1
00:00:02,800 --> 00:00:07,000
Speaker 1: Zon Media Hell and welcome to a very special better

2
00:00:07,040 --> 00:00:18,520
Speaker 1: offline monologue. I'm your host ed Zron And. Wednesday, the

3
00:00:18,560 --> 00:00:21,239
Speaker 1: Wall Street Journal ran a story about how Anthropic is

4
00:00:21,280 --> 00:00:25,000
Speaker 1: about to have its first profitable quarter, specifically an operating

5
00:00:25,040 --> 00:00:28,600
Speaker 1: profit or a bit duh profitability so earnings before interests,

6
00:00:28,800 --> 00:00:32,280
Speaker 1: tax and bad stuff, really, saying that anthropics revenue was

7
00:00:32,320 --> 00:00:34,160
Speaker 1: said to more than double to ten point nine billion

8
00:00:34,159 --> 00:00:36,960
Speaker 1: dollars in the second quarter of twenty twenty six, which

9
00:00:37,000 --> 00:00:40,519
Speaker 1: is when said profit would happen. But the journal Anthropic

10
00:00:40,600 --> 00:00:43,040
Speaker 1: generated four point eight billion dollars in sales in the

11
00:00:43,040 --> 00:00:46,040
Speaker 1: first quarter, and in Q two, that's that five hundred

12
00:00:46,040 --> 00:00:51,960
Speaker 1: and fifty nine million dollar operating profits home interesting, there's

13
00:00:51,960 --> 00:00:54,760
Speaker 1: a lot of certainty, considering we're barely through the first

14
00:00:54,760 --> 00:00:57,800
Speaker 1: half of the second quarter, and quite a specific number

15
00:00:57,800 --> 00:01:01,400
Speaker 1: given the fact that June hasn't start, and all of

16
00:01:01,480 --> 00:01:05,840
Speaker 1: these numbers are mysteriously leaking exactly while Alanthropic is raising

17
00:01:05,840 --> 00:01:08,600
Speaker 1: its funding round. And there's another important note from the

18
00:01:08,680 --> 00:01:11,200
Speaker 1: story too. The journal adds at the very bottom of

19
00:01:11,240 --> 00:01:14,319
Speaker 1: the article that it and I quote is unclear what

20
00:01:14,440 --> 00:01:17,360
Speaker 1: accounting methods Anthropic has used to book revenue and costs.

21
00:01:17,520 --> 00:01:20,000
Speaker 1: As the company isn't yet required to follow the financial

22
00:01:20,040 --> 00:01:23,840
Speaker 1: reporting requirements of a public company. That's right, Anthropic is

23
00:01:23,920 --> 00:01:26,480
Speaker 1: possibly going to be a bit to profitable for a

24
00:01:26,520 --> 00:01:30,440
Speaker 1: single quarter on a non gap so generally is accepted

25
00:01:30,480 --> 00:01:34,040
Speaker 1: accounting principles basis. Anyway, I wonder how they did it,

26
00:01:34,280 --> 00:01:38,120
Speaker 1: because based on the unhelpfully labeled diagram from the Journal's article,

27
00:01:38,160 --> 00:01:40,960
Speaker 1: it appears, as I said last year, the Anthropics costs

28
00:01:40,959 --> 00:01:45,120
Speaker 1: scale linearly with its revenues, except they magically didn't in

29
00:01:45,160 --> 00:01:49,360
Speaker 1: the second quarter. How and I wonder if Anthropic will

30
00:01:49,480 --> 00:01:53,480
Speaker 1: somehow stay profitable. Let's see if the journal has a quote, Oh,

31
00:01:53,160 --> 00:01:56,600
Speaker 1: there we go, Okay, the company might not remain profitable

32
00:01:56,640 --> 00:01:59,320
Speaker 1: for the full year, as it plans spending increases due

33
00:01:59,320 --> 00:02:05,559
Speaker 1: to its fasting needs. Interesting. That's very interesting. So Anthropic

34
00:02:05,760 --> 00:02:09,320
Speaker 1: may be profitable very specifically in the second quarter of

35
00:02:09,320 --> 00:02:12,960
Speaker 1: twenty twenty six, but might not be afterwards. It's almost

36
00:02:13,000 --> 00:02:15,399
Speaker 1: as if it found a way to specifically cut its

37
00:02:15,400 --> 00:02:19,760
Speaker 1: costs in May and June somehow, because it did. Remember

38
00:02:19,800 --> 00:02:23,679
Speaker 1: that deal Anthropic sign with SpaceX to take over Colossus one, Well,

39
00:02:23,720 --> 00:02:27,120
Speaker 1: it's also extending to Colossus two, and they're going to

40
00:02:27,120 --> 00:02:30,240
Speaker 1: be paying SpaceX one point two five billion dollars a

41
00:02:30,240 --> 00:02:34,200
Speaker 1: month starting in May and June, and in those two

42
00:02:34,320 --> 00:02:38,440
Speaker 1: months they're going to have reduced fees. You know, the

43
00:02:38,560 --> 00:02:42,360
Speaker 1: quarter in which Anthropic is magically profitable, the one with

44
00:02:42,440 --> 00:02:46,679
Speaker 1: the lower costs ah per SpaceX is s one. Pursuant

45
00:02:46,680 --> 00:02:49,520
Speaker 1: to its agreement, Anthropic will pay SpaceX one point two

46
00:02:49,560 --> 00:02:52,239
Speaker 1: to five billion dollars a month through May twenty twenty nine,

47
00:02:52,240 --> 00:02:55,320
Speaker 1: with capacity ramping up in May and June at a

48
00:02:55,440 --> 00:02:59,560
Speaker 1: reduced fee as fifteen billion dollars a year in compute

49
00:02:59,560 --> 00:03:02,600
Speaker 1: cost but due to an indeterminately discounted level for the

50
00:03:02,639 --> 00:03:05,600
Speaker 1: precise months that Anthropic is using to tell its investors

51
00:03:05,639 --> 00:03:08,799
Speaker 1: in the media that it has an operating profit. Five

52
00:03:08,880 --> 00:03:11,200
Speaker 1: hundred and fifty nine million dollars in operating profit in

53
00:03:11,240 --> 00:03:13,800
Speaker 1: a three month period is absolutely possible when you're not

54
00:03:13,840 --> 00:03:17,280
Speaker 1: paying for all your costs. While I wouldn't say this

55
00:03:17,360 --> 00:03:19,600
Speaker 1: is cooking the books, I would say it's definitely a

56
00:03:19,639 --> 00:03:23,600
Speaker 1: Shiatsu grade massaging of the numbers. Anthropic is deliberately leaked

57
00:03:23,600 --> 00:03:26,240
Speaker 1: a quarterly profit, and I'm doing air quotes where it

58
00:03:26,280 --> 00:03:29,480
Speaker 1: knows it can suppress its costs. Specifically made sure that

59
00:03:29,520 --> 00:03:32,280
Speaker 1: the journalist gave it an out of costs might increase

60
00:03:32,560 --> 00:03:34,600
Speaker 1: and released it on the day of Nvidia's earnings as

61
00:03:34,600 --> 00:03:38,320
Speaker 1: a means of keeping the AI bubble inflated. Nothing has changed.

62
00:03:39,000 --> 00:03:41,200
Speaker 1: If Anthropic was paying the full rate for its compute

63
00:03:41,200 --> 00:03:44,120
Speaker 1: in those two months, its economics would shift right back

64
00:03:44,120 --> 00:03:46,320
Speaker 1: to where it's always been. Per my reporting from last

65
00:03:46,400 --> 00:03:49,040
Speaker 1: year on its aws costs. A business that has its

66
00:03:49,040 --> 00:03:53,800
Speaker 1: costs linearly increase with its revenues, I also severely doubt

67
00:03:53,800 --> 00:03:56,040
Speaker 1: that Anthropic managed to make the cost of running their

68
00:03:56,120 --> 00:03:59,840
Speaker 1: services profitable in the space of six months, per the information.

69
00:03:59,880 --> 00:04:03,000
Speaker 1: In January, Anthropic missed on its gross margin projections to

70
00:04:03,080 --> 00:04:05,680
Speaker 1: twenty twenty five, saying that its inference costs were twenty

71
00:04:05,720 --> 00:04:10,320
Speaker 1: three percent higher than the company had anticipated. How did Anthropic,

72
00:04:10,440 --> 00:04:12,640
Speaker 1: who faced a massive influx of new business to the

73
00:04:12,680 --> 00:04:14,640
Speaker 1: point that it was forced to buy more compute from

74
00:04:14,640 --> 00:04:18,640
Speaker 1: elon fucking Musk, suddenly have its revenue outpaced its costs.

75
00:04:19,560 --> 00:04:21,840
Speaker 1: Have had a few guesses other than the obvious way

76
00:04:21,880 --> 00:04:26,080
Speaker 1: in which they reduced their costs. Number one, for large enterprises,

77
00:04:26,080 --> 00:04:29,160
Speaker 1: they're taking prepayment of tokens, so fifty million dollars intended

78
00:04:29,200 --> 00:04:31,600
Speaker 1: to be spread over twelve months that it's taking us

79
00:04:31,680 --> 00:04:35,040
Speaker 1: upfront revenue. This would both inflate revenue numbers and depressed

80
00:04:35,040 --> 00:04:38,680
Speaker 1: costs because Anthropic wouldn't have actually provided the compute necessary

81
00:04:38,800 --> 00:04:41,480
Speaker 1: to earn that revenue. Yet perfectly legal, because this isn't

82
00:04:41,520 --> 00:04:44,360
Speaker 1: non gap. This is non gap accounting like you could

83
00:04:44,400 --> 00:04:47,280
Speaker 1: just you could say whatever in a real company or

84
00:04:47,279 --> 00:04:51,440
Speaker 1: a public company, that would be considered deferred revenue. Anthropic

85
00:04:51,520 --> 00:04:54,360
Speaker 1: is already offering discounted tokens for Claude users too, through

86
00:04:54,400 --> 00:04:56,800
Speaker 1: the by Extra Credits page on their accounts, with discounts

87
00:04:56,880 --> 00:05:00,400
Speaker 1: ranging from ten percent to thirty percent. It again, maybe

88
00:05:00,400 --> 00:05:03,680
Speaker 1: booking this upfront rather than saying, okay, let's book this

89
00:05:03,800 --> 00:05:06,599
Speaker 1: as the person uses the compute. They could also be

90
00:05:06,600 --> 00:05:09,400
Speaker 1: front loading annual commitments of basically any kind, subscriptions to

91
00:05:09,440 --> 00:05:12,480
Speaker 1: Claude enterprise team agreements and so on. And they could

92
00:05:12,520 --> 00:05:15,400
Speaker 1: have deliberately ratcheted downtraining to ease the burden on its

93
00:05:15,440 --> 00:05:20,880
Speaker 1: infrastructure to provide inference. But if I'm honest, the revenue

94
00:05:20,920 --> 00:05:24,800
Speaker 1: side is where the real problems lie. So Anthropic has

95
00:05:24,800 --> 00:05:27,200
Speaker 1: said it brought in four point eight billion dollars in

96
00:05:27,240 --> 00:05:29,240
Speaker 1: revenue in the first quarter of twenty twenty six and

97
00:05:29,240 --> 00:05:32,360
Speaker 1: projects to ten point nine billion dollars in Q two

98
00:05:32,480 --> 00:05:36,680
Speaker 1: twenty twenty six. This is pretty tough to reconcile with

99
00:05:36,760 --> 00:05:41,880
Speaker 1: previous reporting. On February twelve, twenty twenty six, Anthropic claimed

100
00:05:41,880 --> 00:05:44,760
Speaker 1: it had reached fourteen billion dollars in annual recurring revenue.

101
00:05:44,839 --> 00:05:47,520
Speaker 1: As a reminder, ARR is the accounting tool largely used

102
00:05:47,520 --> 00:05:50,120
Speaker 1: by startups but some public companies too, where a snapshot

103
00:05:50,160 --> 00:05:52,839
Speaker 1: of a single month's income is taken and multiplied by twelve.

104
00:05:53,360 --> 00:05:55,440
Speaker 1: This gives you an implied monthly revenue of like one

105
00:05:55,440 --> 00:05:59,160
Speaker 1: point one six one point one seven billion dollars. On

106
00:05:59,200 --> 00:06:01,719
Speaker 1: March third, twenty two, twenty six, Warrio Ama Day would

107
00:06:01,720 --> 00:06:05,080
Speaker 1: claim Anthropic had reached nineteen billion dollars an ARR, so

108
00:06:05,120 --> 00:06:07,160
Speaker 1: about one point five to eight billion dollars per month.

109
00:06:07,760 --> 00:06:11,240
Speaker 1: Three days later, on March sixth, Krishna Rao, chief financial

110
00:06:11,240 --> 00:06:14,280
Speaker 1: officer of Anthropic, would declare under oath in a court

111
00:06:14,320 --> 00:06:18,880
Speaker 1: filing that Anthropic had brought in revenues exceeding five billion

112
00:06:18,960 --> 00:06:23,640
Speaker 1: dollars to date, so just to be clear lifetime revenues. Also,

113
00:06:23,720 --> 00:06:26,080
Speaker 1: keep in mind that the Information had previously reported that

114
00:06:26,120 --> 00:06:28,640
Speaker 1: Anthropic had four point five billion dollars in revenue in

115
00:06:28,640 --> 00:06:31,480
Speaker 1: twenty twenty five, which already puts US at the limits

116
00:06:31,480 --> 00:06:35,800
Speaker 1: of credulity. While boosters may claim that exceeding could mean

117
00:06:35,880 --> 00:06:39,760
Speaker 1: literally any number they want above five billion dollars. I

118
00:06:39,800 --> 00:06:43,919
Speaker 1: found it doubtful that the CFO of Anthropic would, under oath,

119
00:06:44,240 --> 00:06:46,480
Speaker 1: lead the court to believe its business was thirty to

120
00:06:46,480 --> 00:06:49,279
Speaker 1: forty percent smaller than it was, especially when trying to

121
00:06:49,320 --> 00:06:51,719
Speaker 1: convince it that the damage of being labeled a supply

122
00:06:51,839 --> 00:06:54,000
Speaker 1: chain risk would ruin its business. They would want it

123
00:06:54,040 --> 00:06:57,320
Speaker 1: to seem bigger than it is, not smaller. And at

124
00:06:57,320 --> 00:06:59,760
Speaker 1: this point it's impossible to reconcile that twenty twenty five

125
00:07:00,160 --> 00:07:02,200
Speaker 1: point with that five billion dollar number. No one wants

126
00:07:02,240 --> 00:07:04,280
Speaker 1: to talk to me about it. None of the journalists involved,

127
00:07:04,320 --> 00:07:06,520
Speaker 1: no no analysts. That no one wants to talk about

128
00:07:06,520 --> 00:07:10,200
Speaker 1: this at all. It makes them uncomfortable and angry, I assume,

129
00:07:10,280 --> 00:07:23,840
Speaker 1: probably because it's impossible to reconcile. If we assume that

130
00:07:23,880 --> 00:07:26,880
Speaker 1: the arr claims made by Anthropica correct, we can presume

131
00:07:26,880 --> 00:07:29,840
Speaker 1: that it made revenues of roughly two point four to

132
00:07:29,840 --> 00:07:32,080
Speaker 1: two point five billion dollars in March, given that it

133
00:07:32,080 --> 00:07:34,600
Speaker 1: claimed it had thirty billion dollars of ARR and April

134
00:07:34,640 --> 00:07:36,680
Speaker 1: sixth if you get in confused with the numbers. By

135
00:07:36,680 --> 00:07:40,960
Speaker 1: the way, there's a newsletter attached to this. Nevertheless, when

136
00:07:41,000 --> 00:07:42,840
Speaker 1: you add up all those arrs, you get to about

137
00:07:42,840 --> 00:07:45,280
Speaker 1: five point two five billion dollars, which is in excess

138
00:07:45,320 --> 00:07:47,240
Speaker 1: of what the Wall Street Jurnal has said. And in

139
00:07:47,280 --> 00:07:50,560
Speaker 1: some world maybe there's a way of cherry picking using

140
00:07:50,600 --> 00:07:53,160
Speaker 1: particular periods to the point that the arrs makes sense

141
00:07:53,320 --> 00:07:55,080
Speaker 1: and would actually be in the region of four point

142
00:07:55,080 --> 00:07:58,560
Speaker 1: eight billion dollars. Fine, but they don't make a lick

143
00:07:58,560 --> 00:08:00,320
Speaker 1: of sense when you bring up what Christian to Row

144
00:08:00,400 --> 00:08:03,800
Speaker 1: their CFO said. If we believe anthropics leaks, putting inside

145
00:08:03,960 --> 00:08:06,760
Speaker 1: all the R figures for a second, this means that

146
00:08:06,800 --> 00:08:10,000
Speaker 1: Anthropic A made over ninety percent of its lifetime revenues

147
00:08:10,040 --> 00:08:12,920
Speaker 1: in the first quarter of twenty twenty six, and B

148
00:08:13,400 --> 00:08:16,520
Speaker 1: made virtually no revenue its previous years, and c leaked

149
00:08:16,560 --> 00:08:21,120
Speaker 1: completely imaginary run rates of the media for years. While

150
00:08:21,160 --> 00:08:24,680
Speaker 1: I acknowledge that anthropic abys to have grown significantly, that

151
00:08:24,760 --> 00:08:29,880
Speaker 1: level of stratospheric growth does stretch the limits of their credibility. Moreover,

152
00:08:29,920 --> 00:08:32,720
Speaker 1: the fact that the previous era. ARR figures are inconsistent with

153
00:08:32,760 --> 00:08:35,760
Speaker 1: the leaked charts from Anthropic. Further raises questions about the

154
00:08:35,760 --> 00:08:39,760
Speaker 1: credibility of well any numbers from the company. The only

155
00:08:39,880 --> 00:08:42,960
Speaker 1: real defense that anybody has here is that Krishna Rau

156
00:08:43,320 --> 00:08:45,959
Speaker 1: under oath low balled the US government and adjudged to

157
00:08:46,160 --> 00:08:49,600
Speaker 1: such a dramatic extent that he hid an excess of

158
00:08:49,679 --> 00:08:53,000
Speaker 1: four billion dollars in revenue. And as I've discussed before,

159
00:08:53,040 --> 00:08:55,960
Speaker 1: and Flying Penguin helpfully collated, which I'll link in the notes.

160
00:08:56,160 --> 00:08:59,440
Speaker 1: Adding up anthropics previously reported ARR from January twenty twenty

161
00:08:59,480 --> 00:09:01,880
Speaker 1: five to my twenty twenty six already gets us to

162
00:09:01,920 --> 00:09:05,320
Speaker 1: around six point six six billion dollars. Now, I know

163
00:09:05,520 --> 00:09:08,080
Speaker 1: there are going to be some boosters who hear about

164
00:09:08,080 --> 00:09:09,440
Speaker 1: this and they're going to be like, well, this is

165
00:09:09,480 --> 00:09:11,640
Speaker 1: proof there's a business model emerging from AI and I'm

166
00:09:12,400 --> 00:09:17,320
Speaker 1: I'm sorry, that's not what's happening. Dario Amadianelon Musk worked

167
00:09:17,320 --> 00:09:19,640
Speaker 1: out a sweetheart deal framed as a ramp up that

168
00:09:19,679 --> 00:09:22,920
Speaker 1: allowed Anthropic to artificially depress its costs. I also question

169
00:09:23,000 --> 00:09:24,880
Speaker 1: how much of a ramp up there really was, or

170
00:09:25,080 --> 00:09:30,000
Speaker 1: what Anthropic's actual compute constraints were because Anthropic immediately loosened

171
00:09:30,040 --> 00:09:32,959
Speaker 1: rate limits for Claude subscribers on announcing the deal, meaning

172
00:09:32,960 --> 00:09:37,840
Speaker 1: that it immediately started having higher inference costs, which somehow

173
00:09:37,960 --> 00:09:42,400
Speaker 1: led to it making a higher profit or did. Musk

174
00:09:42,480 --> 00:09:46,000
Speaker 1: has literally described in its s one have SpaceX charge

175
00:09:46,040 --> 00:09:49,280
Speaker 1: Anthropic less for two specific months to make the numbers

176
00:09:49,320 --> 00:09:53,800
Speaker 1: look better. In July, Anthropic will start paying SpaceX one

177
00:09:53,840 --> 00:09:56,480
Speaker 1: point twenty five billion dollars a month or about fifteen

178
00:09:56,480 --> 00:09:58,520
Speaker 1: billion dollars a year, on top of all its other

179
00:09:58,600 --> 00:10:02,000
Speaker 1: compute deals with Google, Amazon, the Microsoft. If we assume

180
00:10:02,000 --> 00:10:05,200
Speaker 1: that its spend is comparable on AWS and Google Cloud,

181
00:10:05,200 --> 00:10:08,480
Speaker 1: and it's most assuredly more, that means Anthropic is spending

182
00:10:08,520 --> 00:10:10,920
Speaker 1: around three point seventy five billion dollars a month in

183
00:10:11,000 --> 00:10:13,599
Speaker 1: compute costs or eleven point twenty five billion dollars a

184
00:10:13,679 --> 00:10:16,800
Speaker 1: quarter or forty five billion dollars a year. I'm one

185
00:10:16,840 --> 00:10:19,599
Speaker 1: hundred percent sure it's more than that. There's also a

186
00:10:19,720 --> 00:10:22,680
Speaker 1: very compelling argument that the Anthropics costs will increase and

187
00:10:22,720 --> 00:10:25,400
Speaker 1: will eat up that profitability. To once again repeat the

188
00:10:25,480 --> 00:10:28,280
Speaker 1: quote from the Journal, the company might not remain profitable

189
00:10:28,360 --> 00:10:31,120
Speaker 1: for the full year as it plans spending increases due

190
00:10:31,160 --> 00:10:36,120
Speaker 1: to its vast computing needs. Also have to wonder this

191
00:10:36,240 --> 00:10:39,400
Speaker 1: company's profitable. Just humoring here for a second. If you're

192
00:10:39,440 --> 00:10:42,160
Speaker 1: so profitable, why aren't you ipoing? Why not take this

193
00:10:42,200 --> 00:10:45,600
Speaker 1: to the public markets, unless, of course, your only non

194
00:10:45,679 --> 00:10:47,760
Speaker 1: gap a bit to profitable based on a two month

195
00:10:47,880 --> 00:10:50,720
Speaker 1: long discount specifically covering the period in which you're profitable.

196
00:10:51,679 --> 00:10:55,120
Speaker 1: I will give Diario Amida credit. Nobody does it better.

197
00:10:55,520 --> 00:10:59,720
Speaker 1: Nobody does financial engineering and press lead information war better

198
00:10:59,720 --> 00:11:04,160
Speaker 1: than our Anthropic. The utter willingness of the press to

199
00:11:04,160 --> 00:11:06,880
Speaker 1: eat up incongruent numbers and the eagerness of many to

200
00:11:06,920 --> 00:11:09,480
Speaker 1: jump up and find obtuse ways to explain away the

201
00:11:09,520 --> 00:11:12,640
Speaker 1: obvious problems is only made possible when a company's perfected

202
00:11:12,640 --> 00:11:15,360
Speaker 1: the art of manipulation and ingratiation of those who want

203
00:11:15,400 --> 00:11:18,360
Speaker 1: to feel like they're first. If you take this as

204
00:11:18,400 --> 00:11:21,839
Speaker 1: an incontrovertible proof that Anthropic is profitable, you are deliberately

205
00:11:21,880 --> 00:11:24,920
Speaker 1: ignoring the blatantly obvious ways these numbers are being massaged.

206
00:11:25,320 --> 00:11:27,920
Speaker 1: We've got its CFO saying numbers that don't match up

207
00:11:27,920 --> 00:11:30,560
Speaker 1: with any of their leaks or their own marketing materials.

208
00:11:30,640 --> 00:11:32,840
Speaker 1: In the aggressive and deluded way in which people ignore

209
00:11:32,880 --> 00:11:36,760
Speaker 1: them is equal parts frustrating and depressing. So I want

210
00:11:36,760 --> 00:11:39,400
Speaker 1: to speak to any AI boosters in the audience, and

211
00:11:39,440 --> 00:11:41,560
Speaker 1: I want to speak with a little more empathy than usual.

212
00:11:42,559 --> 00:11:45,520
Speaker 1: If you want Anthropic to win, you should be just

213
00:11:45,559 --> 00:11:48,360
Speaker 1: as skeptical of these numbers as I am. You should

214
00:11:48,360 --> 00:11:50,360
Speaker 1: want to smash my face in the tarmac with the

215
00:11:50,400 --> 00:11:53,760
Speaker 1: most crystal clear impossible to argue with numbers, but reft

216
00:11:53,760 --> 00:11:58,679
Speaker 1: of asterisks or discounts from suppliers or obfiscated accounting metrics.

217
00:11:59,240 --> 00:12:03,280
Speaker 1: You should want better from your heroes. If you truly

218
00:12:03,280 --> 00:12:05,880
Speaker 1: think this company is amazing, unstoppable and leading the tech

219
00:12:05,960 --> 00:12:09,199
Speaker 1: industry to a glorious era of innovation, there shouldn't be

220
00:12:09,280 --> 00:12:11,800
Speaker 1: this many questions and the metrics shouldn't be this murky.

221
00:12:12,559 --> 00:12:14,520
Speaker 1: Every other time when a company has played this level

222
00:12:14,520 --> 00:12:18,040
Speaker 1: of silly, weird bullshit, it's led to disaster. For example,

223
00:12:18,080 --> 00:12:20,280
Speaker 1: we Work claimed to be profitable since the second month

224
00:12:20,360 --> 00:12:23,760
Speaker 1: of its operations and repeated those claims of profitability throughout

225
00:12:23,760 --> 00:12:26,480
Speaker 1: its existence until it turned out it was only profitable

226
00:12:26,520 --> 00:12:28,960
Speaker 1: if you remove things like some of the costs of

227
00:12:29,040 --> 00:12:32,400
Speaker 1: doing business. And I get where you're so defensive, and

228
00:12:32,440 --> 00:12:34,080
Speaker 1: I get where you want this to work. A lot

229
00:12:34,080 --> 00:12:36,280
Speaker 1: of you are very excited about generative AI, and being

230
00:12:36,320 --> 00:12:38,440
Speaker 1: excited about it has given you a tremendous community of

231
00:12:38,480 --> 00:12:41,280
Speaker 1: equally excited people. I get that you like these tools,

232
00:12:42,200 --> 00:12:43,880
Speaker 1: and I need you to know that these companies are

233
00:12:43,920 --> 00:12:47,360
Speaker 1: laughing at you. Anthropic timed this lead to focus on

234
00:12:47,400 --> 00:12:51,080
Speaker 1: a specific quarter where it artificially suppressed costs and gave

235
00:12:51,080 --> 00:12:55,000
Speaker 1: you the flimsiest proof imaginable, specifically crafted for you to

236
00:12:55,040 --> 00:12:57,160
Speaker 1: share it as a triumph and spread the idea that

237
00:12:57,200 --> 00:13:01,280
Speaker 1: AI labs are actually profitable when their core economics haven't changed.

238
00:13:01,800 --> 00:13:04,680
Speaker 1: Costs increase linearly with revenue, and will continue to do

239
00:13:04,760 --> 00:13:08,160
Speaker 1: so in perpetuity. I genuinely can't wait for Open AI

240
00:13:08,240 --> 00:13:10,439
Speaker 1: and Anthropic to file that goddamn s once