1
00:00:00,100 --> 00:00:01,966
welcome to money is freedom

2
00:00:01,966 --> 00:00:02,866
a podcast

3
00:00:02,866 --> 00:00:06,066
exploring how finance and freedom connect in our lives

4
00:00:07,033 --> 00:00:08,833
created from thoughtful research

5
00:00:08,833 --> 00:00:11,300
and narrated by Notebook lmaai

6
00:00:11,400 --> 00:00:13,266
this series brings you clear

7
00:00:13,266 --> 00:00:15,966
meaningful insights into finance and beyond

8
00:00:16,166 --> 00:00:18,166
welcome to episode 52

9
00:00:18,900 --> 00:00:22,500
Today we dive into AI's ongoing valuation correction

10
00:00:22,566 --> 00:00:23,800
the margin trap

11
00:00:24,400 --> 00:00:26,200
AI valuations are soaring

12
00:00:26,200 --> 00:00:28,866
at approximately 40.6 times revenue

13
00:00:29,066 --> 00:00:32,333
yet margins lag 20 to 25 points below SAS

14
00:00:33,000 --> 00:00:36,200
this isn't hype it's a structural economics problem

15
00:00:36,200 --> 00:00:38,466
where unsustainable growth will hit a wall

16
00:00:38,866 --> 00:00:40,233
welcome to the Deep Dive

17
00:00:40,233 --> 00:00:41,366
where we cut through the noise

18
00:00:41,366 --> 00:00:43,033
to deliver those critical insights

19
00:00:43,033 --> 00:00:45,066
you need to truly stay informed

20
00:00:45,200 --> 00:00:47,433
today we're diving deep into a topic that's well

21
00:00:47,433 --> 00:00:49,600
it's everywhere artificial intelligence

22
00:00:49,800 --> 00:00:51,966
groundbreaking stuff transformative even

23
00:00:51,966 --> 00:00:54,066
but is the financial foundation

24
00:00:54,066 --> 00:00:56,533
actually as solid as all the hype suggests

25
00:00:56,700 --> 00:00:59,033
we're looking closely at a really insightful article

26
00:00:59,033 --> 00:01:01,300
it's titled The AI Margin Trap

27
00:01:01,300 --> 00:01:02,833
why billion Dollar valuations

28
00:01:02,833 --> 00:01:05,100
are built on 50 Cent Economics

29
00:01:05,100 --> 00:01:06,200
by Danny the VC

30
00:01:06,200 --> 00:01:08,666
published August 26th, 2025

31
00:01:08,666 --> 00:01:10,866
and our mission here is really to peel back the layers

32
00:01:10,866 --> 00:01:13,000
you know get past the buzzwords

33
00:01:13,033 --> 00:01:15,300
and understand the actual economics driving this

34
00:01:15,300 --> 00:01:17,133
exactly and this article

35
00:01:17,233 --> 00:01:18,000
it just

36
00:01:18,000 --> 00:01:20,400
nails this fundamental disconnect that's brewing

37
00:01:20,400 --> 00:01:22,466
under the surface it's a huge tension point

38
00:01:22,466 --> 00:01:23,100
you've got AI

39
00:01:23,100 --> 00:01:26,433
companies valued like pure software businesses right

40
00:01:26,433 --> 00:01:28,400
sky high expectations effortless scale

41
00:01:28,400 --> 00:01:29,500
huge profits

42
00:01:29,500 --> 00:01:31,000
but then you look at how they actually operate

43
00:01:31,000 --> 00:01:33,766
their costs and it looks much more like uh

44
00:01:34,166 --> 00:01:35,200
manufacturing yeah

45
00:01:35,200 --> 00:01:36,200
capital intensive it's

46
00:01:36,200 --> 00:01:37,433
this hidden conflict

47
00:01:37,433 --> 00:01:39,066
that's really critical to understand

48
00:01:39,066 --> 00:01:41,866
okay so you see all this money pouring in right

49
00:01:42,066 --> 00:01:43,666
the article mentions what

50
00:01:43,666 --> 00:01:47,766
a staggering $104 billion in venture funding

51
00:01:47,966 --> 00:01:50,100
just in the first half of 2025

52
00:01:50,100 --> 00:01:51,566
feels unstoppable

53
00:01:52,133 --> 00:01:53,800
but are we starting to see

54
00:01:53,800 --> 00:01:55,966
maybe some cracks in that foundation

55
00:01:55,966 --> 00:01:56,966
is this AI

56
00:01:56,966 --> 00:01:59,466
gold rush heading for a bit of a reality check

57
00:01:59,466 --> 00:02:00,400
oh absolutely

58
00:02:00,433 --> 00:02:01,566
the signs are definitely there

59
00:02:01,566 --> 00:02:03,700
you're seeing AI equities drop

60
00:02:04,000 --> 00:02:05,966
maybe 15% in recent months

61
00:02:05,966 --> 00:02:08,466
and that's bringing uh unprecedented scrutiny

62
00:02:08,466 --> 00:02:10,400
onto the business models themselves

63
00:02:10,966 --> 00:02:12,166
veteran investor David Sax

64
00:02:12,166 --> 00:02:14,066
he put it pretty bluntly he warned about founders

65
00:02:14,066 --> 00:02:16,500
turning down multi billion dollar offers

66
00:02:16,600 --> 00:02:19,000
for companies that haven't even launched a product yet

67
00:02:19,000 --> 00:02:21,300
he said basically these offers don't grow on trees

68
00:02:21,300 --> 00:02:22,166
they just don't

69
00:02:22,166 --> 00:02:25,100
he even mentioned Meyer and Ilya suggesting maybe uh

70
00:02:25,100 --> 00:02:26,600
maybe they should have taken those deals

71
00:02:26,600 --> 00:02:28,033
that's quite a warning from sax

72
00:02:28,033 --> 00:02:29,500
and it really hits that core problem

73
00:02:29,500 --> 00:02:32,500
he pointed out AI companies valued like software

74
00:02:32,500 --> 00:02:34,266
operating like manufacturing

75
00:02:34,433 --> 00:02:35,800
so can you give us a sense of

76
00:02:35,800 --> 00:02:37,633
how big this disconnect really is

77
00:02:37,633 --> 00:02:38,900
like with some numbers

78
00:02:38,900 --> 00:02:40,966
how extreme is this valuation gap

79
00:02:40,966 --> 00:02:42,666
it is astonishing honestly

80
00:02:42,666 --> 00:02:45,700
the AI premium is just off the charts

81
00:02:45,833 --> 00:02:47,666
these startups are trading at get this

82
00:02:47,666 --> 00:02:51,400
40.6 times revenue 40.6 x

83
00:02:51,600 --> 00:02:53,000
now put that in perspective

84
00:02:53,000 --> 00:02:54,200
look at other hot sectors

85
00:02:54,200 --> 00:02:56,466
fintech maybe 20.4 x

86
00:02:56,466 --> 00:03:01,866
payments 17.7 x proptech 16.3 x and traditional saws

87
00:03:01,866 --> 00:03:04,200
you know the usual gold standard for software that's

88
00:03:04,200 --> 00:03:07,666
baseline 8.3 x so AI at 40.6 x

89
00:03:07,666 --> 00:03:10,233
that's a 390% premium over sauce

90
00:03:10,233 --> 00:03:12,566
it's just it's the most extreme valuation premium

91
00:03:12,566 --> 00:03:14,066
we've ever seen in tech history

92
00:03:14,200 --> 00:03:15,833
even higher than the dot com bubble

93
00:03:15,833 --> 00:03:19,166
those multiples rarely went above maybe 20

94
00:03:19,166 --> 00:03:20,766
thirty x this is something else wow

95
00:03:20,766 --> 00:03:23,200
40.6 x that's hard to wrap your head around

96
00:03:23,200 --> 00:03:24,900
so the market isn't just optimistic

97
00:03:24,900 --> 00:03:27,466
it's basically betting that AI will be more profitable

98
00:03:27,466 --> 00:03:29,900
more efficient than any other tech sector ever

99
00:03:29,900 --> 00:03:31,366
that seems like an incredibly high bar

100
00:03:31,366 --> 00:03:33,466
doesn't it it's an extraordinary bet yeah

101
00:03:33,466 --> 00:03:34,300
essentially saying AI

102
00:03:34,300 --> 00:03:36,033
economics will beat everything else we've ever seen

103
00:03:36,033 --> 00:03:36,700
in software it's well

104
00:03:36,700 --> 00:03:38,200
it's built on hope frankly right

105
00:03:38,200 --> 00:03:39,600
but are they actually earning that hope

106
00:03:39,600 --> 00:03:41,800
what do the real numbers the actual profit margins

107
00:03:41,800 --> 00:03:43,866
tell us about whether that Assumption holds water

108
00:03:43,866 --> 00:03:45,366
yeah and that's where the picture gets really

109
00:03:45,366 --> 00:03:48,066
really stark the margin data directly contradicts

110
00:03:48,066 --> 00:03:49,466
those sky high valuations

111
00:03:49,900 --> 00:03:50,700
investors are paying

112
00:03:50,700 --> 00:03:53,900
like five times more for AI startups than SOS right

113
00:03:54,000 --> 00:03:55,400
but look at the gross margins

114
00:03:55,500 --> 00:03:58,166
traditional SOS averages what 77%

115
00:03:58,233 --> 00:04:00,700
you've got Gitlab way up at 91%

116
00:04:00,800 --> 00:04:04,066
Adobe 88% Salesforce 76%

117
00:04:04,200 --> 00:04:06,233
pretty healthy now look at AI

118
00:04:06,233 --> 00:04:07,900
the leaders like anthropic

119
00:04:07,900 --> 00:04:09,100
maybe 50 55%

120
00:04:09,100 --> 00:04:12,366
open AI around 50% perplexity maybe 60%

121
00:04:12,366 --> 00:04:14,100
even the best are operating 20

122
00:04:14,100 --> 00:04:16,300
25 points below the sauce average

123
00:04:16,366 --> 00:04:18,200
and for others gets worse

124
00:04:18,200 --> 00:04:20,833
stack blitz 40% lovable 35%

125
00:04:20,833 --> 00:04:23,266
replitz down to 23% and this isn't just

126
00:04:23,266 --> 00:04:23,633
you know oh

127
00:04:23,633 --> 00:04:24,200
they're young they'll

128
00:04:24,200 --> 00:04:26,300
scale out of it it seems structural

129
00:04:26,433 --> 00:04:29,000
it's driven by those variable inference costs

130
00:04:29,066 --> 00:04:30,500
every time someone uses the AI

131
00:04:30,500 --> 00:04:31,833
it costs real money for compute

132
00:04:31,833 --> 00:04:34,000
that difference is just massive

133
00:04:34,266 --> 00:04:35,600
but couldn't you argue I mean

134
00:04:35,600 --> 00:04:37,466
people do argue these are just early days

135
00:04:37,500 --> 00:04:39,833
scaling costs want the margins naturally improve

136
00:04:39,833 --> 00:04:41,633
as these companies grow get more efficient

137
00:04:41,633 --> 00:04:43,300
maybe closer to those sauce numbers

138
00:04:43,300 --> 00:04:44,266
that's the hope right

139
00:04:44,266 --> 00:04:46,166
that's definitely what investors are banking on

140
00:04:46,200 --> 00:04:48,100
but the article really pushes back on that

141
00:04:48,100 --> 00:04:50,300
it argues the nature of AI inference

142
00:04:50,300 --> 00:04:52,966
that cost is for every single user query

143
00:04:53,033 --> 00:04:55,533
creates this ongoing high variable cost

144
00:04:55,766 --> 00:04:57,466
software usually doesn't have that

145
00:04:57,466 --> 00:04:58,766
not to the same extent

146
00:04:59,100 --> 00:05:01,866
sauce is more about high upfront costs

147
00:05:02,000 --> 00:05:04,100
than near zero cost per user

148
00:05:04,200 --> 00:05:06,866
AI flips that so getting those margins up is a much

149
00:05:06,866 --> 00:05:09,000
much tougher climb than just waiting for scales

150
00:05:09,000 --> 00:05:10,233
not automatic okay

151
00:05:10,233 --> 00:05:11,566
so these thin margins

152
00:05:11,566 --> 00:05:13,200
they're not just an accounting issue

153
00:05:13,200 --> 00:05:14,800
they create real fragility

154
00:05:14,800 --> 00:05:16,466
real operational risks

155
00:05:16,633 --> 00:05:18,900
does the article give an example of where this uh

156
00:05:18,900 --> 00:05:21,566
margin fragility really caused problems for a company

157
00:05:21,566 --> 00:05:22,600
oh definitely

158
00:05:22,766 --> 00:05:25,600
the cursor case study is that was pretty dramatic

159
00:05:25,600 --> 00:05:27,633
cursor was an AI powered code editor

160
00:05:27,633 --> 00:05:28,833
they ran into huge trouble

161
00:05:28,833 --> 00:05:29,866
because essentially

162
00:05:29,866 --> 00:05:32,166
their flat pricing hit negative unit economics

163
00:05:32,166 --> 00:05:34,066
they were losing money on heavy users

164
00:05:34,300 --> 00:05:37,400
then they switched to this opaque token based billing

165
00:05:37,800 --> 00:05:40,400
and some users suddenly got hit with bills for like

166
00:05:40,433 --> 00:05:42,266
$7,000 a day yeah

167
00:05:42,266 --> 00:05:44,566
huge shock and then to make it worse

168
00:05:44,566 --> 00:05:46,900
when people complained their own AI support

169
00:05:46,900 --> 00:05:49,700
bots started making up fake restricted policies

170
00:05:49,700 --> 00:05:51,966
yeah total mess result

171
00:05:52,000 --> 00:05:53,466
mass customer exodus

172
00:05:53,900 --> 00:05:56,000
and because anthropic relied on partners

173
00:05:56,000 --> 00:05:57,700
like cursor for distribution

174
00:05:57,800 --> 00:06:00,566
it actually threatened Anthropic's revenue stream too

175
00:06:00,600 --> 00:06:03,200
yeah it just shows when your margins are that thin

176
00:06:03,200 --> 00:06:05,166
there's zero room for error

177
00:06:05,433 --> 00:06:08,700
no buffer for mistakes or unexpected usage spikes

178
00:06:08,700 --> 00:06:10,400
it's incredibly fragile

179
00:06:10,566 --> 00:06:12,233
that's a complete nightmare scenario

180
00:06:12,233 --> 00:06:13,800
and it brings us back to the money

181
00:06:13,833 --> 00:06:15,900
this level of capital intensity

182
00:06:16,100 --> 00:06:17,200
it really makes you wonder if

183
00:06:17,200 --> 00:06:18,866
investors have fully grasped it

184
00:06:18,866 --> 00:06:21,166
are they funding based on this new reality

185
00:06:21,166 --> 00:06:23,633
or still using the old sauce playbook

186
00:06:23,633 --> 00:06:25,266
what does this margin profile mean

187
00:06:25,266 --> 00:06:26,900
for how much cash these companies need

188
00:06:26,900 --> 00:06:28,233
just to survive and grow

189
00:06:28,233 --> 00:06:30,200
it means they need a ton more capital right

190
00:06:30,200 --> 00:06:32,000
way more than a similar sauce company

191
00:06:32,200 --> 00:06:32,600
David

192
00:06:32,600 --> 00:06:35,433
Sax's framework on capital efficiency really drives

193
00:06:35,433 --> 00:06:36,966
this home it's eye opening

194
00:06:37,100 --> 00:06:40,100
say you want to scale to $150 million in ARR

195
00:06:40,466 --> 00:06:42,866
an 80% gross margin sauce company

196
00:06:42,866 --> 00:06:44,466
it probably hits cash flow positive

197
00:06:44,466 --> 00:06:47,300
maybe even generates around $27 million by then

198
00:06:47,433 --> 00:06:49,666
now take a 50% gross margin AI company

199
00:06:49,666 --> 00:06:51,266
aiming for that same hundred fifty mil

200
00:06:51,266 --> 00:06:53,300
a R R it burns through an extra

201
00:06:53,300 --> 00:06:54,966
80 to 100 million dollars in capital

202
00:06:54,966 --> 00:06:57,300
just to get there 80 to 100 million more

203
00:06:57,300 --> 00:07:00,933
yeah and if you're a 23% margin company like Replit

204
00:07:00,966 --> 00:07:01,766
you're burning over

205
00:07:01,766 --> 00:07:03,366
a hundred and sixty million dollars

206
00:07:03,366 --> 00:07:06,533
more than the sauce company that difference

207
00:07:06,866 --> 00:07:07,366
we're talking

208
00:07:07,366 --> 00:07:10,100
over $130 million in additional funding needed

209
00:07:10,100 --> 00:07:11,100
just because of the margins

210
00:07:11,100 --> 00:07:13,366
it's a fundamentally different capital game goodness

211
00:07:13,366 --> 00:07:15,666
with that kind of cash burn and those low margins

212
00:07:16,366 --> 00:07:21,100
the path to an IPO to actual sustainable profitability

213
00:07:21,100 --> 00:07:22,633
must look incredibly steep

214
00:07:22,633 --> 00:07:23,700
it's not just a long road

215
00:07:23,700 --> 00:07:25,033
it sounds like climbing a cliff face

216
00:07:25,033 --> 00:07:26,666
it is incredibly steep yeah

217
00:07:26,666 --> 00:07:27,566
you know public markets

218
00:07:27,566 --> 00:07:29,033
they generally want to see what 10

219
00:07:29,033 --> 00:07:30,900
15% operating margins uh huh

220
00:07:30,900 --> 00:07:32,766
so if the AI leaders are starting at 50

221
00:07:32,766 --> 00:07:35,066
60% gross margins they need to find

222
00:07:35,066 --> 00:07:36,600
another 35 to 45 percentage

223
00:07:36,600 --> 00:07:38,133
points of operational leverage

224
00:07:38,366 --> 00:07:39,633
and what that means you know

225
00:07:39,633 --> 00:07:40,233
for listeners

226
00:07:40,233 --> 00:07:42,600
is they have to drastically cut their other costs

227
00:07:42,600 --> 00:07:44,100
R&D sales

228
00:07:44,100 --> 00:07:46,066
admin relative to revenue as they scale

229
00:07:46,066 --> 00:07:47,600
that's tough it implies like

230
00:07:47,600 --> 00:07:50,166
three to five years of absolutely perfect execution

231
00:07:50,200 --> 00:07:52,400
with basically no major competition popping up

232
00:07:52,400 --> 00:07:54,500
that's a very big ask in this market

233
00:07:54,500 --> 00:07:56,300
okay so the climb is already steep

234
00:07:56,300 --> 00:07:58,100
but then the article mentions this

235
00:07:58,233 --> 00:08:01,200
accounting gamesmanship what's going on there

236
00:08:01,200 --> 00:08:02,600
how could that be making the picture

237
00:08:02,600 --> 00:08:04,600
look even rosier than it actually is

238
00:08:04,600 --> 00:08:05,366
ah yeah

239
00:08:05,366 --> 00:08:07,200
this is a really crucial detail

240
00:08:07,466 --> 00:08:09,000
turns out many AI

241
00:08:09,000 --> 00:08:10,033
startups are not

242
00:08:10,033 --> 00:08:12,066
including the cost of serving their free users

243
00:08:12,066 --> 00:08:13,633
in their main gross margin calculations

244
00:08:13,633 --> 00:08:15,166
really they just leave that out

245
00:08:15,166 --> 00:08:16,100
seems so

246
00:08:16,300 --> 00:08:19,366
Openai reportedly does include those free tier costs

247
00:08:19,566 --> 00:08:20,600
and that's part of why

248
00:08:20,600 --> 00:08:22,600
their reported margin is around 50%

249
00:08:22,600 --> 00:08:23,633
which the article suggests

250
00:08:23,633 --> 00:08:26,933
is probably closer to the true economic reality

251
00:08:27,200 --> 00:08:28,566
but many competitors

252
00:08:28,566 --> 00:08:31,500
by excluding those costs can show higher margins

253
00:08:31,500 --> 00:08:34,366
maybe closer to 60% it looks better on paper right

254
00:08:34,400 --> 00:08:35,433
but the article argues look

255
00:08:35,433 --> 00:08:36,700
if you factor in the compute costs

256
00:08:36,700 --> 00:08:37,966
for all those free users

257
00:08:38,100 --> 00:08:40,633
most AI companies are probably operating closer to 40

258
00:08:40,633 --> 00:08:42,600
50% gross margins in reality

259
00:08:42,700 --> 00:08:44,533
and that given the valuations

260
00:08:44,700 --> 00:08:46,966
is well into unsustainable territory

261
00:08:46,966 --> 00:08:49,100
wow so let me make sure I understand this for

262
00:08:49,100 --> 00:08:50,133
for you listening

263
00:08:50,233 --> 00:08:53,066
companies might be getting valued based on numbers that

264
00:08:53,166 --> 00:08:55,533
well that don't fully capture the real costs

265
00:08:55,833 --> 00:08:59,266
which makes that already steep climb to profitability

266
00:08:59,266 --> 00:09:00,266
even harder

267
00:09:00,266 --> 00:09:03,066
pushes that need for perfect execution even further out

268
00:09:03,066 --> 00:09:04,666
requires even more capital

269
00:09:04,866 --> 00:09:05,700
it feels like

270
00:09:05,700 --> 00:09:07,666
investors might be chasing a bit of a mirage here

271
00:09:07,666 --> 00:09:08,666
that's a good way to put it

272
00:09:08,666 --> 00:09:13,100
it definitely masks the true capital intensity and the

273
00:09:13,100 --> 00:09:15,500
uh the underlying economic challenge

274
00:09:15,500 --> 00:09:18,466
yeah puts a massive bet on future efficiency gains

275
00:09:18,466 --> 00:09:19,966
that just haven't materialized yet

276
00:09:19,966 --> 00:09:21,300
okay so the article basically says

277
00:09:21,300 --> 00:09:22,500
the correction isn't on its way

278
00:09:22,500 --> 00:09:23,666
it's kind of already happening

279
00:09:23,666 --> 00:09:26,166
market forces economic reality setting in

280
00:09:26,400 --> 00:09:27,966
so as things shake out

281
00:09:27,966 --> 00:09:30,233
what does the article say companies need to do

282
00:09:30,233 --> 00:09:32,400
what are the strategies for survival here right

283
00:09:32,400 --> 00:09:33,666
at least of the survival framework

284
00:09:33,666 --> 00:09:36,933
three main paths for it first is margin optimization

285
00:09:37,200 --> 00:09:40,466
the efficiency play this means things like uh

286
00:09:40,466 --> 00:09:41,666
deep inference optimization

287
00:09:41,666 --> 00:09:43,300
quantization distillation

288
00:09:43,300 --> 00:09:45,666
making models run leaner faster cheaper

289
00:09:45,866 --> 00:09:48,900
also developing custom chips like Google did with TPUs

290
00:09:48,966 --> 00:09:50,500
or vertical integration

291
00:09:50,600 --> 00:09:52,833
controlling more of the stack to keep costs down

292
00:09:52,833 --> 00:09:54,233
okay efficiency makes sense

293
00:09:54,233 --> 00:09:55,233
what's the second path

294
00:09:55,233 --> 00:09:57,866
second is business model evolution

295
00:09:58,000 --> 00:09:59,400
the platform play

296
00:09:59,600 --> 00:10:02,266
this is about shifting away from just selling compute

297
00:10:02,266 --> 00:10:03,866
heavy API calls

298
00:10:03,866 --> 00:10:06,233
maybe becoming more of a platform orchestrator

299
00:10:06,233 --> 00:10:07,533
building marketplaces

300
00:10:07,600 --> 00:10:09,866
or focusing on high value enterprise solutions

301
00:10:09,866 --> 00:10:11,033
where you can charge more

302
00:10:11,033 --> 00:10:12,433
create stickiness and hopefully

303
00:10:12,433 --> 00:10:14,900
get better margins than just per token billing

304
00:10:15,100 --> 00:10:16,200
change the game you're playing

305
00:10:16,200 --> 00:10:17,433
got it and the third

306
00:10:17,433 --> 00:10:21,266
third is scale economics the infrastructure play

307
00:10:21,400 --> 00:10:22,900
this is for the giants really

308
00:10:22,900 --> 00:10:25,300
it involves negotiating super preferential

309
00:10:25,300 --> 00:10:26,966
deals with cloud providers

310
00:10:27,066 --> 00:10:29,066
cause you're buying compute at massive scale

311
00:10:29,066 --> 00:10:31,533
or even building your own proprietary infrastructure

312
00:10:31,633 --> 00:10:33,233
basically getting so big

313
00:10:33,233 --> 00:10:35,666
that you can fundamentally change your cost structure

314
00:10:35,666 --> 00:10:36,966
through sheer volume

315
00:10:36,966 --> 00:10:39,166
and maybe justify that custom hardware investment

316
00:10:39,166 --> 00:10:42,166
efficiency new business models or massive scale

317
00:10:42,166 --> 00:10:45,000
OK so looking at those paths hmm

318
00:10:45,366 --> 00:10:47,500
it does feel like we've seen similar patterns

319
00:10:47,500 --> 00:10:48,566
in tech before right

320
00:10:48,566 --> 00:10:50,966
given these strategies who does the article suggest

321
00:10:50,966 --> 00:10:53,100
is actually well positioned to navigate this

322
00:10:53,100 --> 00:10:55,000
and who's really at risk in the coming shakeout

323
00:10:55,000 --> 00:10:56,966
sounds like clear winners and losers are emerging

324
00:10:56,966 --> 00:10:59,000
absolutely the article is pretty clear here

325
00:10:59,066 --> 00:11:01,366
position for success they point to opening

326
00:11:01,466 --> 00:11:03,266
I why well

327
00:11:03,266 --> 00:11:05,233
diversified revenue consumer subs

328
00:11:05,300 --> 00:11:06,666
enterprise API

329
00:11:06,666 --> 00:11:09,566
plus those huge Microsoft subsidies acting as a buffer

330
00:11:09,566 --> 00:11:10,966
and infrastructure advantage

331
00:11:10,966 --> 00:11:13,733
anthropic is also mentioned as potentially well

332
00:11:14,166 --> 00:11:16,066
if they maintain their focus on enterprise

333
00:11:16,066 --> 00:11:18,033
and strong margins and crucially

334
00:11:18,033 --> 00:11:20,666
diversify away from just being an API provider

335
00:11:20,700 --> 00:11:23,166
and you know generally any company that has a clear

336
00:11:23,333 --> 00:11:25,500
believable path to getting those gross margins

337
00:11:25,500 --> 00:11:26,700
above 70% OK

338
00:11:26,700 --> 00:11:29,300
so open AI Anthropic if they execute well

339
00:11:29,300 --> 00:11:31,166
and anyone with a path to high margins

340
00:11:31,166 --> 00:11:33,033
who's on the other side who's at risk

341
00:11:33,033 --> 00:11:35,100
at existential risk yeah

342
00:11:35,100 --> 00:11:37,966
basically companies stuck with sub 40% margins

343
00:11:38,033 --> 00:11:39,933
the article names Replit Stacklet

344
00:11:39,933 --> 00:11:42,433
lovable as examples facing potential funding cliffs

345
00:11:42,433 --> 00:11:43,966
as investors get more discerning

346
00:11:44,166 --> 00:11:45,633
also those pre product companies

347
00:11:45,633 --> 00:11:48,300
still turning down huge acquisition offers

348
00:11:48,300 --> 00:11:50,733
Sachs's warning seems very relevant here

349
00:11:50,733 --> 00:11:52,833
and really any business that's heavily relied on

350
00:11:52,833 --> 00:11:54,033
just selling API access

351
00:11:54,033 --> 00:11:55,833
without a clear plan for how to dramatically

352
00:11:55,833 --> 00:11:56,966
improve those margins

353
00:11:57,133 --> 00:11:59,166
they're stuck with those manufacturing like costs

354
00:11:59,166 --> 00:12:01,266
without the scale or strategy to escape

355
00:12:01,333 --> 00:12:01,700
yeah

356
00:12:01,700 --> 00:12:04,433
it really is a classic tech story playing out again

357
00:12:04,433 --> 00:12:06,266
isn't it growth at all costs

358
00:12:06,266 --> 00:12:08,733
eventually runs into the wall of unit economics

359
00:12:09,033 --> 00:12:10,033
what we've talked about today

360
00:12:10,033 --> 00:12:12,133
these 40 x revenue multiples

361
00:12:12,133 --> 00:12:15,633
for businesses struggling with 50% gross margins

362
00:12:15,833 --> 00:12:18,266
it really does feel like one of the most concentrated

363
00:12:18,266 --> 00:12:21,266
high risk bets we've seen in venture capital history

364
00:12:21,266 --> 00:12:23,200
it truly does look the AI revolution

365
00:12:23,633 --> 00:12:24,800
it's real it will

366
00:12:24,933 --> 00:12:26,066
form industries no doubt about that

367
00:12:26,066 --> 00:12:27,333
that's not the debate

368
00:12:27,433 --> 00:12:29,233
but the current valuation environment

369
00:12:29,233 --> 00:12:30,500
it's just not sustainable

370
00:12:30,500 --> 00:12:31,900
based on the underlying economics

371
00:12:31,900 --> 00:12:32,966
we're seeing right now

372
00:12:32,966 --> 00:12:35,100
so the critical question for you as an investor

373
00:12:35,100 --> 00:12:39,900
for founders is will today's high flying AI startups

374
00:12:39,900 --> 00:12:41,966
with these business models and these valuations

375
00:12:42,033 --> 00:12:43,933
actually be the long term winners

376
00:12:44,233 --> 00:12:45,633
or are they just paving the way

377
00:12:45,633 --> 00:12:47,500
burning cash to build the infrastructure

378
00:12:47,500 --> 00:12:48,433
for a different set of

379
00:12:48,433 --> 00:12:50,033
more economically sound companies

380
00:12:50,033 --> 00:12:52,000
to eventually capture the real value

381
00:12:52,000 --> 00:12:54,900
it's a huge question and that really leaves us

382
00:12:54,900 --> 00:12:58,366
and you with a provocative thought to chew on

383
00:12:58,433 --> 00:13:00,433
thinking about everything we've just discussed

384
00:13:00,433 --> 00:13:02,866
does taking that billion dollar offer for a pre

385
00:13:02,866 --> 00:13:03,700
product company

386
00:13:03,700 --> 00:13:05,800
suddenly sound like maybe the smartest move

387
00:13:05,800 --> 00:13:07,266
some founders aren't making

388
00:13:07,366 --> 00:13:09,266
what does this deep dive make you consider

389
00:13:09,266 --> 00:13:10,300
about where AI

390
00:13:10,300 --> 00:13:13,200
investment and innovation is really headed next