Now, the question becomes, is there a short-term bubble? Are we overspending?

285
00:17:33,950 --> 00:17:37,050
Josh:
That is to be determined, but this is a good type of bubble.

286
00:17:37,150 --> 00:17:41,170
Josh:
This is one that even if it does explode, we are left with unbelievable technology

287
00:17:41,170 --> 00:17:44,870
Josh:
across the board and a scaling infrastructure that will continue to be able

288
00:17:44,870 --> 00:17:47,810
Josh:
to support this new type of technology that's permeating throughout society.

289
00:17:47,810 --> 00:17:50,710
Josh:
So is this a good thing yes is nvidia going

290
00:17:50,710 --> 00:17:53,750
Josh:
to suffer maybe sure the headlines suck it's like okay

291
00:17:53,750 --> 00:17:56,430
Josh:
we're not the coolest person in the world now there's someone else who's like also a

292
00:17:56,430 --> 00:17:59,250
Josh:
cool kid but they're still going to continue to produce

293
00:17:59,250 --> 00:18:01,910
Josh:
the best products in the world and i just to the point that you made

294
00:18:01,910 --> 00:18:05,270
Josh:
earlier they're just different types of chips like um a

295
00:18:05,270 --> 00:18:08,310
Josh:
gpu is very different than a tpu and a

296
00:18:08,310 --> 00:18:11,050
Josh:
lot of people also need to understand that the whole

297
00:18:11,050 --> 00:18:13,690
Josh:
world isn't actually training ai there's still a lot of other things that are

298
00:18:13,690 --> 00:18:17,450
Josh:
happening like graphics or simulation or financial technology

299
00:18:17,450 --> 00:18:20,130
Josh:
scientific research tpus just don't do

300
00:18:20,130 --> 00:18:22,990
Josh:
that and gpus do so there's there's a lot more

301
00:18:22,990 --> 00:18:25,910
Josh:
going on to the story there's a lot more gpus being sold there's a lot of tpus

302
00:18:25,910 --> 00:18:29,110
Josh:
there's enough for everybody and then there's still not enough for everybody

303
00:18:29,110 --> 00:18:33,130
Josh:
so i think in the long run like this is just great for both companies this is

304
00:18:33,130 --> 00:18:37,230
Josh:
positive some uh there's a lot of excitement around this rightfully so because

305
00:18:37,230 --> 00:18:39,990
Josh:
i think it's it's great there's another person stepping in but it's not the

306
00:18:39,990 --> 00:18:43,530
Josh:
end of of anyone it's just the beginning for for so many of these companies still i.

307
00:18:43,530 --> 00:18:47,450
Ejaaz:
Mean like don't take your and my opinions either, right? Why don't you just

308
00:18:47,450 --> 00:18:50,590
Ejaaz:
listen to one of the smartest men or smartest businessmen? Oh, we got Elon.

309
00:18:51,310 --> 00:18:55,770
Ejaaz:
Yeah, we got Elon. And he was asked this question in this interview clip that I'm about to show.

310
00:18:56,370 --> 00:19:00,650
Ejaaz:
This was released yesterday where he was asked, Elon, if you had to invest in

311
00:19:00,650 --> 00:19:05,070
Ejaaz:
any AI companies today and hold it for a decade, what would you buy?

312
00:19:05,250 --> 00:19:07,650
Ejaaz:
And you think, you know, Elon would shill his own companies.

313
00:19:07,790 --> 00:19:12,430
Ejaaz:
He didn't. He shilled two companies, Josh, Google and NVIDIA. Let me show you a clip.

314
00:19:12,630 --> 00:19:16,470
Ejaaz:
I think, you know, Google is going are gonna be pretty valuable in the future.

315
00:19:16,930 --> 00:19:19,050
Ejaaz:
They've laid the groundwork for,

316
00:19:20,520 --> 00:19:23,760
Ejaaz:
an immense amount of value creation from an AI standpoint.

317
00:19:27,020 --> 00:19:28,700
Ejaaz:
NVIDIA is obvious at this point.

318
00:19:31,660 --> 00:19:37,020
Ejaaz:
I mean, there's an argument that companies that do AI and robotics and maybe

319
00:19:37,020 --> 00:19:45,080
Ejaaz:
space flight are gonna be overwhelmingly all the value, almost all the value.

320
00:19:46,890 --> 00:19:51,130
Ejaaz:
So the output of goods and services from AI and robotics is so high that it

321
00:19:51,130 --> 00:19:52,130
Ejaaz:
will dwarf everything else.

322
00:19:52,790 --> 00:19:58,030
Ejaaz:
And so, you know, you hear it there from Selfway is basically describing NVIDIA

323
00:19:58,030 --> 00:20:03,650
Ejaaz:
as a sort of toll collector because you kind of like need to basically pay the

324
00:20:03,650 --> 00:20:07,450
Ejaaz:
toll man for his GPUs to get access to the intelligence that you're trying to build.

325
00:20:07,810 --> 00:20:11,970
Ejaaz:
And then Google's mode is kind of similar but quite different in the sense that

326
00:20:11,970 --> 00:20:18,870
Ejaaz:
they create the GPUs, their own TPUs, but they also like kind of own dominance

327
00:20:18,870 --> 00:20:21,210
Ejaaz:
across the entire AI stack, right, Josh?

328
00:20:21,650 --> 00:20:25,690
Ejaaz:
And just to kind of maybe like round things up, I was looking at these crazy

329
00:20:25,690 --> 00:20:28,350
Ejaaz:
charts from the Financial Times this week,

330
00:20:28,370 --> 00:20:34,270
Ejaaz:
which basically showed that Google's Gemini model has now almost caught up in

331
00:20:34,270 --> 00:20:38,250
Ejaaz:
the number of users or monthly downloads that ChatGPT has,

332
00:20:38,350 --> 00:20:40,610
Ejaaz:
which is just insane. That's the chart that we see here on the left.

333
00:20:40,770 --> 00:20:44,990
Ejaaz:
And then on the right, which I found the most interesting, is the amount of

334
00:20:44,990 --> 00:20:51,450
Ejaaz:
time that each Gemini user is spending on the app using the Gemini model has now beaten ChatGPT.

335
00:20:51,630 --> 00:20:54,850
Ejaaz:
This kind of blew my mind because I was like, surely everyone's still using

336
00:20:54,850 --> 00:20:58,410
Ejaaz:
ChatGPT because people tend to use ChatGPT for their own personalized things.

337
00:20:58,570 --> 00:21:01,310
Ejaaz:
They kind of like confer it, therapy, ask about personal stuff.

338
00:21:01,490 --> 00:21:05,470
Ejaaz:
That probably spends a lot more time, but it seems like the productivity aspect

339
00:21:05,470 --> 00:21:12,010
Ejaaz:
that people are getting from task orientation-based AI stuff using Gemini seems to be extending.

340
00:21:12,010 --> 00:21:14,930
Ejaaz:
And that just kind of like shows I've heard anecdotes from friends you

341
00:21:14,930 --> 00:21:17,810
Ejaaz:
and I were talking to a team member just before recording this and he

342
00:21:17,810 --> 00:21:20,150
Ejaaz:
was like yeah I was talking to a bunch of my friends and they've fully switched

343
00:21:20,150 --> 00:21:24,410
Ejaaz:
to the Gemini app so I think we're going to continue seeing this trend of Google

344
00:21:24,410 --> 00:21:28,010
Ejaaz:
gaining the advantage not because of their infrastructure mode but because they

345
00:21:28,010 --> 00:21:32,250
Ejaaz:
like own all the popular apps that anyone and everyone wants to use and so all

346
00:21:32,250 --> 00:21:35,790
Ejaaz:
they have to do is plug in their AI model with whatever app Gmail,

347
00:21:36,150 --> 00:21:40,350
Ejaaz:
Maps whatever you might kind of think of and suddenly you have a really productive

348
00:21:40,350 --> 00:21:42,410
Ejaaz:
useful app that you and I want to use every day.

349
00:21:42,630 --> 00:21:46,410
Ejaaz:
Josh, you mentioned that you want to use down to banana or you're using down to banana, right?

350
00:21:46,960 --> 00:21:51,080
Josh:
Yeah, there's a there's an important shift that I found that has happened recently

351
00:21:51,080 --> 00:21:54,420
Josh:
that hasn't happened before, which is I have Gemini on my home screen on my phone.

352
00:21:54,620 --> 00:21:58,660
Josh:
And that to me is very high signal because I've been resistant of it because

353
00:21:58,660 --> 00:21:59,720
Josh:
it just hasn't been good.

354
00:22:00,080 --> 00:22:05,300
Josh:
And while it's still not great, I think the ChatGPT app is engineered far better than Gemini.

355
00:22:05,480 --> 00:22:09,580
Josh:
It is good enough to make me want to use it. So I went from not being a user

356
00:22:09,580 --> 00:22:13,840
Josh:
like I would really I'd use Gemini 3 Pro on desktop whenever I had a hard question,

357
00:22:13,980 --> 00:22:15,300
Josh:
but I wasn't reaching for it.

358
00:22:15,300 --> 00:22:20,240
Josh:
And Nano Banana Pro, it really was the killer use case that had me like,

359
00:22:20,440 --> 00:22:24,680
Josh:
oh my God, I need this quickly accessible and in my pocket at all times because

360
00:22:24,680 --> 00:22:28,480
Josh:
it is so far superior to any other product in the space like that.

361
00:22:28,800 --> 00:22:32,760
Josh:
And I think as Google starts to roll out these products, as this is something

362
00:22:32,760 --> 00:22:35,280
Josh:
that we talk about a lot with OpenAI, where they're really good at creating

363
00:22:35,280 --> 00:22:37,520
Josh:
innovation and wrapping it in a product and selling it.

364
00:22:38,080 --> 00:22:42,000
Josh:
As Google gets better at doing that, I really, I strongly suspect Gemini will

365
00:22:42,000 --> 00:22:44,740
Josh:
continue this trend of taking over broader and broader people.

366
00:22:44,740 --> 00:22:47,660
Josh:
Because when you think about how many monthly active users google

367
00:22:47,660 --> 00:22:50,280
Josh:
has it's like gigantic they're one of the few in the

368
00:22:50,280 --> 00:22:53,120
Josh:
world that actually has more than than chat gpt and open ai

369
00:22:53,120 --> 00:22:56,120
Josh:
and if they can convert all of these services to

370
00:22:56,120 --> 00:22:59,760
Josh:
pack ai into it into one coherent service

371
00:22:59,760 --> 00:23:03,840
Josh:
and package that's incredible we talk about a lot of times like with meta for

372
00:23:03,840 --> 00:23:07,200
Josh:
example the they had their awesome hardware product but no one really wanted

373
00:23:07,200 --> 00:23:10,940
Josh:
to use it because the ecosystem sucked well google has like the best ecosystem

374
00:23:10,940 --> 00:23:14,840
Josh:
ever it's funny on my even on my iphone i have an iphone hardware with software

375
00:23:14,840 --> 00:23:16,980
Josh:
from Google because their software is so superior.

376
00:23:17,120 --> 00:23:20,460
Josh:
So as they're able to integrate these top tier models into all the products,

377
00:23:20,640 --> 00:23:24,400
Josh:
this is a serious shift. And I'm very bullish on Google.

378
00:23:25,180 --> 00:23:28,700
Ejaaz:
Yeah, I don't think we're going to see this trend reverse. We already know that

379
00:23:28,700 --> 00:23:33,860
Ejaaz:
Apple is going to be using a Google Gemini-based model in their phone for Siri.

380
00:23:34,180 --> 00:23:38,980
Ejaaz:
That's right, yeah. And we know that we're going to start seeing a lot of Gemini-based

381
00:23:38,980 --> 00:23:41,720
Ejaaz:
apps just kind of like appear in our regular day-to-day.

382
00:23:41,880 --> 00:23:45,720
Ejaaz:
One other final point is like when you compare like Google versus OpenAI,

383
00:23:46,100 --> 00:23:47,840
Ejaaz:
remember OpenAI still isn't profitable.

384
00:23:48,260 --> 00:23:52,380
Ejaaz:
Google is massively, massively profitable, so they don't need to turn on ads.

385
00:23:52,380 --> 00:23:56,080
Ejaaz:
They don't need to kind of like like demean the

386
00:23:56,080 --> 00:23:58,880
Ejaaz:
user experience in any way they could just keep giving you the stuff for free

387
00:23:58,880 --> 00:24:01,820
Ejaaz:
and gaining millions and millions more monthly active

388
00:24:01,820 --> 00:24:05,060
Ejaaz:
users whereas open air at some point is going to turn on ads and when they turn

389
00:24:05,060 --> 00:24:08,420
Ejaaz:
on ads it's going to be an inferior performance it's going to be a inferior

390
00:24:08,420 --> 00:24:13,040
Ejaaz:
product to some extent and that might shift to more people using google gemini's

391
00:24:13,040 --> 00:24:16,500
Ejaaz:
products and google knows this so they're willing to just kind of sit back they

392
00:24:16,500 --> 00:24:19,880
Ejaaz:
own kind of every infrastructure layer and they're just going to see how things play out.

393
00:24:20,380 --> 00:24:23,600
Ejaaz:
But I think that is it for today's episode.

394
00:24:24,600 --> 00:24:28,480
Ejaaz:
Super exciting to kind of like see where Google and NVIDIA ultimately end up.

395
00:24:28,600 --> 00:24:32,080
Ejaaz:
In my opinion, I think that both companies, to your point earlier,

396
00:24:32,160 --> 00:24:34,240
Ejaaz:
Josh, are going to do extremely, extremely well.

397
00:24:34,360 --> 00:24:37,600
Ejaaz:
It's a positive sum game. And the fact of the matter is there is not enough compute.

398
00:24:37,740 --> 00:24:41,960
Ejaaz:
There's not enough energy to feed the compute that both of these companies are pushing out.

399
00:24:42,140 --> 00:24:45,260
Ejaaz:
So I think we're just going to see both these companies grow into two of the

400
00:24:45,260 --> 00:24:50,100
Ejaaz:
largest and most valuable companies in the world. Now, I need to take a quick

401
00:24:50,100 --> 00:24:55,900
Ejaaz:
victory lap for all listeners here who aren't subscribed and who haven't rated our show just yet.

402
00:24:56,220 --> 00:24:59,220
Ejaaz:
We released an episode about the bull case for Google.

403
00:24:59,620 --> 00:25:03,520
Ejaaz:
What was it? Two months ago now. And a lot of it has now played out right now.

404
00:25:03,640 --> 00:25:06,960
Ejaaz:
If you'd invested in Google back then, you would have participated in the hundreds

405
00:25:06,960 --> 00:25:10,420
Ejaaz:
of billions of dollars that their market cap is up relative to NVIDIA right now.

406
00:25:10,660 --> 00:25:12,520
Ejaaz:
Now, I'm not going to say that we triggered it. Maybe we did.

407
00:25:12,760 --> 00:25:16,580
Ejaaz:
Maybe we didn't. But if you want to hear more bull cases like this or more alpha

408
00:25:16,580 --> 00:25:21,880
Ejaaz:
in advance, subscribe to us, rate us, give us a thumbs up, give us feedback. We love it.

409
00:25:22,160 --> 00:25:25,420
Ejaaz:
And we will hear more from you or we will, you will hear more from us rather

410
00:25:25,420 --> 00:25:27,560
Ejaaz:
on the next episode. See you then.