1
00:00:00,020 --> 00:00:04,100
Josh:
Everyone needs to stop forgetting about Meta, the social media company turned

2
00:00:04,100 --> 00:00:08,620
Josh:
AI company, because this week they have released quite a few new plans and options

3
00:00:08,620 --> 00:00:11,000
Josh:
that I think are going to be interesting to someone.

4
00:00:11,180 --> 00:00:14,380
Josh:
If you're watching this, chances are you will be affected by these new subscriptions.

5
00:00:14,560 --> 00:00:17,340
Josh:
What we're going to do is not only share the new meta subscriptions,

6
00:00:17,500 --> 00:00:21,060
Josh:
but also compare them to all the existing subscriptions that you can go and

7
00:00:21,060 --> 00:00:25,380
Josh:
pay for, whether it be for social media platforms like X or through AI like

8
00:00:25,380 --> 00:00:27,000
Josh:
Gemini, Claude, Anthropic.

9
00:00:27,000 --> 00:00:31,660
Josh:
There is now a full spread of all the different types of subscriptions you could

10
00:00:31,660 --> 00:00:32,580
Josh:
have to different models.

11
00:00:32,780 --> 00:00:35,360
Josh:
And we're going to walk through them systematically, starting with Meta,

12
00:00:35,500 --> 00:00:40,420
Josh:
who started on the social front this week, offering $4 plans for Instagram,

13
00:00:40,760 --> 00:00:41,980
Josh:
Facebook, and WhatsApp.

14
00:00:42,580 --> 00:00:45,420
Ejaaz:
Yeah, exactly. So Meta is getting into the subscription game,

15
00:00:45,580 --> 00:00:48,220
Ejaaz:
which they're probably following in the lead of like Anthropik and OpenAIR.

16
00:00:48,360 --> 00:00:51,620
Ejaaz:
They're seeing these annual recurring revenue numbers just kind of skyrocket.

17
00:00:51,740 --> 00:00:53,140
Ejaaz:
And they're like, I want some of that as well.

18
00:00:53,300 --> 00:00:58,020
Ejaaz:
Also, So like a bit of important context here, Meta's now spent around $30 billion

19
00:00:58,020 --> 00:01:03,320
Ejaaz:
between AI CapEx GPUs and like acquiring talent and trying to build their own

20
00:01:03,320 --> 00:01:06,220
Ejaaz:
AI products. But I mean, Josh, I think you share the same opinion.

21
00:01:06,770 --> 00:01:10,270
Ejaaz:
Which of the meta AI products have you actually used. This is their kind of

22
00:01:10,270 --> 00:01:14,410
Ejaaz:
like foray to try and get people to use it. So it comes in the form of a few

23
00:01:14,410 --> 00:01:15,110
Ejaaz:
different subscriptions.

24
00:01:15,370 --> 00:01:21,290
Ejaaz:
Now, the consumer ones look like Instagram Plus for not $4, but $3.99 a month.

25
00:01:21,390 --> 00:01:25,750
Ejaaz:
You've got Facebook Plus at $3.99 a month, and you've got WhatsApp Plus at $2.99 a month.

26
00:01:25,910 --> 00:01:29,250
Ejaaz:
Now, if you're wondering like why anyone would be using these kind of like individual

27
00:01:29,250 --> 00:01:33,350
Ejaaz:
apps and individual subscriptions, it's because Instagram kind of caters towards

28
00:01:33,350 --> 00:01:36,290
Ejaaz:
a content creator-focused type of platform.

29
00:01:36,630 --> 00:01:40,870
Ejaaz:
Facebook Plus, a lot of enterprises and business run their business on there via ads.

30
00:01:41,110 --> 00:01:44,370
Ejaaz:
The ad business for Meta alone is absolutely huge.

31
00:01:44,550 --> 00:01:47,430
Ejaaz:
And they're actually growing at a faster rate than Google did at the equivalent

32
00:01:47,430 --> 00:01:49,750
Ejaaz:
time that they existed in their lifecycle right now.

33
00:01:49,890 --> 00:01:53,070
Ejaaz:
And then you've got WhatsApp as well, which handles a lot of customer support

34
00:01:53,070 --> 00:01:56,990
Ejaaz:
for a lot of these enterprises. So you can think about one particular user,

35
00:01:56,990 --> 00:01:59,730
Ejaaz:
not just subscribing to one of these subscriptions, but many.

36
00:01:59,950 --> 00:02:03,490
Ejaaz:
And then collectively, there's this new enterprise plan. I think it's on the

37
00:02:03,490 --> 00:02:06,610
Ejaaz:
MetaOne AI subscription plan, which we'll get into in a second,

38
00:02:07,030 --> 00:02:09,270
Ejaaz:
which is kind of like this collective suite of access.

39
00:02:09,470 --> 00:02:12,370
Ejaaz:
Now, the thing that jumped out to me the most here is

40
00:02:12,760 --> 00:02:15,680
Ejaaz:
honestly, how cheap the accessibility is here.

41
00:02:15,800 --> 00:02:19,560
Ejaaz:
If you are a content creator, the chances are it's a very competitive game and

42
00:02:19,560 --> 00:02:22,940
Ejaaz:
you're not really earning as much money as like some of these like Fortune 500 companies.

43
00:02:23,360 --> 00:02:27,180
Ejaaz:
Now, if you want to get access to some, maybe not frontier intelligence,

44
00:02:27,780 --> 00:02:31,120
Ejaaz:
But a model that's good enough to do the thing that you're hyper-focused on,

45
00:02:31,260 --> 00:02:35,080
Ejaaz:
such as advertising or creating content or using the right SEO-timed words,

46
00:02:35,480 --> 00:02:36,880
Ejaaz:
meta is in the attention game.

47
00:02:37,020 --> 00:02:39,720
Ejaaz:
That is what all their models have been trained on, using their own data.

48
00:02:39,820 --> 00:02:43,520
Ejaaz:
They've been in this game for over a decade and they know how to capture people's attention.

49
00:02:43,700 --> 00:02:47,020
Ejaaz:
That's very useful for the particular type of users that they're going after.

50
00:02:47,180 --> 00:02:50,660
Ejaaz:
But I wonder how this compares directly to some of the other major frontier

51
00:02:50,660 --> 00:02:52,140
Ejaaz:
labs like Google or Anthropic.

52
00:02:52,380 --> 00:02:56,460
Josh:
Yeah, a helpful way of looking at Metis plans is there's three different types of them.

53
00:02:56,620 --> 00:03:00,800
Josh:
One is the one that we just mentioned for $4 a month, you get access to Facebook

54
00:03:00,800 --> 00:03:02,560
Josh:
and Instagram plus, which gives

55
00:03:02,560 --> 00:03:05,540
Josh:
access to extra features like profile customization, super reactions,

56
00:03:06,060 --> 00:03:09,700
Josh:
story insights, among other things this is still very new we're still figuring

57
00:03:09,700 --> 00:03:13,100
Josh:
out what exactly this includes second to this sitting on top of it are their

58
00:03:13,100 --> 00:03:18,340
Josh:
standard plus premium and max plans those are if you are an actual creator on

59
00:03:18,340 --> 00:03:21,860
Josh:
the platform that gives you the verified badge if you see people with blue check marks

60
00:03:22,200 --> 00:03:25,180
Josh:
not on a platform like X, they are paying for these plans.

61
00:03:25,360 --> 00:03:30,480
Josh:
That includes lots of things noteworthy. It doesn't include the skipping of

62
00:03:30,480 --> 00:03:32,800
Josh:
ads. No matter what, you cannot escape ads, it appears.

63
00:03:33,240 --> 00:03:36,640
Josh:
And then third to this, we have Meta's AI plan. And Meta's AI plan is the one

64
00:03:36,640 --> 00:03:37,220
Josh:
that we're going to sit on.

65
00:03:37,300 --> 00:03:39,940
Josh:
We're going to spend some time on and we're going to start comparing to the

66
00:03:39,940 --> 00:03:42,160
Josh:
other companies because it's starting pretty cheap.

67
00:03:42,460 --> 00:03:46,420
Josh:
Meta's AI subscription service starts at $8 a month.

68
00:03:46,620 --> 00:03:49,860
Josh:
What does that get you? We're not entirely sure just yet. It's

69
00:03:49,860 --> 00:03:53,240
Josh:
called meta one plus and it will give you access to presumably the latest

70
00:03:53,240 --> 00:03:56,180
Josh:
in meta's ai tools what those ai tools look like

71
00:03:56,180 --> 00:04:00,780
Josh:
we're not entirely sure this is all very new and novel but the second thing

72
00:04:00,780 --> 00:04:04,840
Josh:
is the 20 a month plan and that 20 a month plan is seeming to be the one that

73
00:04:04,840 --> 00:04:09,620
Josh:
wants to compete head to head with the other companies like claude and chat

74
00:04:09,620 --> 00:04:13,920
Josh:
gpt and gemini so how does it compare is what we're going to get into here i.

75
00:04:13,920 --> 00:04:17,700
Ejaaz:
Think a good place to start is uh looking at their creative plans and pricing

76
00:04:17,700 --> 00:04:19,740
Ejaaz:
that they have for their existing subscriptions.

77
00:04:20,260 --> 00:04:23,440
Ejaaz:
Now, this isn't the latest subscriptions that we just spoke about,

78
00:04:23,560 --> 00:04:27,980
Ejaaz:
but it kind of implies what the other kind of new subscription types will include.

79
00:04:28,160 --> 00:04:32,240
Ejaaz:
Now, I've noticed that across Plus, Premium, and Max, you have this thing being

80
00:04:32,240 --> 00:04:34,920
Ejaaz:
brandished right in front of you, which is protect your brand.

81
00:04:35,100 --> 00:04:38,740
Ejaaz:
I think a lot of Meta's users who are business users who care about the kind

82
00:04:38,740 --> 00:04:42,400
Ejaaz:
of social image and branding online across all their different platforms...

83
00:04:42,770 --> 00:04:45,470
Ejaaz:
Get a lot of impersonators so this is something that like i

84
00:04:45,470 --> 00:04:48,170
Ejaaz:
think ai is going to result in a proliferation of it's

85
00:04:48,170 --> 00:04:51,470
Ejaaz:
just kind of like fake content people that copy your content copyright issues etc

86
00:04:51,470 --> 00:04:54,330
Ejaaz:
this kind of like collectively protects you against

87
00:04:54,330 --> 00:04:57,370
Ejaaz:
that i'm wondering how that actually works but it seems like there's some

88
00:04:57,370 --> 00:05:01,610
Ejaaz:
form of like agent interaction uh whether they're human or digital and there's

89
00:05:01,610 --> 00:05:05,810
Ejaaz:
also maximized discovery i think this is really where meta kind of like shines

90
00:05:05,810 --> 00:05:10,650
Ejaaz:
compared to like the google ai models the anthropic ai models and open air themselves

91
00:05:10,650 --> 00:05:15,910
Ejaaz:
um Meta really has been focused on the search engine optimization role of things.

92
00:05:16,230 --> 00:05:21,670
Ejaaz:
Now, in a world of AI, you can think that the search engine kind of like evolves quite a bit.

93
00:05:21,770 --> 00:05:25,010
Ejaaz:
Google's been speaking about this quite a bit, which is the AI model itself

94
00:05:25,010 --> 00:05:27,090
Ejaaz:
or the AI agent ends up becoming the customer.

95
00:05:27,230 --> 00:05:29,610
Ejaaz:
And you need to cater to an AI agent versus a human.

96
00:05:29,870 --> 00:05:34,490
Ejaaz:
That SEO role looks very different. And Meta is trying to pioneer what that model looks like.

97
00:05:34,730 --> 00:05:38,010
Ejaaz:
The first model that they released a few months ago, I believe, is MewSpark.

98
00:05:38,110 --> 00:05:43,550
Ejaaz:
And it excels at this type of work. Whereas compared to like regular LLM work, it kind of sucks.

99
00:05:43,810 --> 00:05:46,970
Ejaaz:
Now, I want to look at like a direct comparison between Google,

100
00:05:47,150 --> 00:05:50,650
Ejaaz:
Meta and the other likes. And we have this really cool artifact here.

101
00:05:50,830 --> 00:05:53,930
Josh:
Yes. So thank you, Claude, for creating this artifact. We are clearly paying

102
00:05:53,930 --> 00:05:56,390
Josh:
for the Claude plan, or at least I am to say the least.

103
00:05:56,510 --> 00:05:59,470
Josh:
But this is a really helpful table that compares the different types of offerings

104
00:05:59,470 --> 00:06:04,250
Josh:
from all these companies. We have Meta, Google, Claude, ChatGPT and X slash Grok.

105
00:06:04,530 --> 00:06:08,550
Josh:
The entry level pricing is something that is used by almost all companies except

106
00:06:08,550 --> 00:06:10,010
Josh:
for Claude at $8 per month.

107
00:06:10,010 --> 00:06:13,050
Josh:
You can get access to meta one google's ai plan chat gpt

108
00:06:13,050 --> 00:06:17,450
Josh:
and x slash grok on x that gives you the blue check mark plus access to grok

109
00:06:17,450 --> 00:06:22,270
Josh:
chat gpt gives you the go plan so you get ads inside of your offerings that

110
00:06:22,270 --> 00:06:26,890
Josh:
when you type into your llm claude noteworthy doesn't do ads so they have no

111
00:06:26,890 --> 00:06:31,010
Josh:
plan at this level google plus is ai supported and then the same with meta one

112
00:06:31,010 --> 00:06:32,890
Josh:
now standard is the 20 a month.

113
00:06:33,630 --> 00:06:36,450
Josh:
X is noteworthy and the fact that they don't have one they

114
00:06:36,450 --> 00:06:39,290
Josh:
actually started $30 but among the others this is

115
00:06:39,290 --> 00:06:42,550
Josh:
the point where i think a lot of people will pay the most attention because people

116
00:06:42,550 --> 00:06:46,190
Josh:
are willing to pay $20 a month and in most cases it's very much worth it if

117
00:06:46,190 --> 00:06:50,910
Josh:
you are using a chat gpt plan if you are using a cloud plan that's free you're

118
00:06:50,910 --> 00:06:55,790
Josh:
currently on an old cheap model that doesn't really work nearly as well as the

119
00:06:55,790 --> 00:06:59,650
Josh:
frontier models are and if you want to judge the progress of ai models,

120
00:06:59,850 --> 00:07:03,030
Josh:
you really need to be on this $20 a month plan because it at least gives you

121
00:07:03,030 --> 00:07:06,030
Josh:
access to the Frontier models as they're being released.

122
00:07:06,190 --> 00:07:11,790
Josh:
To Opus 4.7, to ChatGPT 5.5, and Gemini and soon Meta's new offerings.

123
00:07:11,950 --> 00:07:16,190
Josh:
So if we're comparing these, the standard level, I'm not sure who would really

124
00:07:16,190 --> 00:07:21,090
Josh:
want to pay $20 for Meta's offering when you can go and get ChatGPT premium,

125
00:07:21,390 --> 00:07:25,490
Josh:
Claude premium, Google premium even at the same $20 price point. That's a bit difficult.

126
00:07:25,750 --> 00:07:28,590
Josh:
Then these go up and really just the the higher tiers

127
00:07:28,590 --> 00:07:31,390
Josh:
of this are more tokens so there's not a huge difference in

128
00:07:31,390 --> 00:07:34,190
Josh:
terms of feature set it's mostly just the ability to use it

129
00:07:34,190 --> 00:07:38,350
Josh:
more productively so as a starting point the 20 month plans are all very competitive

130
00:07:38,350 --> 00:07:43,010
Josh:
except for the fact that two of these are frontier models and the others are

131
00:07:43,010 --> 00:07:46,890
Josh:
not so if you're looking for frontier level intelligence top intelligence per

132
00:07:46,890 --> 00:07:51,470
Josh:
token it's really cloud and chat gpt google has a lot of fun tools if that's

133
00:07:51,470 --> 00:07:52,490
Josh:
where you like to spend your time.

134
00:07:52,930 --> 00:07:55,990
Josh:
But aside from that, Meta and Grok's offerings leave a lot to be desired.

135
00:07:56,530 --> 00:08:00,330
Ejaaz:
I'm kind of split on this, okay? So like on the one half, I agree with you.

136
00:08:00,590 --> 00:08:03,910
Ejaaz:
On the other half, I'm like, I do think Meta is going after a different type of market.

137
00:08:04,050 --> 00:08:10,830
Ejaaz:
Like when I look at like a $7.99 a month plan, I instantly think of OpenAI's ChatGPT Go plan.

138
00:08:10,970 --> 00:08:12,870
Ejaaz:
Do you remember they launched that specifically just in India?

139
00:08:12,990 --> 00:08:15,110
Ejaaz:
And I think it was like five bucks a month.

140
00:08:15,210 --> 00:08:19,050
Ejaaz:
That resulted in adding, I think it was like tens of millions of new users over

141
00:08:19,050 --> 00:08:20,950
Ejaaz:
the next couple of weeks since they launched it, right?

142
00:08:21,150 --> 00:08:23,670
Ejaaz:
And I think that Metz is trying to do a similar thing here. They're going for

143
00:08:23,670 --> 00:08:26,130
Ejaaz:
a very specific archetype of individual, which is like,

144
00:08:26,410 --> 00:08:29,230
Ejaaz:
You're out in the content game. They understand that, like, you may not have

145
00:08:29,230 --> 00:08:32,330
Ejaaz:
an abundance of money yet, but they want to kind of, like, lead you on that exponential.

146
00:08:32,590 --> 00:08:35,310
Ejaaz:
And they know that they have search and discovery kind of, like,

147
00:08:35,410 --> 00:08:38,150
Ejaaz:
nailed to the bone. And so they want to kind of, like, offer that or be the

148
00:08:38,150 --> 00:08:39,970
Ejaaz:
first major AI provider for that.

149
00:08:40,190 --> 00:08:44,650
Ejaaz:
Now, on the other hand of things, like, I can imagine as I see all these,

150
00:08:44,650 --> 00:08:47,830
Ejaaz:
like, new models released from Anthropic and OpenAI, the next model release

151
00:08:47,830 --> 00:08:50,310
Ejaaz:
means that the previous models released is, like, even cheaper.

152
00:08:50,470 --> 00:08:53,550
Ejaaz:
In fact, the newer model releases end up being cheaper than the old model releases,

153
00:08:53,670 --> 00:08:56,810
Ejaaz:
which is kind of antithetical because you like get a smarter model which

154
00:08:56,810 --> 00:08:59,510
Ejaaz:
is kind of cheaper than the previous model so it's like bang for your

155
00:08:59,510 --> 00:09:02,390
Ejaaz:
buck i also think that with more gpus and

156
00:09:02,390 --> 00:09:05,350
Ejaaz:
semiconductors as that all scales you end up just getting uh

157
00:09:05,350 --> 00:09:09,530
Ejaaz:
more tokens per watt i believe it's performance per watt it's just going to

158
00:09:09,530 --> 00:09:14,190
Ejaaz:
end up kind of driving lower so i don't know about the staying power of meta

159
00:09:14,190 --> 00:09:17,990
Ejaaz:
subscriptions here their models do need to improve pretty rapidly and this kind

160
00:09:17,990 --> 00:09:21,790
Ejaaz:
of leads me on to like an overarching theme that i have on meta which is this

161
00:09:21,790 --> 00:09:23,610
Ejaaz:
is actually like good news for Meta.

162
00:09:23,850 --> 00:09:27,630
Ejaaz:
And this is desperately needed by Meta because they've blown around $30 billion

163
00:09:27,630 --> 00:09:32,350
Ejaaz:
over the last 12 months, acquiring talent and investing in this AI thing with

164
00:09:32,350 --> 00:09:34,910
Ejaaz:
nothing to show for it. They released a MuseSpark model.

165
00:09:35,030 --> 00:09:36,530
Ejaaz:
It was kind of like whatever people have forgotten about it.

166
00:09:36,610 --> 00:09:40,530
Ejaaz:
People aren't really kind of using it aside from people who actually use Meta's different platforms.

167
00:09:40,690 --> 00:09:45,290
Ejaaz:
But this could technically or theoretically on paper result in kind of like a goldmine for them.

168
00:09:45,490 --> 00:09:49,270
Ejaaz:
If you imagine like they can convert like 100 million users for like their $20

169
00:09:49,270 --> 00:09:52,150
Ejaaz:
dollar a month plan, that is reasonable income coming in.

170
00:09:52,290 --> 00:09:55,910
Ejaaz:
But I've also seen comments from Zuck earlier this week, which talks about if

171
00:09:55,910 --> 00:09:57,930
Ejaaz:
this doesn't play out, this subscription thing,

172
00:09:58,130 --> 00:10:01,410
Ejaaz:
if their AI models aren't used by a lot of people, they're willing to go down

173
00:10:01,410 --> 00:10:04,070
Ejaaz:
the NeoCloud route, which is what Elon Musk has done recently,

174
00:10:04,210 --> 00:10:07,950
Ejaaz:
and partner with major AI labs to rent out their compute and data classes,

175
00:10:08,110 --> 00:10:09,770
Ejaaz:
which they've invested billions of dollars in.

176
00:10:09,970 --> 00:10:14,410
Ejaaz:
So I think Zuck is being pressured quite a lot by shareholders at this point.

177
00:10:14,530 --> 00:10:18,030
Ejaaz:
He's being forced to kind of show some kind of revenue, he fired or laid off

178
00:10:18,030 --> 00:10:20,470
Ejaaz:
8,000 plus people over the last seven days.

179
00:10:20,710 --> 00:10:24,410
Ejaaz:
And that was like one of like, I think five cuts in the last couple of months.

180
00:10:24,590 --> 00:10:28,330
Ejaaz:
So Meta's like backed into a corner right now and they need to kind of like

181
00:10:28,330 --> 00:10:30,490
Ejaaz:
show the goods or they need to pivot pretty aggressively.

182
00:10:31,050 --> 00:10:34,210
Josh:
You mentioned the word neocloud, which is funny because I have some neocloud news.

183
00:10:34,350 --> 00:10:36,550
Josh:
This is, again, if you're listening to this, this is a Friday roundup.

184
00:10:36,650 --> 00:10:38,270
Josh:
We're covering all the news. So we're going to speed run this.

185
00:10:38,430 --> 00:10:40,410
Josh:
It's a very busy morning at Limitless HQ.

186
00:10:40,590 --> 00:10:42,050
Josh:
So we're going to try to get through these as quick as we can,

187
00:10:42,270 --> 00:10:46,070
Josh:
starting with the StarCloud news, which I found to be really interesting.

188
00:10:46,870 --> 00:10:50,330
Josh:
Elon and SpaceX have a new partner. They are working with, it's funny enough,

189
00:10:50,550 --> 00:10:54,750
Josh:
Gavin Baker here is in the photo. We just recorded an episode with him yesterday about his portfolio.

190
00:10:55,190 --> 00:10:58,290
Josh:
Clearly he is bullish on space. And the announcement is that StarCloud

191
00:10:58,290 --> 00:11:00,930
Josh:
is building star cloud three that is going to

192
00:11:00,930 --> 00:11:03,870
Josh:
be launching on the new spacex starship this is

193
00:11:03,870 --> 00:11:06,550
Josh:
starship version three this is the test flight that they just had last week that

194
00:11:06,550 --> 00:11:09,830
Josh:
we covered and the idea is that that three-ton

195
00:11:09,830 --> 00:11:14,950
Josh:
spacecraft is designed to fit inside of the pes dispenser inside of starship

196
00:11:14,950 --> 00:11:18,670
Josh:
the ceo philip johnson claims that star cloud three would be the first orbital

197
00:11:18,670 --> 00:11:22,550
Josh:
data center that's actually cost competitive with terrestrial at about five

198
00:11:22,550 --> 00:11:27,810
Josh:
cents per kilowatt if commercial costs land near $500 per kilogram.

199
00:11:28,030 --> 00:11:31,670
Josh:
So again, this is fully contingent on Starship being able to get it up into orbit.

200
00:11:31,830 --> 00:11:38,790
Josh:
The idea is that commercial access should open up around 2028-2029 given the

201
00:11:38,790 --> 00:11:42,690
Josh:
amount of time that it's going to take to get Starship online and then start launching these,

202
00:11:43,310 --> 00:11:46,830
Josh:
aircraft in space but it's a really cool announcement where we have an actual,

203
00:11:47,390 --> 00:11:50,890
Josh:
client for SpaceX that is going to be launching serious payloads into space

204
00:11:50,890 --> 00:11:53,830
Josh:
and fully pursuing the AI data centers in orbit plan.

205
00:11:54,110 --> 00:11:57,390
Josh:
It's going to take a little while but this is where we build the foundation

206
00:11:57,390 --> 00:12:01,650
Josh:
this is where companies are starting to plan these launches for the next couple of years.

207
00:12:02,110 --> 00:12:06,990
Ejaaz:
We have another SpaceX AI alliance member Josh you know over the last couple

208
00:12:06,990 --> 00:12:11,630
Ejaaz:
of weeks you've got Elon partnering with Anthropic twice you've got them partnering

209
00:12:11,630 --> 00:12:12,770
Ejaaz:
pretty aggressively with cursor,

210
00:12:12,930 --> 00:12:16,770
Ejaaz:
which is going to result in a $60 billion acquisition 30 days after the IPO,

211
00:12:16,910 --> 00:12:18,790
Ejaaz:
which is expected to be sometime next month.

212
00:12:19,250 --> 00:12:22,470
Ejaaz:
And now you've got this partnership with StarCloud. Now, what I love about this

213
00:12:22,470 --> 00:12:27,310
Ejaaz:
is if you think of SpaceX owning the highway or the toll booth into space,

214
00:12:27,670 --> 00:12:30,630
Ejaaz:
they're going to need the goods to be putting out into space, right?

215
00:12:30,730 --> 00:12:33,270
Ejaaz:
So they've got Anthropik to kind of tentatively agree to say like,

216
00:12:33,330 --> 00:12:36,530
Ejaaz:
yeah, we're going to actually use some of your space AI data centers,

217
00:12:36,530 --> 00:12:39,030
Ejaaz:
but also you need the goods to get out there.

218
00:12:39,210 --> 00:12:42,930
Ejaaz:
Now, Elon Musk is investing a ton of money over the next five years into something

219
00:12:42,930 --> 00:12:47,350
Ejaaz:
called the TerraFab, which will end up becoming the largest chip fab in America

220
00:12:47,350 --> 00:12:49,390
Ejaaz:
if it plays out how he plans it to.

221
00:12:49,670 --> 00:12:53,250
Ejaaz:
And there's an interesting statistic from that goal, which is 70% of the chips

222
00:12:53,250 --> 00:12:56,870
Ejaaz:
that are produced are going to be radiation hardened and sent out into space.

223
00:12:56,870 --> 00:13:00,710
Ejaaz:
So he's very much like committing his entire focus and goal into this.

224
00:13:00,810 --> 00:13:06,110
Ejaaz:
That's why the SpaceX IPO is valued at expectedly a $2 trillion valuation.

225
00:13:06,110 --> 00:13:10,210
Ejaaz:
But he needs the goods to be able to do that now I'm looking at this tweet up here

226
00:13:10,760 --> 00:13:13,840
Ejaaz:
Saying that StarCloud recently, or is in the process of raising $200 million

227
00:13:13,840 --> 00:13:16,180
Ejaaz:
at a $2 billion valuation, that is cheap.

228
00:13:16,460 --> 00:13:21,080
Ejaaz:
If StarCloud can actually put out radiation-hardened GPUs, which, by the way, they have.

229
00:13:21,200 --> 00:13:25,120
Ejaaz:
There's currently, I think it's like a stack of like H100s that are currently

230
00:13:25,120 --> 00:13:28,720
Ejaaz:
orbiting space that are training and inferencing on it, as Gavin mentioned in the previous tweet.

231
00:13:28,960 --> 00:13:32,300
Ejaaz:
What's stopping Elon from acquiring them and now having the actual goods to

232
00:13:32,300 --> 00:13:34,540
Ejaaz:
be able to do that? Obviously, he's still relying on NVIDIA.

233
00:13:34,640 --> 00:13:37,760
Ejaaz:
He has a good partnership with them. But I just I'm extremely bullish about

234
00:13:37,760 --> 00:13:41,100
Ejaaz:
this, and I like that he's choosing to kind of partner with different firms

235
00:13:41,100 --> 00:13:42,940
Ejaaz:
versus trying to build everything himself.

236
00:13:43,120 --> 00:13:47,240
Ejaaz:
But moving on, there is another company that is raising a pretty decent valuation

237
00:13:47,240 --> 00:13:51,260
Ejaaz:
right now, Cognition, which is the, I believe they started Devin,

238
00:13:51,400 --> 00:13:54,760
Ejaaz:
yeah, two years ago, which was the OG AI agent.

239
00:13:55,320 --> 00:13:59,520
Ejaaz:
They raised over a billion dollars at $26 billion valuation.

240
00:13:59,520 --> 00:14:03,760
Ejaaz:
Now, the coolest part about these guys is everyone thought that they were completely

241
00:14:03,760 --> 00:14:09,060
Ejaaz:
cooked after Anthropic and OpenClaw kind of like launched their like autonomous agents.

242
00:14:09,260 --> 00:14:12,360
Ejaaz:
But what I liked about Cognition is they have a lot of staying power.

243
00:14:12,560 --> 00:14:16,360
Ejaaz:
And we've spoken about on previous episodes that there's this company called Cursor.

244
00:14:16,620 --> 00:14:19,600
Ejaaz:
And everyone blamed them for being like kind of like, or criticized them for

245
00:14:19,600 --> 00:14:21,760
Ejaaz:
being like a rapper, a thin rapper around an AI model.

246
00:14:21,880 --> 00:14:25,200
Ejaaz:
But what ended up being the case was they had this sort of agent harness,

247
00:14:25,380 --> 00:14:28,680
Ejaaz:
this agent mode, which kind of like orchestrates how models work.

248
00:14:28,990 --> 00:14:33,250
Ejaaz:
Uses the right models, uses the right prompts, and it's an incredibly valuable

249
00:14:33,250 --> 00:14:35,170
Ejaaz:
mode when it comes to training a model.

250
00:14:35,490 --> 00:14:40,690
Ejaaz:
Cognition is probably second in the ranking of having or owning this particular type of mode.

251
00:14:41,010 --> 00:14:43,890
Ejaaz:
And Josh, I know you wrote an entire essay about this recently.

252
00:14:44,270 --> 00:14:48,070
Josh:
I'm a big fan of harnesses. I really like harnesses. If you haven't subscribed

253
00:14:48,070 --> 00:14:49,410
Josh:
to a newsletter, I would highly recommend.

254
00:14:49,570 --> 00:14:52,470
Josh:
I released an entire essay about harnesses earlier this week.

255
00:14:52,510 --> 00:14:57,010
Josh:
It's basically all of the peripherals required to turn the brain of an LLM into

256
00:14:57,010 --> 00:15:01,190
Josh:
a full body, into something fully capable of doing agentic tasks that are long-running.

257
00:15:01,450 --> 00:15:04,870
Josh:
One thing I found funny from this announcement was the idea that Cognition uses

258
00:15:04,870 --> 00:15:06,910
Josh:
now 90% of their own code.

259
00:15:07,050 --> 00:15:10,850
Josh:
And if they're valued at a $26 billion valuation, clearly there's a lot of value

260
00:15:10,850 --> 00:15:13,510
Josh:
there. So this is a continuation of the wrapper wars.

261
00:15:13,690 --> 00:15:17,490
Josh:
We now kind of have very clear winners on who's winning the LLM wars,

262
00:15:17,650 --> 00:15:20,930
Josh:
the large language models that sit on the frontier, how you package them,

263
00:15:21,230 --> 00:15:23,370
Josh:
how you add a harness for them will be the differentiator.

264
00:15:23,450 --> 00:15:28,430
Josh:
We saw Cursor with a $60 billion deal with SpaceX. Now we have Cognition raising another big round.

265
00:15:28,850 --> 00:15:34,010
Josh:
The current war, that new frontier, is being fought on the wrapper level with the harnesses.

266
00:15:34,150 --> 00:15:37,550
Josh:
Cognition's the new eccentric in that. I'm pretty bullish, pretty excited on Cognition.

267
00:15:38,070 --> 00:15:42,770
Ejaaz:
Josh, on your bingo card for this week, did you expect the Pope to sign a partnership

268
00:15:42,770 --> 00:15:44,190
Ejaaz:
with the Major Frontier AI Lab?

269
00:15:44,510 --> 00:15:47,790
Josh:
A partnership? Tell me about this partnership. What do you mean partnership?

270
00:15:48,290 --> 00:15:53,790
Ejaaz:
Okay, so listen, we don't typically get into the swing of reviewing biblical

271
00:15:53,790 --> 00:15:56,590
Ejaaz:
architecture here on this show, but we are going to today.

272
00:15:56,930 --> 00:15:57,890
Josh:
Who got the mandate of heaven?

273
00:15:58,330 --> 00:16:03,930
Ejaaz:
Yeah, yeah, exactly. So Pope Leo kind of blew the newsroom wide open this week,

274
00:16:04,050 --> 00:16:08,990
Ejaaz:
releasing his Magnifica Humanitas of His Holiness, Pope Leo,

275
00:16:09,330 --> 00:16:12,770
Ejaaz:
on safeguarding the human person in the time of artificial intelligence.

276
00:16:12,930 --> 00:16:15,070
Ejaaz:
Now, my initial reaction to this is,

277
00:16:15,540 --> 00:16:19,620
Ejaaz:
I actually kind of love it because kind of like the religious pious leaders

278
00:16:19,620 --> 00:16:23,200
Ejaaz:
don't tend to kind of like technology a lot or get involved with it much.

279
00:16:23,540 --> 00:16:28,840
Ejaaz:
The fact that he's written, I believe this is like a 50 page plus script on

280
00:16:28,840 --> 00:16:30,660
Ejaaz:
his kind of like thoughts of how this plays out,

281
00:16:30,800 --> 00:16:37,100
Ejaaz:
how AI is going to evolve, how it kind of equates to like ethical use and religious

282
00:16:37,100 --> 00:16:39,080
Ejaaz:
scripts. He gets into a whole thing.

283
00:16:39,180 --> 00:16:41,820
Ejaaz:
Like there's this thing I was reading earlier on where it's like between two

284
00:16:41,820 --> 00:16:45,500
Ejaaz:
biblical images and talks about the construction of the Tower of Babel and these

285
00:16:45,500 --> 00:16:46,680
Ejaaz:
two scenes within the Bible.

286
00:16:46,820 --> 00:16:50,480
Ejaaz:
And the main takeaway that I got from this entire thing is he's trying to be

287
00:16:50,480 --> 00:16:54,120
Ejaaz:
extremely thoughtful and careful about how these models are developed because

288
00:16:54,120 --> 00:16:58,140
Ejaaz:
he knows about the power of influence that these models will have on kids,

289
00:16:58,420 --> 00:17:03,000
Ejaaz:
young adults, and everyone across every single age in the future.

290
00:17:03,000 --> 00:17:06,740
Ejaaz:
And so he knows that it's gonna guide a lot of education and learning about

291
00:17:06,740 --> 00:17:09,220
Ejaaz:
religious scripts and all these kinds of things. And so he wants to make sure

292
00:17:09,220 --> 00:17:10,460
Ejaaz:
that it's developed in the right way.

293
00:17:10,560 --> 00:17:13,940
Ejaaz:
Now, one of the biggest call to actions that he has out of this entire document

294
00:17:13,940 --> 00:17:17,660
Ejaaz:
and the preceding hearing or talk or discussion that they had after this was,

295
00:17:18,000 --> 00:17:21,200
Ejaaz:
He wants there to be governance instilled into this.

296
00:17:21,300 --> 00:17:24,260
Ejaaz:
Now, the governance, he suggests, comes from two particular parties,

297
00:17:24,400 --> 00:17:27,440
Ejaaz:
one being the national government that runs the entire country,

298
00:17:27,560 --> 00:17:29,700
Ejaaz:
respective country where it's operating in,

299
00:17:29,860 --> 00:17:34,780
Ejaaz:
the model it's operating in, as well as the church or religious groups involved

300
00:17:34,780 --> 00:17:37,580
Ejaaz:
so that they can kind of mold the personality and ethics of this.

301
00:17:37,700 --> 00:17:41,560
Ejaaz:
Now, when I kind of like speak about this, I think of one particular AI lab,

302
00:17:41,740 --> 00:17:45,020
Ejaaz:
which is Anthropic, who have been speaking to religious leaders independently

303
00:17:45,020 --> 00:17:48,540
Ejaaz:
of their own accord over the last, I believe, couple of months.

304
00:17:48,800 --> 00:17:52,200
Ejaaz:
And the Pope actually calls that Anthropic saying like, we would like to walk

305
00:17:52,200 --> 00:17:54,740
Ejaaz:
together. So I thought that was a pretty interesting update.

306
00:17:55,060 --> 00:17:59,840
Josh:
That's fun. It's nice when there's more awareness being raised around the idea.

307
00:18:00,140 --> 00:18:03,340
Josh:
It's like this will bring a lot of new eyes to the world of AI.

308
00:18:03,580 --> 00:18:06,300
Josh:
It will start a conversation that is going to need to happen.

309
00:18:06,300 --> 00:18:09,280
Josh:
It is going to bring awareness in a way that is constructive,

310
00:18:09,720 --> 00:18:13,040
Josh:
not damaging, hopefully. And I think it's just, it's a fun...

311
00:18:13,840 --> 00:18:16,780
Josh:
Cool thing and it's a testament to how large ai has

312
00:18:16,780 --> 00:18:19,480
Josh:
gotten where this is now sitting at a global scale and it is top of

313
00:18:19,480 --> 00:18:23,040
Josh:
mind and i like the idea that all of these people are going to be working together

314
00:18:23,040 --> 00:18:27,480
Josh:
to figure out the best way to navigate forward as a society and that that's

315
00:18:27,480 --> 00:18:31,780
Josh:
pretty exciting so glad i'm glad the pope's getting on board and uh hopefully

316
00:18:31,780 --> 00:18:35,440
Josh:
a lot of people who you know respect and trust him will take ai a little bit

317
00:18:35,440 --> 00:18:37,840
Josh:
more seriously and start considering this conversation and.

318
00:18:37,840 --> 00:18:40,820
Ejaaz:
Then next on the docket we have two more topics we're going to breeze through

319
00:18:40,820 --> 00:18:41,800
Josh:
It All right, bang them out.

320
00:18:41,980 --> 00:18:46,060
Ejaaz:
Josh, something that you and I complain about a lot is when a new model releases,

321
00:18:46,380 --> 00:18:47,900
Ejaaz:
you have these fancy benchmarks, right?

322
00:18:48,380 --> 00:18:52,880
Ejaaz:
But we don't know what that means. And often these benchmarks are maxed out

323
00:18:52,880 --> 00:18:57,660
Ejaaz:
or gamified and doesn't actually tell us or portray what the model can actually do.

324
00:18:57,780 --> 00:19:00,740
Ejaaz:
So let's say a model is super good at coding and then a software engineer actually

325
00:19:00,740 --> 00:19:04,900
Ejaaz:
uses it for their specific work. It's actually not good enough.

326
00:19:05,160 --> 00:19:08,940
Ejaaz:
Now, what a bunch of people ended up doing was they developed this new benchmark

327
00:19:08,940 --> 00:19:15,880
Ejaaz:
called DeepSWE, and it focuses on agentic AI engineering specifically.

328
00:19:16,200 --> 00:19:20,080
Ejaaz:
So when you spin up multiple instances of Claude or ChatGPT,

329
00:19:20,180 --> 00:19:24,480
Ejaaz:
for instance, to do a bunch of your coding work in parallel, how good is it actually?

330
00:19:24,640 --> 00:19:29,280
Ejaaz:
Now, the typical benchmark, and this is, by the way, like the holy Bible, to use a pun here,

331
00:19:29,780 --> 00:19:34,660
Ejaaz:
of the holy grail of benchmarks, which is software engineering verified,

332
00:19:34,940 --> 00:19:39,940
Ejaaz:
which is what every AI lab kind of like shows off when they release a new model.

333
00:19:40,580 --> 00:19:42,640
Ejaaz:
Currently, Claude and ChatGPT

334
00:19:43,250 --> 00:19:50,170
Ejaaz:
at the top of this, but it's tested against pre-approved solutions and answers

335
00:19:50,170 --> 00:19:52,970
Ejaaz:
against a particular problem. So the model already knows the answer.

336
00:19:53,210 --> 00:19:55,410
Ejaaz:
And so it's not really authentic or genuine.

337
00:19:55,670 --> 00:19:59,470
Ejaaz:
What this benchmark ends up doing is providing not only new problems of which

338
00:19:59,470 --> 00:20:02,350
Ejaaz:
the models have no idea what the answer is, they're not given the answers beforehand,

339
00:20:02,610 --> 00:20:06,550
Ejaaz:
but also real long horizon tasks that go for between 7 to 12 hours.

340
00:20:06,690 --> 00:20:09,710
Ejaaz:
And so they're put to the test. And we have a new ranking over here.

341
00:20:10,070 --> 00:20:16,570
Ejaaz:
GPT 5.5 and GPT 5.4, their predecessor model is number one and two of this new benchmark.

342
00:20:16,770 --> 00:20:21,310
Ejaaz:
You've got Claude Opus 4.7 just behind on 54% and Sonnet 4.6 as well.

343
00:20:21,590 --> 00:20:26,550
Ejaaz:
Gemini 3.5 Flash, even though it is their latest model, comes out at a pretty poor 28%.

344
00:20:26,550 --> 00:20:31,150
Josh:
Yikes. I do like the idea of more benchmarks. I think benchmarks are fun. They are important.

345
00:20:31,370 --> 00:20:34,490
Josh:
They are oftentimes games. So if we can come up with ways in which we could

346
00:20:34,490 --> 00:20:35,430
Josh:
organically test these,

347
00:20:35,590 --> 00:20:38,430
Josh:
if you ask me the way that i choose to

348
00:20:38,430 --> 00:20:41,190
Josh:
do my benchmarks it's purely vibes based i'll get

349
00:20:41,190 --> 00:20:44,010
Josh:
in i'll do the day-to-day task you could very much feel a model

350
00:20:44,010 --> 00:20:47,170
Josh:
if you use it every day you could feel where it's better feel where it's worse i find

351
00:20:47,170 --> 00:20:50,130
Josh:
that they're oftentimes spiky in the sense that they're very very

352
00:20:50,130 --> 00:20:53,310
Josh:
good at specific things and oftentimes a lot weaker than others and that

353
00:20:53,310 --> 00:20:56,350
Josh:
kind of spike complex changes over time as these

354
00:20:56,350 --> 00:20:59,790
Josh:
new models get released but i'm glad we have a more formal way

355
00:20:59,790 --> 00:21:02,430
Josh:
of doing this greater than vibes and greater than the

356
00:21:02,430 --> 00:21:05,530
Josh:
things that have been gamed recently now last thing on the agenda is

357
00:21:05,530 --> 00:21:11,590
Josh:
snowflake stock um the headline i'm seeing is snowflake stock jumps 30 percent

358
00:21:11,590 --> 00:21:16,350
Josh:
in a day this was after the earnings report that just came out this week and

359
00:21:16,350 --> 00:21:20,050
Josh:
for those that don't know snowflake is a cloud compute providing platform and

360
00:21:20,050 --> 00:21:24,130
Josh:
clearly they've done something right because uh wow this is a lot of revenue in one day.

361
00:21:24,730 --> 00:21:27,670
Ejaaz:
Yeah, so to give a bit of context here,

362
00:21:27,950 --> 00:21:31,830
Ejaaz:
there's something, there's a phenomenon happening called the SaaS-pocalypse,

363
00:21:32,010 --> 00:21:36,850
Ejaaz:
software-as-a-service apocalypse, where basically as Claude and ChatGPT end

364
00:21:36,850 --> 00:21:38,610
Ejaaz:
up upgrading their own models,

365
00:21:38,770 --> 00:21:44,530
Ejaaz:
they become able to replace a bunch of subscription software tools and services

366
00:21:44,530 --> 00:21:47,290
Ejaaz:
because they could just vibe code it or rebuild it from scratch.

367
00:21:47,490 --> 00:21:51,630
Ejaaz:
And this has led to a massive decline. I'm talking tens to hundreds of billions

368
00:21:51,630 --> 00:21:54,210
Ejaaz:
of dollars lost in the stock market for all these different SaaS products.

369
00:21:54,210 --> 00:21:57,530
Ejaaz:
I remember a time where Anthropic would release a plugin update for,

370
00:21:57,630 --> 00:22:02,130
Ejaaz:
let's say, a security review tool, and all the cybersecurity stocks were absolutely done, right?

371
00:22:02,430 --> 00:22:07,090
Ejaaz:
And so the point here is the trend has been kind of not only broken, but inverted.

372
00:22:07,430 --> 00:22:10,710
Ejaaz:
Snowflake, which is a major SaaS provider for a lot of enterprise companies,

373
00:22:11,130 --> 00:22:15,510
Ejaaz:
not only had their stock jump up 35%, but it's because they defined a moat that

374
00:22:15,510 --> 00:22:17,770
Ejaaz:
sat on top of Claude and ChatGPT.

375
00:22:17,770 --> 00:22:21,850
Ejaaz:
Now, in previous episodes, we've spoken about Anthropic and OpenAI signing what's

376
00:22:21,850 --> 00:22:24,730
Ejaaz:
known as a joint venture with some major private equity firms.

377
00:22:24,870 --> 00:22:28,810
Ejaaz:
And the intention of this is to place their own engineers, they're known as

378
00:22:28,810 --> 00:22:33,390
Ejaaz:
forward-deployed engineers, into Fortune 100 companies so that they can kind

379
00:22:33,390 --> 00:22:37,430
Ejaaz:
of like design and architect-specific workflows, so AI-agentic workflows,

380
00:22:37,690 --> 00:22:39,510
Ejaaz:
using Core or using ChatGPT.

381
00:22:39,630 --> 00:22:42,870
Ejaaz:
Now, the only reason why they would be doing that is because their own generalized

382
00:22:42,870 --> 00:22:47,230
Ejaaz:
LLM isn't good enough. And so you need someone to kind of like guide this kind of orchestration.

383
00:22:47,830 --> 00:22:51,450
Ejaaz:
Snowflake is basically arguing in their latest quarterly earnings that they already have that data.

384
00:22:51,710 --> 00:22:56,670
Ejaaz:
They already have that design and they're doing it themselves using a mixture of Claude and ChatGPT.

385
00:22:56,870 --> 00:23:00,570
Ejaaz:
And that's why the stock jumped up massively. It just shows that not only are

386
00:23:00,570 --> 00:23:04,630
Ejaaz:
SaaS companies here to stay, but they also might be having or creating a defensible moat.

387
00:23:04,970 --> 00:23:07,590
Josh:
I feel like we need a dartboard that we can start throwing darts at to pick

388
00:23:07,590 --> 00:23:10,130
Josh:
the companies that are going to go up 30% because every week it seems like there

389
00:23:10,130 --> 00:23:12,530
Josh:
is 10 new ones that are going absolutely nuclear.

390
00:23:12,770 --> 00:23:15,510
Josh:
But that wraps up the episode today. Thank you all so much for watching.

391
00:23:15,510 --> 00:23:18,110
Josh:
If you have made it through all of our previous episodes this week,

392
00:23:18,230 --> 00:23:19,230
Josh:
you are now fully caught up.

393
00:23:19,370 --> 00:23:21,930
Josh:
You can go touch grass, enjoy your weekend, have an amazing time.

394
00:23:21,930 --> 00:23:25,750
Josh:
And we will be back as always with a new episode starting bright and early next week.

395
00:23:25,930 --> 00:23:28,470
Josh:
Thank you guys so much for watching. Have an amazing weekend and we'll see you next time.