1
00:00:00,018 --> 00:00:05,078
Ejaaz:
Leopold Aschenbrenner is a 23-year-old kid that raised one of the most successful

2
00:00:05,078 --> 00:00:10,344
Ejaaz:
AI funds ever, but he might be in disastrous trouble right now.

3
00:00:10,565 --> 00:00:16,608
Ejaaz:
News reports this morning states that his fund might be down between 30 to 50%

4
00:00:16,730 --> 00:00:20,512
Ejaaz:
as semiconductor stocks have drawn down aggressively over the last four weeks.

5
00:00:20,512 --> 00:00:22,338
Ejaaz:
We're going to get into that and a lot more.

6
00:00:22,338 --> 00:00:25,838
Josh:
Yeah, let's kick it off with our poster boy, golden child, Leopold Aschenbrenner,

7
00:00:25,838 --> 00:00:29,998
Josh:
who has basically done no wrong for the last, what is it, two years of investing

8
00:00:30,318 --> 00:00:34,048
Josh:
where he's rolled this fund into something like 30 billion dollars well for

9
00:00:34,048 --> 00:00:39,625
Josh:
the first time ever since inception it seems like things are not doing super hot in paradise

10
00:00:40,269 --> 00:00:43,957
Josh:
looks like he's starting to sell and not just sell, but sell a lot.

11
00:00:43,957 --> 00:00:48,019
Josh:
In fact, liquidate a lot of his positions that he's been accumulating over time

12
00:00:48,234 --> 00:00:49,650
Josh:
because it seems like the fund is in trouble.

13
00:00:49,842 --> 00:00:53,887
Josh:
We know the kind of directive. We filmed enough episodes on Leopold to understand

14
00:00:53,887 --> 00:00:59,067
Josh:
his general thesis, which is long only with leverage to try to just maximize

15
00:00:59,067 --> 00:01:01,376
Josh:
upside returns, which has worked incredibly well.

16
00:01:01,937 --> 00:01:05,467
Josh:
But we understand that, you know, well, I remember vividly in the last episode,

17
00:01:05,467 --> 00:01:08,327
Josh:
you know, as we were comparing him to someone like Gavin Baker and how Gavin's

18
00:01:08,327 --> 00:01:11,807
Josh:
been around for 30 years because he's able to survive the upswings,

19
00:01:11,807 --> 00:01:14,594
Josh:
the downswings. They have that enduring capability.

20
00:01:14,913 --> 00:01:18,097
Josh:
Leopold is getting his first test of strength right now, and it looks like it's

21
00:01:18,097 --> 00:01:21,887
Josh:
not going super hot. I mean, on screen, I'm looking at a tweet from Martin Screlli

22
00:01:21,887 --> 00:01:26,777
Josh:
that says, Situation has sold half of its Anthropoc stake, which is like,

23
00:01:26,777 --> 00:01:28,113
Josh:
I mean, granted, this is all rumors.

24
00:01:28,432 --> 00:01:32,931
Josh:
None of this is confirmed to be true. But man, if that is, that is a painful, painful loss.

25
00:01:33,244 --> 00:01:41,037
Ejaaz:
To be clear, he invested in their series, a trownd in 2025 when they were worth $60 billion.

26
00:01:41,929 --> 00:01:46,207
Ejaaz:
So he must be in pretty dire trouble because it made up for like 40% of his

27
00:01:46,207 --> 00:01:49,337
Ejaaz:
funds. So we're talking about like three to maybe $6 billion.

28
00:01:50,597 --> 00:01:54,617
Ejaaz:
And for him to sell that, he must be in drastic trouble. So what are the rumors saying?

29
00:01:54,617 --> 00:01:59,677
Ejaaz:
Again, very reliable source here, Martin Shkreli from the Godel News says that

30
00:01:59,677 --> 00:02:04,632
Ejaaz:
Situational Awareness LP fund is down more than 50% month to date,

31
00:02:04,893 --> 00:02:09,497
Ejaaz:
and it was up 200% year to date. So that is a massive drawdown.

32
00:02:10,152 --> 00:02:13,777
Ejaaz:
And in terms of like what positions he's liquidating, I'm a little confused

33
00:02:13,777 --> 00:02:16,787
Ejaaz:
here, Josh, because when we covered his recent 13F,

34
00:02:17,438 --> 00:02:22,928
Ejaaz:
he was hedged. He wasn't particularly bullish and like long on everything,

35
00:02:23,143 --> 00:02:26,008
Ejaaz:
but he also wasn't particularly bearish. If you remember, he had one of the

36
00:02:26,008 --> 00:02:28,054
Ejaaz:
largest shorts on NVIDIA itself.

37
00:02:28,193 --> 00:02:32,135
Ejaaz:
So I was wondering if he had that sort of portfolio composition back then,

38
00:02:32,384 --> 00:02:36,168
Ejaaz:
he might have even like, you know, not drawn down too much.

39
00:02:36,168 --> 00:02:39,548
Ejaaz:
You know, rumors are saying that he's down like 30% overall on the fund,

40
00:02:39,548 --> 00:02:43,588
Ejaaz:
which, you know, is disastrous for the size of the fund, but it's still not

41
00:02:43,588 --> 00:02:46,548
Ejaaz:
the worst that it could be given that he was long all these semiconductor stocks.

42
00:02:46,548 --> 00:02:50,403
Ejaaz:
Now, the reason, you know, the question about like whether this is good or bad,

43
00:02:50,682 --> 00:02:53,224
Ejaaz:
I think investors have really been feeling it broadly.

44
00:02:53,708 --> 00:02:57,448
Ejaaz:
A lot of AI stocks in general have just drawn down a lot over the last four

45
00:02:57,448 --> 00:02:59,948
Ejaaz:
weeks. We actually had an episode that covered this yesterday.

46
00:02:59,948 --> 00:03:03,104
Ejaaz:
You should definitely go check that episode out. We explained kind of like why and what's going on.

47
00:03:03,429 --> 00:03:07,028
Ejaaz:
But I think his thesis still holds true. And listen, I hope he makes it out.

48
00:03:07,284 --> 00:03:08,688
Ejaaz:
I hope he's able to come out of this.

49
00:03:08,748 --> 00:03:12,298
Josh:
Yeah, dude, you and me both. I mean, we'll see. It's the difficult thing about

50
00:03:12,298 --> 00:03:16,009
Josh:
this again is it's all speculation because this is all insider rumors. No one actually knows.

51
00:03:16,322 --> 00:03:20,108
Josh:
We will see dude his next 13f filing is going to be one of the most exciting

52
00:03:20,108 --> 00:03:25,748
Josh:
episodes of all time did ilia survive or did he not or so speaking of ilia no did

53
00:03:26,097 --> 00:03:29,908
Josh:
leopold survive but speaking of ilia we have some news in the ilia front as

54
00:03:29,908 --> 00:03:33,348
Josh:
well which comes in the form of his company safe super intelligence they've been

55
00:03:33,728 --> 00:03:37,088
Josh:
radio silent and i give this company and the team a tremendous amount of credit

56
00:03:37,088 --> 00:03:39,298
Josh:
because gosh who is doing this much work remind

57
00:03:39,298 --> 00:03:40,168
Ejaaz:
Us who ilia is.

58
00:03:40,673 --> 00:03:46,768
Josh:
Ilia setzkever he is ex co-founder of open ai he left open ai to go start his

59
00:03:46,768 --> 00:03:50,718
Josh:
own company no one really knows what this company is they're just named safe

60
00:03:50,718 --> 00:03:54,622
Josh:
super intelligence the idea is that they want to figure out a way to safely align ai

61
00:03:54,959 --> 00:03:59,808
Josh:
with the human intention and it sounds like they have been making progress i

62
00:03:59,808 --> 00:04:04,206
Josh:
mean this company has been radio silent there's no leaks there's no information getting out

63
00:04:04,531 --> 00:04:08,043
Josh:
but they recently just this week announced a long-term strategic partnership

64
00:04:08,258 --> 00:04:13,098
Josh:
with nvidia which is like huge it says nvidia is making a substantial investment

65
00:04:13,098 --> 00:04:17,278
Josh:
in ssi that will let us 10xr compute in the next 12 months we reach the point

66
00:04:17,278 --> 00:04:19,338
Josh:
where our research is worth scaling

67
00:04:19,976 --> 00:04:24,179
Josh:
And with this partnership, we will be able to. We are honored by NVIDIA's conviction.

68
00:04:24,417 --> 00:04:25,114
Ejaaz:
That's crazy.

69
00:04:25,427 --> 00:04:27,627
Josh:
They made a breakthrough. They have some sort of a breakthrough.

70
00:04:27,627 --> 00:04:30,987
Josh:
Like Jensen saw this and he was like, oh yeah, bro, we got you.

71
00:04:30,987 --> 00:04:31,867
Josh:
We're investing in this.

72
00:04:31,867 --> 00:04:33,387
Ejaaz:
That's huge. That's huge.

73
00:04:33,387 --> 00:04:37,967
Josh:
So I have to imagine it's only a matter of time until we actually wind up getting

74
00:04:37,967 --> 00:04:41,877
Josh:
one of these breakthroughs published. But I know Ilya in particular has been

75
00:04:41,877 --> 00:04:43,678
Josh:
focused on the human anatomy of the brain.

76
00:04:44,125 --> 00:04:48,347
Josh:
And I have to imagine that this is some breakthrough relating to a part of the

77
00:04:48,347 --> 00:04:52,780
Josh:
brain that we haven't quite tapped into. It leaves very bullish on the idea that

78
00:04:53,494 --> 00:04:56,402
Josh:
AI is very similar to a human in the way that it learns and the way that it

79
00:04:56,402 --> 00:05:00,032
Josh:
thinks. And there's a lot of places we could draw from. So I am stoked to see

80
00:05:00,032 --> 00:05:04,663
Josh:
what this looks like. But the rumors are true. They're building something and it is very badass.

81
00:05:05,110 --> 00:05:10,382
Ejaaz:
So just to conceptualize, you know, why it's super important for these smaller

82
00:05:10,382 --> 00:05:14,717
Ejaaz:
AI labs that aren't called Anthropic, OpenAI or Google to announce something

83
00:05:14,973 --> 00:05:18,672
Ejaaz:
is they're basically just overvalued research labs.

84
00:05:18,672 --> 00:05:21,842
Ejaaz:
When they're funded, it's just a small team of researchers. and they're like,

85
00:05:21,842 --> 00:05:26,332
Ejaaz:
well, hey, hopefully we make a breakthrough that's better than what Anthropic and OpenAI is doing.

86
00:05:26,512 --> 00:05:30,352
Ejaaz:
And that's what you're betting on. So it's actually a very low hit rate that

87
00:05:30,352 --> 00:05:32,624
Ejaaz:
these AI labs actually get anything out of the door.

88
00:05:32,828 --> 00:05:37,262
Ejaaz:
So this announcement, like raising $5 billion, what did Jensen see is the meme

89
00:05:37,262 --> 00:05:41,605
Ejaaz:
that I'm seeing going around, is fantastic news for Ilya and what they're building.

90
00:05:41,808 --> 00:05:44,983
Ejaaz:
Now, the question is, if they've made an architectural breakthrough,

91
00:05:45,152 --> 00:05:47,522
Ejaaz:
how different is it to the LLM

92
00:05:47,522 --> 00:05:50,922
Ejaaz:
architecture, which is how Claude and Chachupiti work behind the scenes.

93
00:05:51,067 --> 00:05:55,722
Ejaaz:
If SSI has made a meaningful breakthrough, then there is an argument to be said

94
00:05:55,722 --> 00:05:58,392
Ejaaz:
that they could be competitive with these Frontier Labs. Because of course,

95
00:05:58,392 --> 00:06:02,112
Ejaaz:
remember, the argument is these Frontier Labs are so far ahead and eventually

96
00:06:02,112 --> 00:06:04,709
Ejaaz:
they're going to be building a model that can recursively build itself.

97
00:06:04,987 --> 00:06:08,848
Ejaaz:
How can you possibly catch up? So this news from Ilya is super cool.

98
00:06:09,039 --> 00:06:15,282
Ejaaz:
Now, the rumor has it is his model will release as soon as mid to late August.

99
00:06:15,282 --> 00:06:20,222
Ejaaz:
And if strawberry man over here on the screen is is is right he goes uh it'll

100
00:06:20,222 --> 00:06:22,401
Ejaaz:
be superior to fable now listen.

101
00:06:22,825 --> 00:06:26,251
Ejaaz:
By the time we get to the end of August, Anthropic will put out Fable 6 and

102
00:06:26,251 --> 00:06:27,091
Ejaaz:
it'll crush everything else.

103
00:06:27,091 --> 00:06:30,701
Ejaaz:
But the point is, for a smaller lab to catch up is very interesting.

104
00:06:30,701 --> 00:06:35,761
Ejaaz:
Now, on the topic of recursive self-improvement, Josh, there was a big and very

105
00:06:35,761 --> 00:06:40,141
Ejaaz:
weird alignment between two of the biggest enemies in AI.

106
00:06:40,344 --> 00:06:45,661
Ejaaz:
Anthropic and OpenAI researchers, I think 1,300 spanning across other frontier

107
00:06:45,661 --> 00:06:47,629
Ejaaz:
AI labs and the AI industry in general,

108
00:06:47,920 --> 00:06:52,781
Ejaaz:
signed an open petition calling for the US government to deliberately slow down

109
00:06:52,781 --> 00:06:56,421
Ejaaz:
or pace AI development in order to prevent it from advancing too quickly.

110
00:06:56,421 --> 00:07:01,021
Ejaaz:
And the general thesis behind this letter is we are very close to building an

111
00:07:01,021 --> 00:07:05,190
Ejaaz:
AI model that can self-improve, that can build the next version of itself.

112
00:07:05,323 --> 00:07:10,871
Ejaaz:
And in that world, all that matters is the AI model could get much smarter too

113
00:07:10,871 --> 00:07:12,561
Ejaaz:
quickly for humans to comprehend.

114
00:07:12,561 --> 00:07:15,601
Ejaaz:
And we could end up in a situation where the AI model is just autonomously working

115
00:07:15,601 --> 00:07:21,801
Ejaaz:
for itself and humans, for lack of a better fantasy, end up becoming the slaves

116
00:07:21,801 --> 00:07:23,828
Ejaaz:
of this thing i think that's what's happening here josh.

117
00:07:24,113 --> 00:07:27,892
Josh:
Yeah and funny enough ilia who we just mentioned actually signed this this morning

118
00:07:28,118 --> 00:07:29,952
Josh:
the petition so he is number say like

119
00:07:30,324 --> 00:07:33,813
Josh:
1301 to be involved in this and there's basically three points to this

120
00:07:34,039 --> 00:07:37,701
Josh:
uh the first being that they believe that ai may soon be able to build better

121
00:07:37,701 --> 00:07:41,231
Josh:
ai which is that recursive self-improvement loop that's shocking about um that

122
00:07:41,231 --> 00:07:45,831
Josh:
can make progress accelerate uh really fast and make things really scary we

123
00:07:45,831 --> 00:07:48,610
Josh:
started to see this with the chat gpt 6.0

124
00:07:48,813 --> 00:07:51,031
Josh:
leak and how sam altman just came from

125
00:07:51,451 --> 00:07:54,746
Josh:
the government this week he's like he's in washington dc talking to the government

126
00:07:54,989 --> 00:07:59,191
Josh:
and he's saying words like we have permanently deactivated the model as if it

127
00:07:59,191 --> 00:08:02,171
Josh:
is this dangerous malicious thing because they are actually like afraid of

128
00:08:02,631 --> 00:08:05,621
Josh:
what it's able to do and don't quite know how to sandbox just yet the second

129
00:08:05,621 --> 00:08:09,101
Josh:
part of this is that no single company or country can slow down alone so they

130
00:08:09,101 --> 00:08:10,738
Josh:
believe that everyone needs to do this in unison

131
00:08:11,388 --> 00:08:15,090
Josh:
So they're asking the U.S. government to lead this international effort and

132
00:08:15,090 --> 00:08:18,933
Josh:
to build the tools and agreements needed to slow things down if it becomes necessary.

133
00:08:19,949 --> 00:08:22,800
Josh:
It seems very clear that the labs are getting spooked out pretty quickly because

134
00:08:22,800 --> 00:08:25,533
Josh:
the capabilities of these things are getting good quick and they're getting

135
00:08:25,817 --> 00:08:28,488
Josh:
way better, way faster than anticipated.

136
00:08:28,784 --> 00:08:32,847
Josh:
And there's this like kind of looming fear like, hey, how do we actually safely

137
00:08:33,352 --> 00:08:37,073
Josh:
even test these things privately, let alone release them publicly?

138
00:08:37,369 --> 00:08:40,170
Josh:
So it's this like really strange disconnect that's happening.

139
00:08:40,170 --> 00:08:43,314
Josh:
And it's interesting to see all the labs align around this thing, including ILIA,

140
00:08:44,010 --> 00:08:48,381
Josh:
And all of the major players from a lot of the major labs. So we'll see how this plays out.

141
00:08:48,503 --> 00:08:51,204
Josh:
It is definitely interesting, though. It's like everyone's thinking now,

142
00:08:51,204 --> 00:08:54,152
Josh:
hey, if we slow things down, actually, it probably is a good thing.

143
00:08:54,471 --> 00:08:56,450
Josh:
It's just the question now of whether it's even possible.

144
00:08:56,706 --> 00:09:00,994
Ejaaz:
Yeah, well, I mean, on the topic of whether something is potentially good or

145
00:09:00,994 --> 00:09:03,904
Ejaaz:
bad, whether you should slow down model development, Anthropic came out with

146
00:09:03,904 --> 00:09:07,334
Ejaaz:
a statement on the debate between open and closed source.

147
00:09:07,334 --> 00:09:12,084
Ejaaz:
Now, for those of you who aren't aware, there's been a big point of contention between the U.S.

148
00:09:12,084 --> 00:09:16,871
Ejaaz:
Government and open source weights or open source AI models,

149
00:09:17,097 --> 00:09:20,364
Ejaaz:
where they're hinting at potentially banning open source models,

150
00:09:20,364 --> 00:09:23,694
Ejaaz:
which would like, you know, which would receive, you know, quite a lot of disdain

151
00:09:23,694 --> 00:09:26,364
Ejaaz:
from a lot of founders in America and in the West in general.

152
00:09:26,364 --> 00:09:28,934
Ejaaz:
Now, Anthropic came out with their positioning on this, which is,

153
00:09:29,416 --> 00:09:32,864
Ejaaz:
we have never stated that we want to ban open models.

154
00:09:32,864 --> 00:09:36,794
Ejaaz:
We just want to make sure that we can slow the development enough such that

155
00:09:36,985 --> 00:09:42,324
Ejaaz:
this model, these models, these AGI type intelligences don't escape us and become

156
00:09:42,324 --> 00:09:47,690
Ejaaz:
a problem. So Dario, in his own words, presented three very reasonable options, in my opinion.

157
00:09:48,254 --> 00:09:51,514
Ejaaz:
One, we shouldn't sell powerful chips or chip making equipment to China.

158
00:09:51,514 --> 00:09:55,004
Ejaaz:
The point, of course, that he's making there is China doesn't have next-gen chips.

159
00:09:55,004 --> 00:09:57,704
Ejaaz:
They can't train next-gen models, which they can open source and release to

160
00:09:57,704 --> 00:09:59,994
Ejaaz:
the world to potentially malicious actors.

161
00:09:59,994 --> 00:10:03,314
Ejaaz:
Number two, we have to crack down on distillation operations.

162
00:10:03,314 --> 00:10:06,964
Ejaaz:
For those of you who don't know, distillation is basically the act of taking

163
00:10:06,964 --> 00:10:11,854
Ejaaz:
an already smart release model and using it to create prompts or outputs that

164
00:10:11,854 --> 00:10:15,796
Ejaaz:
you can use to train a completely different model. It's kind of like sealing the blueprint in a way.

165
00:10:16,011 --> 00:10:19,534
Ejaaz:
And Anthropic and other labs like OpenAI and Google banned this explicitly,

166
00:10:19,534 --> 00:10:23,554
Ejaaz:
but China has been accused of potentially doing this at a wide scale.

167
00:10:23,554 --> 00:10:24,771
Ejaaz:
So just making that illegal.

168
00:10:24,997 --> 00:10:29,474
Ejaaz:
And the third one is whatever your model is, if it's sufficiently capable to

169
00:10:29,474 --> 00:10:32,354
Ejaaz:
perform a cybersecurity threat or biothreat or whatever that might be,

170
00:10:32,674 --> 00:10:37,114
Ejaaz:
closed or open, it should go through a framework of processing beforehand had

171
00:10:37,114 --> 00:10:39,754
Ejaaz:
to make sure that it is safe and has the necessary safeguards.

172
00:10:39,754 --> 00:10:43,194
Ejaaz:
Now, of course, the obvious point of contention there is open models can't have

173
00:10:43,194 --> 00:10:46,754
Ejaaz:
safeguards really. You just run it at home privately on your own servers.

174
00:10:47,462 --> 00:10:50,261
Ejaaz:
It's a really good way to kind of shape the conversation. I don't know how far

175
00:10:50,261 --> 00:10:53,459
Ejaaz:
we're going to get, but it was good to see Darius take finally on this.

176
00:10:53,889 --> 00:10:56,731
Josh:
Yeah, it's a big conversation that's been coming up a lot more.

177
00:10:56,731 --> 00:10:59,881
Josh:
Just as everything starts to move faster, open source gets a little bit scarier,

178
00:10:59,881 --> 00:11:01,603
Josh:
closed source gets a little bit more powerful.

179
00:11:01,911 --> 00:11:05,401
Josh:
And there is this contradiction happening in the world where people who are

180
00:11:05,401 --> 00:11:08,789
Josh:
AGI-pilled cannot simultaneously be open source maximalists.

181
00:11:09,091 --> 00:11:14,851
Josh:
Because if you do believe in the impact of agi and what it will do on the world it is difficult to

182
00:11:15,591 --> 00:11:19,870
Josh:
Also say that you want that to be available and open sourced as quickly as the

183
00:11:20,172 --> 00:11:23,098
Josh:
as like progress is kind of making so it's an interesting area

184
00:11:23,464 --> 00:11:26,121
Josh:
um there's a lot of commentary that's been going around this week i think a

185
00:11:26,121 --> 00:11:29,721
Josh:
lot of the news has been around like thought boy pieces commentary duarkesh

186
00:11:29,721 --> 00:11:34,231
Josh:
had one that we really loved he had a bit of a contrarian take where he was

187
00:11:34,231 --> 00:11:36,937
Josh:
basically if i could summarize this i would say everything up

188
00:11:37,245 --> 00:11:41,971
Josh:
everything up gpus up usage up utilization up like up only is how i interpreted

189
00:11:41,971 --> 00:11:43,416
Josh:
this if i could summarize this in two words

190
00:11:43,648 --> 00:11:47,491
Josh:
but basically the idea is that um recently we've had this phenomenon and this

191
00:11:47,491 --> 00:11:50,641
Josh:
is the counter michael burry phenomenon if you remember he said the price of

192
00:11:50,641 --> 00:11:53,963
Josh:
gpus was going to go down over time dora kesha saying hey wait a second

193
00:11:54,178 --> 00:12:00,314
Josh:
um because there is going to be so few gpus and so much high value intelligence

194
00:12:00,581 --> 00:12:04,331
Josh:
that the value of that intelligence per token is actually just going to keep going up

195
00:12:04,614 --> 00:12:07,631
Josh:
and the value of the gpus is actually going to keep going up and if you are

196
00:12:07,631 --> 00:12:11,876
Josh:
trying to lock in rental and compute now that's probably a good idea because

197
00:12:12,497 --> 00:12:16,056
Josh:
There's a fixed amount of capability of running intelligence,

198
00:12:16,056 --> 00:12:19,086
Josh:
but an infinite amount of high quality intelligence that is coming.

199
00:12:19,336 --> 00:12:23,451
Josh:
And the premium on that, while we figure out this dispersion of resources,

200
00:12:23,654 --> 00:12:27,896
Josh:
is going to increase to a level that's fairly high. And the price of these tokens,

201
00:12:27,896 --> 00:12:31,207
Josh:
the price of these GPUs, will just continue to kind of steadily trend upward

202
00:12:31,451 --> 00:12:34,028
Josh:
until we somehow manage to figure out the supply-demand problem.

203
00:12:34,533 --> 00:12:39,206
Ejaaz:
Yeah, I think another really good point that he makes is these AI chips would

204
00:12:39,206 --> 00:12:43,776
Ejaaz:
only make sense to use for the best ai models in the world in that scenario

205
00:12:43,776 --> 00:12:45,655
Ejaaz:
so like just imagine we have

206
00:12:45,917 --> 00:12:51,726
Ejaaz:
claude fable 6 in the future and it is as good as doing the tasks and profession

207
00:12:51,726 --> 00:12:56,232
Ejaaz:
as a professional software engineer that's worth like what is that 250k a year

208
00:12:56,569 --> 00:13:02,456
Ejaaz:
his point is if one ai chip can run claude fable 6 and it can do the job of

209
00:13:02,456 --> 00:13:07,207
Ejaaz:
a 250k a year employee why shouldn't it be worth 250k.

210
00:13:08,126 --> 00:13:11,886
Ejaaz:
And therefore, it starts to price out all the smaller labs like safe super intelligence

211
00:13:11,886 --> 00:13:13,566
Ejaaz:
and stuff that doesn't have as much capital.

212
00:13:13,566 --> 00:13:18,106
Ejaaz:
So it only makes sense for NVIDIA to raise the prices. Everything goes downstream

213
00:13:18,106 --> 00:13:19,186
Ejaaz:
of the model capability.

214
00:13:19,186 --> 00:13:23,496
Ejaaz:
So if NVIDIA knows that their next-gen chip, well, let's call it Feynman in

215
00:13:23,496 --> 00:13:28,116
Ejaaz:
2028, which they're releasing, is good enough to train a superior model at Anthropic

216
00:13:28,116 --> 00:13:34,166
Ejaaz:
or Opening or whatever, they're going to charge them the value of that market share, essentially.

217
00:13:34,166 --> 00:13:38,186
Ejaaz:
So that means if you're a smaller AI lab, you can't really get access to these

218
00:13:38,186 --> 00:13:41,126
Ejaaz:
things. So Dworkesh's point is like the value of these GPUs are actually going

219
00:13:41,126 --> 00:13:44,536
Ejaaz:
to go up only. And if you look at any of the recent earnings reports from Microsoft

220
00:13:44,536 --> 00:13:48,166
Ejaaz:
and stuff, their cloud service providing business has gone up.

221
00:13:48,586 --> 00:13:50,988
Ejaaz:
These revenue numbers are jumping up like 40 to 50%.

222
00:13:51,679 --> 00:13:56,314
Ejaaz:
Sam confirms this as well. He says it's also a demand problem where Sam,

223
00:13:56,314 --> 00:14:00,694
Ejaaz:
who is notorious for having spent or committed, I think it's like $500 billion,

224
00:14:00,694 --> 00:14:04,640
Ejaaz:
like half a trillion dollars to compute over the next five to 10 years.

225
00:14:04,965 --> 00:14:09,384
Ejaaz:
He says on this podcast this week that it wasn't enough and that he needs to

226
00:14:09,384 --> 00:14:12,014
Ejaaz:
be even more aggressive. So I don't know who's right at this point,

227
00:14:12,014 --> 00:14:14,671
Ejaaz:
but two very bullish posts from Dwarkesh at this point.

228
00:14:14,886 --> 00:14:17,394
Josh:
Two very bullish takes. And we have another leader in the space,

229
00:14:17,394 --> 00:14:21,324
Josh:
this one in the world of security, and that is Ledger, where if you are building with

230
00:14:21,624 --> 00:14:25,574
Josh:
agents and you're worried about security ledger has a solution for you there's

231
00:14:25,574 --> 00:14:29,770
Josh:
a three-step process the agents propose the humans approve and the ledger signers

232
00:14:29,921 --> 00:14:33,534
Josh:
enforce so they have this thing called the ledger agent stack which is a series

233
00:14:33,534 --> 00:14:35,320
Josh:
of open source tools that let your agent work

234
00:14:35,622 --> 00:14:39,174
Josh:
with crypto while you stay in full control of the entire thing you could think

235
00:14:39,174 --> 00:14:43,304
Josh:
of it like auto mode with cloud code except you have to approve at each step

236
00:14:43,304 --> 00:14:46,674
Josh:
of the way and this works with all of your favorite providers this works with

237
00:14:46,674 --> 00:14:48,828
Josh:
codex it works with cloud code it works with Courser.

238
00:14:49,066 --> 00:14:52,614
Josh:
It is open source and is fully available today. So if you are building with

239
00:14:52,614 --> 00:14:56,624
Josh:
agents, go and check them out at developers.ledger.com. It is linked in our

240
00:14:56,624 --> 00:15:00,251
Josh:
description below. Thank you so much to Ledger for supporting this episode.

241
00:15:00,448 --> 00:15:03,744
Josh:
And now we have to get into earnings season, of course, because,

242
00:15:03,744 --> 00:15:08,511
Josh:
you know, I mean, we do talk about stocks quite a bit and the earnings are all over the place. I mean,

243
00:15:08,929 --> 00:15:13,113
Josh:
We saw the semiconductors and the memory stocks get absolutely crushed,

244
00:15:13,113 --> 00:15:17,893
Josh:
but there is one company who is not, and that is Microsoft, who had $90 billion

245
00:15:18,493 --> 00:15:23,604
Josh:
in revenue with estimate per share of $4.75 versus $4.25.

246
00:15:23,726 --> 00:15:27,323
Josh:
So it looks like, I mean, things are up only at Microsoft. I'm taking a look

247
00:15:27,323 --> 00:15:32,353
Josh:
at their cloud revenue, which is up 43% year over year. I mean,

248
00:15:32,353 --> 00:15:33,548
Josh:
this is a great business.

249
00:15:33,717 --> 00:15:36,943
Josh:
Who would have thought Microsoft, the company that's never released a single AI product.

250
00:15:36,943 --> 00:15:39,963
Ejaaz:
The market absolutely loves it. Like their stock this morning as of recording

251
00:15:39,963 --> 00:15:43,190
Ejaaz:
is up 15%, which is pretty big for a big boy like Microsoft.

252
00:15:43,417 --> 00:15:48,273
Ejaaz:
Now listen, the number one highlight from their earnings is their Azure earnings.

253
00:15:48,273 --> 00:15:51,233
Ejaaz:
So that's their cloud service provider, which they use to basically lease out

254
00:15:51,233 --> 00:15:55,673
Ejaaz:
compute to any AI model lab or AI company that wants to run or train their,

255
00:15:55,673 --> 00:15:56,953
Ejaaz:
or inference their own models.

256
00:15:56,953 --> 00:16:01,033
Ejaaz:
And that's what Microsoft's cloud businesses focus on. That is up 43% and is

257
00:16:01,033 --> 00:16:04,513
Ejaaz:
the thing that everyone is looking at when you look at hyperscalers specifically.

258
00:16:04,513 --> 00:16:07,513
Ejaaz:
So Microsoft, Google, and Amazon.

259
00:16:07,893 --> 00:16:10,953
Ejaaz:
And it's a really good business. And I think the profit margins for all of the

260
00:16:10,953 --> 00:16:13,271
Ejaaz:
revenue that they're bringing in is incredibly impressive.

261
00:16:13,514 --> 00:16:19,354
Ejaaz:
Moving on, there's a few more companies that we covered in the episode yesterday, but TLDR is Samsung.

262
00:16:19,813 --> 00:16:23,284
Ejaaz:
That little memory company where their stock is dumping is doing extremely well.

263
00:16:23,482 --> 00:16:27,173
Ejaaz:
They're signing a lot of LTAs, which are long-term agreements with future customers.

264
00:16:27,173 --> 00:16:32,293
Ejaaz:
So they're locking it in until like, I think, 2028. So 2027's memory supply

265
00:16:32,293 --> 00:16:33,925
Ejaaz:
might already be accounted for.

266
00:16:34,169 --> 00:16:36,873
Ejaaz:
And then we have Intel. Josh, you know, you just spoke about agents.

267
00:16:37,433 --> 00:16:41,333
Ejaaz:
Intel provides the number one bit of infrastructure to help those agents do

268
00:16:41,333 --> 00:16:42,795
Ejaaz:
their thing called CPUs.

269
00:16:43,050 --> 00:16:46,063
Ejaaz:
Intel and AMD, I don't know when AMD's earnings is coming out.

270
00:16:46,731 --> 00:16:48,903
Ejaaz:
Has done really well and they had a banging quarter.

271
00:16:48,903 --> 00:16:53,783
Ejaaz:
So the TLDR is these companies have reported amazing revenue earnings.

272
00:16:53,783 --> 00:16:57,303
Ejaaz:
Their profit margins are going up because of AI, but the stocks are kind of

273
00:16:57,303 --> 00:17:01,683
Ejaaz:
like very volatile, except for the one company that is not doing anything in

274
00:17:01,683 --> 00:17:04,973
Ejaaz:
AI right now that just became the number one most valuable company in the world.

275
00:17:05,353 --> 00:17:08,521
Ejaaz:
We're talking about Apple. Congratulations to Apple. Who's gone on a historic run. Yep.

276
00:17:09,613 --> 00:17:13,187
Josh:
Yeah, I mean, Apple, they deserve this, honestly. Siri is going to be incredible.

277
00:17:13,187 --> 00:17:15,678
Josh:
They're going to have two amazing devices with the iPhone Fold,

278
00:17:15,894 --> 00:17:18,667
Josh:
the iPhone 19, or whatever number we're at now.

279
00:17:20,967 --> 00:17:23,707
Josh:
They're in such a great position for running local inference on machines,

280
00:17:23,944 --> 00:17:25,262
Josh:
and I think people are recognizing that.

281
00:17:25,465 --> 00:17:28,292
Josh:
So I expect to see a lot more coming down the pipe from Apple.

282
00:17:28,478 --> 00:17:30,441
Josh:
I mean, congrats to Apple. This is great news.

283
00:17:31,017 --> 00:17:33,837
Josh:
We also have Kimi K3, as we round up this rapid fire section,

284
00:17:34,137 --> 00:17:37,727
Josh:
who released their open source weights. And remember, we get excited about running

285
00:17:37,727 --> 00:17:42,172
Josh:
open source models at home. Well, it turns out it costs a little bit more than you think it might.

286
00:17:42,276 --> 00:17:45,897
Josh:
It looks like it might cost about, what? $35,000 a month.

287
00:17:45,897 --> 00:17:49,497
Ejaaz:
Dude, it costs like a million dollars per year. It's crazy.

288
00:17:49,497 --> 00:17:53,377
Josh:
No, no, no, no, no, no, no. Half a million dollars. Half a million dollars.

289
00:17:53,737 --> 00:17:54,347
Ejaaz:
One time.

290
00:17:54,347 --> 00:17:58,427
Josh:
So only like, yeah. So maybe twice as many people can do that as a one-time

291
00:17:58,427 --> 00:18:01,333
Josh:
purchase. But yeah, Kimi is expensive. that these model weights are ginormous.

292
00:18:01,658 --> 00:18:04,577
Josh:
But it is open source at least. So hey, if you got half a million dollars laying

293
00:18:04,577 --> 00:18:06,853
Josh:
around, you want to play with it. This would be your time to do it.

294
00:18:07,091 --> 00:18:11,201
Ejaaz:
And that is it. You are all caught up for the week on AI.

295
00:18:11,358 --> 00:18:14,237
Ejaaz:
Let us know if you've missed anything in the comments. But wherever you're listening,

296
00:18:14,237 --> 00:18:18,289
Ejaaz:
if you're listening to us on YouTube, Spotify, Apple, subscribe, leave us a comment.

297
00:18:18,458 --> 00:18:21,389
Ejaaz:
Make sure you give us an amazing rating. I think we're a pretty good show personally.

298
00:18:22,297 --> 00:18:25,037
Ejaaz:
And we have a newsletter that goes out twice a week. Definitely subscribe to

299
00:18:25,037 --> 00:18:28,837
Ejaaz:
that. And Josh, is there any final comments that we want before the week?

300
00:18:28,837 --> 00:18:31,217
Josh:
Share it with your friends. Let them know how much you love the show.

301
00:18:31,217 --> 00:18:33,827
Josh:
If you think they might be interested, go touch some grass for this weekend

302
00:18:33,827 --> 00:18:37,497
Josh:
because you are now fully caught up on all things AI and investing in frontier

303
00:18:37,497 --> 00:18:38,814
Josh:
technology and all the good things.

304
00:18:39,017 --> 00:18:42,564
Josh:
So have an amazing weekend and we will see you all when we come back next week.