1
00:00:00,265 --> 00:00:02,474
After Jack Dorsey cut half of his staff

2
00:00:02,504 --> 00:00:04,174
the company stock jumped 25%

3
00:00:05,264 --> 00:00:07,434
On February 26 2026

4
00:00:07,544 --> 00:00:09,874
one tweet blew up Silicon Valley

5
00:00:10,185 --> 00:00:10,933
Jack Dorsey

6
00:00:11,185 --> 00:00:12,624
co-founder of Twitter

7
00:00:12,625 --> 00:00:14,593
now CEO of Block

8
00:00:14,744 --> 00:00:16,314
posted a message himself

9
00:00:16,464 --> 00:00:18,173
with just one core piece of news

10
00:00:18,545 --> 00:00:21,873
Today we are making the hardest decision in the history of this company

11
00:00:21,945 --> 00:00:23,614
cutting the organization from over ten thousand people

12
00:00:23,864 --> 00:00:25,474
down to under six thousand

13
00:00:25,625 --> 00:00:26,574
More than four thousand people

14
00:00:26,745 --> 00:00:28,193
are leaving starting today

15
00:00:28,988 --> 00:00:30,357
What was even more explosive was the reason

16
00:00:30,627 --> 00:00:32,377
He did not say the macro economy was bad

17
00:00:32,547 --> 00:00:34,796
or the market was down and we need to cut costs

18
00:00:34,948 --> 00:00:35,916
What he said was

19
00:00:36,068 --> 00:00:39,476
AI capability made a leap at the end of 2025

20
00:00:39,668 --> 00:00:42,916
and we believe this company can no longer stay the size it was

21
00:00:43,348 --> 00:00:44,597
The day the news broke

22
00:00:44,668 --> 00:00:46,736
Block stock jumped 25% in after hours trading

23
00:00:48,071 --> 00:00:48,381
Hello

24
00:00:48,751 --> 00:00:49,220
everyone

25
00:00:49,471 --> 00:00:50,520
I am Leo Wang

26
00:00:50,711 --> 00:00:51,620
Have you ever wondered

27
00:00:51,992 --> 00:00:53,720
why cutting half the staff

28
00:00:53,751 --> 00:00:55,340
made Wall Street applaud like crazy

29
00:00:56,031 --> 00:00:57,510
The underlying logic here

30
00:00:57,511 --> 00:01:00,400
goes far deeper than the headline you saw

31
00:01:00,711 --> 00:01:03,840
Today I do not want to just retell a layoff story

32
00:01:03,952 --> 00:01:06,001
I want to show you three things

33
00:01:06,192 --> 00:01:06,540
First

34
00:01:06,912 --> 00:01:08,910
what exactly is Jack Dorsey doing

35
00:01:08,912 --> 00:01:10,641
what is he turning this company into

36
00:01:10,711 --> 00:01:11,061
Second

37
00:01:11,352 --> 00:01:13,320
how he thought his way to this decision

38
00:01:13,431 --> 00:01:16,240
why he had to act while the company was still making money

39
00:01:16,431 --> 00:01:16,781
Third

40
00:01:17,071 --> 00:01:18,751
and this is what I most want to talk about

41
00:01:18,752 --> 00:01:21,880
what this means for every single person who goes to work

42
00:01:22,270 --> 00:01:23,359
Let me start with the first question

43
00:01:23,790 --> 00:01:26,478
What exactly does Jack Dorsey want this company to become

44
00:01:26,829 --> 00:01:27,948
A month after the layoffs

45
00:01:27,949 --> 00:01:31,148
he and Sequoia Capital partner Roelof Botha

46
00:01:31,150 --> 00:01:32,838
co-published an essay

47
00:01:32,949 --> 00:01:35,219
titled From Hierarchy to Intelligence

48
00:01:35,749 --> 00:01:38,238
The core argument of that piece is extremely radical

49
00:01:38,469 --> 00:01:38,818
He says

50
00:01:39,109 --> 00:01:41,269
humans invented the corporate hierarchy

51
00:01:41,270 --> 00:01:42,799
two thousand years ago

52
00:01:43,029 --> 00:01:44,299
starting with the Roman legions

53
00:01:44,589 --> 00:01:46,068
a centurion managed ten men

54
00:01:46,070 --> 00:01:47,178
a general managed the centurions

55
00:01:47,430 --> 00:01:48,719
the emperor managed the generals

56
00:01:49,821 --> 00:01:51,290
What is this structure really

57
00:01:51,782 --> 00:01:53,031
It is information routing

58
00:01:53,382 --> 00:01:56,550
In a system too big for one person to see

59
00:01:56,582 --> 00:01:58,690
you need middle layers to pass information along

60
00:01:59,022 --> 00:02:01,990
pre-process decisions and keep everyone aligned

61
00:02:02,382 --> 00:02:03,430
For two thousand years

62
00:02:03,502 --> 00:02:05,751
the material of this structure never changed

63
00:02:05,805 --> 00:02:06,606
It was people

64
00:02:07,022 --> 00:02:08,011
people passing along messages

65
00:02:08,461 --> 00:02:09,670
people writing reports

66
00:02:09,701 --> 00:02:11,350
people sitting in meetings to align

67
00:02:11,662 --> 00:02:12,510
The problem is

68
00:02:12,622 --> 00:02:15,670
people as a material have a serious bandwidth bottleneck

69
00:02:16,582 --> 00:02:18,430
Every management layer you add

70
00:02:18,462 --> 00:02:20,271
distorts the information one more time

71
00:02:20,501 --> 00:02:21,731
Every extra meeting you hold

72
00:02:22,061 --> 00:02:23,990
delays the decision one more round

73
00:02:24,302 --> 00:02:26,180
Once a company passes a few hundred people

74
00:02:26,182 --> 00:02:28,551
a strange thing starts to happen

75
00:02:28,821 --> 00:02:29,810
Everybody is busy

76
00:02:30,101 --> 00:02:31,990
but the company as a whole gets slower

77
00:02:32,022 --> 00:02:32,530
At the end of the day

78
00:02:32,781 --> 00:02:34,111
this is not a management problem

79
00:02:34,221 --> 00:02:35,630
this is a materials problem

80
00:02:36,101 --> 00:02:37,780
It is like building a house out of wood

81
00:02:37,781 --> 00:02:39,671
six or seven floors is the limit

82
00:02:39,902 --> 00:02:41,910
not because the architect is not smart enough

83
00:02:41,982 --> 00:02:44,710
but because wood cannot carry any more weight

84
00:02:45,145 --> 00:02:46,875
Now AI has arrived

85
00:02:46,946 --> 00:02:50,715
It can pull together all of a company information around the clock

86
00:02:50,865 --> 00:02:52,895
Slack messages code commits

87
00:02:53,145 --> 00:02:56,034
customer feedback financial data meeting notes

88
00:02:56,145 --> 00:02:57,984
and then on top of all that information

89
00:02:57,985 --> 00:03:00,835
build what he calls an intelligence layer

90
00:03:01,265 --> 00:03:02,465
Any single employee

91
00:03:02,466 --> 00:03:05,034
can ask that intelligence layer a question directly

92
00:03:05,066 --> 00:03:08,354
and get all the context they need to make a decision

93
00:03:08,825 --> 00:03:10,254
Do you know what that means

94
00:03:11,066 --> 00:03:14,794
All those middle managers who passed messages wrote reports and ran alignment

95
00:03:14,825 --> 00:03:17,155
their core function has been replaced

96
00:03:17,746 --> 00:03:19,354
It is not that they are not smart enough

97
00:03:19,425 --> 00:03:20,954
it is not that they do not work hard enough

98
00:03:21,066 --> 00:03:24,514
it is that the function they carried information routing

99
00:03:24,666 --> 00:03:28,595
got taken over by a new material orders of magnitude more efficient

100
00:03:29,194 --> 00:03:32,882
So Jack Dorsey compressed the company into three roles

101
00:03:32,994 --> 00:03:35,002
The first is the individual contributor

102
00:03:35,073 --> 00:03:36,762
the people who actually do the work

103
00:03:36,793 --> 00:03:39,762
writing code doing design running sales

104
00:03:39,914 --> 00:03:42,042
With AI agents backing them up

105
00:03:42,194 --> 00:03:45,922
one person can now do what used to take a team or even ten people

106
00:03:46,353 --> 00:03:48,222
The second is the directly responsible individual

107
00:03:48,633 --> 00:03:49,672
They do not manage people

108
00:03:49,673 --> 00:03:51,042
they only own outcomes

109
00:03:51,233 --> 00:03:52,542
understand what the customer needs

110
00:03:52,714 --> 00:03:53,912
break it into tasks

111
00:03:53,914 --> 00:03:55,243
and make sure it gets done

112
00:03:56,146 --> 00:03:57,774
The third is the player coach

113
00:03:58,146 --> 00:03:58,744
Note

114
00:03:58,746 --> 00:04:00,554
this is not a manager in the traditional sense

115
00:04:00,746 --> 00:04:02,154
They are on the field playing themselves

116
00:04:02,225 --> 00:04:03,955
while helping teammates get stronger

117
00:04:04,306 --> 00:04:06,865
This relationship is not a permanent reporting line

118
00:04:06,866 --> 00:04:08,834
it is a temporary enablement relationship

119
00:04:09,146 --> 00:04:10,054
Today I help you

120
00:04:10,266 --> 00:04:11,714
tomorrow you help someone else

121
00:04:11,905 --> 00:04:13,994
The shape of the whole company changed too

122
00:04:14,105 --> 00:04:15,135
It is no longer a pyramid

123
00:04:15,345 --> 00:04:16,395
it is a circle

124
00:04:16,866 --> 00:04:17,774
AI sits at the center

125
00:04:18,105 --> 00:04:19,235
people sit around the edge

126
00:04:19,545 --> 00:04:21,944
Human work is no longer passing information

127
00:04:21,946 --> 00:04:23,314
it is calibrating the system

128
00:04:23,466 --> 00:04:26,874
taking human judgment values and understanding of customers

129
00:04:26,946 --> 00:04:29,235
and feeding it continuously into the intelligence layer at the center

130
00:04:29,913 --> 00:04:31,072
You might say

131
00:04:31,072 --> 00:04:33,462
is this just another Silicon Valley CEO selling a dream

132
00:04:33,993 --> 00:04:35,952
I talked before about the Notion founder

133
00:04:35,952 --> 00:04:37,481
Ivan Zhao and his essay

134
00:04:37,632 --> 00:04:39,762
He used steel and the steam engine as a metaphor

135
00:04:39,832 --> 00:04:42,402
saying we are still stuffing AI into an old house

136
00:04:42,553 --> 00:04:44,642
But what makes Jack Dorsey different

137
00:04:44,673 --> 00:04:46,231
is that he is not just writing essays

138
00:04:46,233 --> 00:04:49,082
he actually cut half of his own company

139
00:04:49,233 --> 00:04:50,632
This is not a thought experiment

140
00:04:50,632 --> 00:04:52,801
this is going all in with real money

141
00:04:53,018 --> 00:04:55,866
And he has one argument especially worth noting

142
00:04:55,977 --> 00:04:56,326
He says

143
00:04:56,737 --> 00:04:58,347
the biggest bottleneck at a company

144
00:04:58,378 --> 00:05:00,746
is actually its own product roadmap

145
00:05:01,241 --> 00:05:01,686
What does that mean

146
00:05:02,498 --> 00:05:04,987
A traditional company works like this

147
00:05:05,097 --> 00:05:06,987
product managers do user research

148
00:05:07,097 --> 00:05:07,967
write requirement docs

149
00:05:08,217 --> 00:05:09,186
rank priorities

150
00:05:09,338 --> 00:05:12,066
and then engineers build step by step along the roadmap

151
00:05:12,537 --> 00:05:13,537
But the problem is

152
00:05:13,537 --> 00:05:16,467
a roadmap is essentially your assumption about the world

153
00:05:16,977 --> 00:05:18,737
You assume what users want

154
00:05:18,737 --> 00:05:20,866
and then you build according to that assumption

155
00:05:21,706 --> 00:05:22,535
Jack Dorsey says

156
00:05:22,826 --> 00:05:23,864
in the AI era

157
00:05:23,866 --> 00:05:25,554
this flow should run in reverse

158
00:05:25,665 --> 00:05:27,475
Users state what they need directly

159
00:05:27,506 --> 00:05:30,395
the system calls on existing capabilities in real time to meet it

160
00:05:30,506 --> 00:05:31,455
and if it cannot

161
00:05:31,706 --> 00:05:33,105
that gap it could not fill

162
00:05:33,105 --> 00:05:35,835
automatically becomes the next thing to build

163
00:05:35,946 --> 00:05:38,064
The roadmap is not planned out

164
00:05:38,066 --> 00:05:39,874
it is forced out of you by users

165
00:05:40,625 --> 00:05:43,993
Block has one unique advantage here

166
00:05:44,145 --> 00:05:46,873
It owns both Cash App and Square

167
00:05:47,025 --> 00:05:51,914
which means it sees transaction data from both buyers and sellers

168
00:05:52,265 --> 00:05:56,313
Millions of real payments flow through its system every day

169
00:05:56,825 --> 00:05:59,513
Jack Dorsey said something really sharp

170
00:05:59,904 --> 00:06:02,594
Money is the most honest signal in the world

171
00:06:02,904 --> 00:06:07,224
What you say on social media what you write on a survey

172
00:06:07,224 --> 00:06:08,553
could all be fake

173
00:06:08,825 --> 00:06:10,503
but where your money goes

174
00:06:10,505 --> 00:06:12,114
that signal cannot lie

175
00:06:12,825 --> 00:06:16,474
When you plug that dense real signal into an AI intelligence layer

176
00:06:16,505 --> 00:06:19,274
your company stops being an execution machine

177
00:06:19,344 --> 00:06:23,553
and becomes an agent that senses adapts and evolves on its own

178
00:06:23,705 --> 00:06:26,393
That is why the episode of their podcast was titled

179
00:06:26,465 --> 00:06:29,654
every company can now become a mini AGI

180
00:06:30,965 --> 00:06:31,154
Alright

181
00:06:31,404 --> 00:06:32,813
now to the second question

182
00:06:32,965 --> 00:06:35,373
and the part I find most worth thinking about

183
00:06:35,565 --> 00:06:39,013
How did Jack Dorsey talk himself into swinging this axe

184
00:06:39,404 --> 00:06:42,614
On the podcast he broke down his own thinking in detail

185
00:06:43,044 --> 00:06:46,094
He says he ran a very cold reverse calculation

186
00:06:46,409 --> 00:06:47,209
Step one

187
00:06:47,284 --> 00:06:50,493
to keep the existing business running at one hundred percent

188
00:06:50,604 --> 00:06:51,974
what is the minimum headcount

189
00:06:52,325 --> 00:06:52,834
Step two

190
00:06:53,205 --> 00:06:55,893
to meet every regulatory and compliance requirement

191
00:06:55,924 --> 00:06:57,334
what is the minimum headcount

192
00:06:57,924 --> 00:06:58,474
Step three

193
00:06:58,804 --> 00:06:59,964
on top of that

194
00:06:59,965 --> 00:07:04,493
to keep growing and rebuild the company as an agent

195
00:07:04,604 --> 00:07:05,654
how many people do you need

196
00:07:06,044 --> 00:07:07,553
Stack those three conditions together

197
00:07:07,885 --> 00:07:09,164
and the number you get

198
00:07:09,164 --> 00:07:10,373
is six thousand people

199
00:07:10,895 --> 00:07:11,285
But

200
00:07:11,576 --> 00:07:13,104
inside that calculation

201
00:07:13,136 --> 00:07:15,984
one judgment mattered more than how many people to cut

202
00:07:16,376 --> 00:07:17,124
He said something

203
00:07:17,616 --> 00:07:18,465
that roughly means

204
00:07:18,616 --> 00:07:21,984
you must restructure by choice from a position of strength

205
00:07:22,056 --> 00:07:25,104
not shrink by force from a position of weakness

206
00:07:25,936 --> 00:07:26,845
The weight of that sentence

207
00:07:27,095 --> 00:07:28,304
is worth sitting with

208
00:07:28,696 --> 00:07:31,225
You have seen how too many companies die

209
00:07:31,576 --> 00:07:32,725
They do not die all at once

210
00:07:33,016 --> 00:07:34,544
they get worn down slowly

211
00:07:34,816 --> 00:07:36,044
Cut ten percent this year

212
00:07:36,496 --> 00:07:38,265
cut twenty percent next year

213
00:07:38,455 --> 00:07:40,564
every round of layoffs is passive defense

214
00:07:40,855 --> 00:07:41,814
no direction

215
00:07:41,816 --> 00:07:42,645
no creativity

216
00:07:43,016 --> 00:07:44,785
just endless shrinking

217
00:07:45,295 --> 00:07:46,645
Employees live in fear

218
00:07:47,056 --> 00:07:48,944
the best people leave first

219
00:07:49,016 --> 00:07:51,424
and whoever stays coasts and waits for the next round

220
00:07:51,816 --> 00:07:53,004
That kind of slow death

221
00:07:53,215 --> 00:07:55,465
hurts more than cutting off an arm in one go

222
00:07:56,145 --> 00:07:58,754
Jack Dorsey chose the opposite path

223
00:07:59,145 --> 00:08:02,033
He moved while Block revenue was still growing

224
00:08:02,504 --> 00:08:04,314
In the fourth quarter of 2025

225
00:08:04,384 --> 00:08:06,953
the company posted revenue of six point two five billion dollars

226
00:08:07,145 --> 00:08:09,134
with gross profit up 24% year over year

227
00:08:09,985 --> 00:08:12,913
and it raised its full year guidance for 2026

228
00:08:13,504 --> 00:08:16,754
This is not a dying company fighting to survive

229
00:08:16,985 --> 00:08:20,473
this is a profitable company choosing to overthrow itself

230
00:08:21,105 --> 00:08:23,554
He says the whole decision took less than three weeks

231
00:08:23,744 --> 00:08:28,554
the team debated ran scenarios fast and kept correcting assumptions

232
00:08:28,785 --> 00:08:30,333
The board not only did not block it

233
00:08:30,704 --> 00:08:32,153
they strongly supported it

234
00:08:32,345 --> 00:08:32,853
Why

235
00:08:33,785 --> 00:08:36,103
Because the whole chain of logic was clear

236
00:08:36,105 --> 00:08:40,154
principles reasoning and a verifiable picture of the future

237
00:08:40,573 --> 00:08:43,741
On the podcast he also shared a very personal reflection

238
00:08:44,132 --> 00:08:45,932
He says he once made a big mistake

239
00:08:45,933 --> 00:08:47,302
over delegating

240
00:08:47,533 --> 00:08:50,462
He split Block into separate business units

241
00:08:50,573 --> 00:08:51,722
one leader for Square

242
00:08:52,053 --> 00:08:53,861
one leader for Cash App

243
00:08:53,972 --> 00:08:55,981
trying to run a multi CEO structure

244
00:08:56,132 --> 00:08:59,062
The result was a loose holding company

245
00:08:59,092 --> 00:09:02,462
split culture mismatched rhythm broken execution

246
00:09:02,972 --> 00:09:05,101
What he learned from that lesson was this

247
00:09:05,212 --> 00:09:07,741
before a company truly becomes an agent

248
00:09:07,813 --> 00:09:10,381
over delegating actually makes the system lose control

249
00:09:10,653 --> 00:09:12,202
You have to build the center first

250
00:09:12,573 --> 00:09:13,621
before you can let go

251
00:09:13,926 --> 00:09:16,855
He told another story that really moved me

252
00:09:17,326 --> 00:09:19,815
He said when Cash App first started

253
00:09:19,967 --> 00:09:21,656
almost everyone was against it

254
00:09:22,007 --> 00:09:22,955
the board did not understand it

255
00:09:23,326 --> 00:09:24,695
the internal team did not buy it

256
00:09:25,046 --> 00:09:26,075
Every day he held the line

257
00:09:26,487 --> 00:09:28,335
he was burning his own credibility

258
00:09:28,727 --> 00:09:29,975
But he said this

259
00:09:30,166 --> 00:09:32,975
every company has multiple founding moments

260
00:09:33,367 --> 00:09:35,256
not just the one at the very beginning

261
00:09:35,647 --> 00:09:39,416
Every time you make a key choice that goes against everyone instincts

262
00:09:39,527 --> 00:09:41,935
you are redefining that company

263
00:09:42,406 --> 00:09:45,215
That is how he can do something very counterintuitive

264
00:09:45,527 --> 00:09:47,616
accept losing trust in the short term

265
00:09:47,686 --> 00:09:49,416
to buy being right in the long term

266
00:09:49,847 --> 00:09:52,416
The condition is that you know why you are doing it

267
00:09:52,527 --> 00:09:55,095
and you know how you will win it back later

268
00:09:56,293 --> 00:09:58,743
He even sums that ability up in one phrase

269
00:09:58,893 --> 00:10:01,263
not being held hostage by outside opinion

270
00:10:01,893 --> 00:10:02,562
But he stresses

271
00:10:02,854 --> 00:10:04,332
this is not blind confidence

272
00:10:04,334 --> 00:10:06,263
it is conviction grounded in principles

273
00:10:06,854 --> 00:10:07,483
At this point

274
00:10:07,773 --> 00:10:09,662
Roelof Botha added one more cut

275
00:10:10,133 --> 00:10:13,223
He said a good CEO needs three core qualities

276
00:10:13,374 --> 00:10:15,702
authenticity rationality and empathy

277
00:10:16,013 --> 00:10:17,042
but in the AI era

278
00:10:17,334 --> 00:10:18,123
you have to add one more

279
00:10:18,574 --> 00:10:19,263
speed

280
00:10:19,734 --> 00:10:20,962
because change is moving too fast

281
00:10:21,254 --> 00:10:22,373
Hesitate for six months

282
00:10:22,374 --> 00:10:24,503
and you might miss an entire generation of opportunity

283
00:10:24,744 --> 00:10:27,263
Then Jack Dorsey said something really interesting

284
00:10:27,264 --> 00:10:30,672
He said most people treat AI as an output tool

285
00:10:30,744 --> 00:10:33,753
having it write your copy your code your designs

286
00:10:33,783 --> 00:10:35,353
but that gets it completely backwards

287
00:10:35,584 --> 00:10:37,993
The more important value of AI is input enhancement

288
00:10:38,183 --> 00:10:39,212
not helping you produce

289
00:10:39,464 --> 00:10:41,182
but giving you better information

290
00:10:41,183 --> 00:10:42,753
so you can make better judgments

291
00:10:42,943 --> 00:10:44,432
He says for a full year

292
00:10:44,504 --> 00:10:47,113
he spent three hours every morning on one thing

293
00:10:47,224 --> 00:10:49,832
feeding the output of various AI tools to himself

294
00:10:50,023 --> 00:10:52,033
not letting AI make decisions for him

295
00:10:52,063 --> 00:10:55,182
but using AI to give himself more complete context

296
00:10:55,183 --> 00:10:56,753
and then deciding himself

297
00:10:57,681 --> 00:10:59,491
The logic behind this is

298
00:10:59,561 --> 00:11:01,281
in an environment of information overload

299
00:11:01,281 --> 00:11:03,600
what is truly scarce is not the ability to produce

300
00:11:03,602 --> 00:11:05,970
it is something he calls taste

301
00:11:06,482 --> 00:11:08,001
Jack Dorsey stresses

302
00:11:08,002 --> 00:11:10,491
this taste is not about whether something looks good

303
00:11:10,521 --> 00:11:11,810
it is three things

304
00:11:12,161 --> 00:11:14,131
do you have a clear direction

305
00:11:14,561 --> 00:11:16,891
are you willing to push it obsessively

306
00:11:17,321 --> 00:11:20,771
and is it actually more to the point than what already exists

307
00:11:21,205 --> 00:11:21,962
He says

308
00:11:21,962 --> 00:11:26,131
the problem now is too many companies are making copies of copies of copies

309
00:11:26,561 --> 00:11:29,251
because AI made copying too easy

310
00:11:29,281 --> 00:11:30,891
templatization happens too fast

311
00:11:31,321 --> 00:11:32,310
What is truly scarce

312
00:11:32,482 --> 00:11:33,350
is no longer capability

313
00:11:33,561 --> 00:11:34,611
it is a point of view

314
00:11:35,267 --> 00:11:37,835
The way people work inside Block changed too

315
00:11:37,987 --> 00:11:41,916
Meetings used to mean bringing slides docs and ideas

316
00:11:42,066 --> 00:11:44,355
now everyone brings a prototype

317
00:11:44,507 --> 00:11:47,195
that can be changed discussed and tested on the spot

318
00:11:47,387 --> 00:11:49,186
Decisions are no longer based on descriptions

319
00:11:49,186 --> 00:11:51,475
but on something real you can actually use

320
00:11:51,867 --> 00:11:54,676
Jack Dorsey says this brought one fundamental change

321
00:11:54,826 --> 00:11:56,436
They used to preach focus

322
00:11:56,466 --> 00:11:58,075
do one thing at a time

323
00:11:58,186 --> 00:12:00,796
now you can explore several directions at once

324
00:12:00,907 --> 00:12:03,315
because the cost of trying has dropped to almost zero

325
00:12:03,426 --> 00:12:05,995
A prototype takes an hour to build

326
00:12:06,535 --> 00:12:07,863
But then he turns around and says

327
00:12:08,015 --> 00:12:10,383
one thing has actually become more important

328
00:12:10,574 --> 00:12:12,503
that last twenty percent

329
00:12:12,935 --> 00:12:15,343
judgment taste creativity

330
00:12:15,574 --> 00:12:18,424
AI can get you through the first eighty percent

331
00:12:18,535 --> 00:12:22,064
but what really decides whether this is right and worth doing

332
00:12:22,135 --> 00:12:23,023
is still a human

333
00:12:23,857 --> 00:12:24,046
Alright

334
00:12:24,497 --> 00:12:26,107
time for the third question

335
00:12:26,218 --> 00:12:28,067
what does this have to do with you

336
00:12:28,377 --> 00:12:30,746
Let me start with an uncomfortable fact

337
00:12:31,017 --> 00:12:33,227
The day Block announced the layoffs

338
00:12:33,338 --> 00:12:37,906
Wall Street voted yes with a 25% after hours jump

339
00:12:38,257 --> 00:12:40,046
Right after that Fortune CNN

340
00:12:40,578 --> 00:12:41,867
and Bloomberg all reported it

341
00:12:41,897 --> 00:12:44,386
and nearly all of them quoted one line from Jack Dorsey

342
00:12:44,497 --> 00:12:45,166
He said at the time

343
00:12:45,458 --> 00:12:46,496
within a year

344
00:12:46,497 --> 00:12:49,827
most companies will end up where Block is

345
00:12:50,097 --> 00:12:51,227
That is not a threat

346
00:12:51,257 --> 00:12:53,467
that is a trend already unfolding

347
00:12:53,821 --> 00:12:54,411
Think about it

348
00:12:54,901 --> 00:12:57,550
if you were the CEO of a public company

349
00:12:57,781 --> 00:13:00,310
and you saw Block cut half its people

350
00:13:00,462 --> 00:13:01,790
with the business not collapsing

351
00:13:01,821 --> 00:13:02,930
and the stock jumping

352
00:13:03,382 --> 00:13:04,210
what would you think

353
00:13:04,742 --> 00:13:06,411
What would your board think

354
00:13:06,862 --> 00:13:08,651
What would your investors think

355
00:13:09,382 --> 00:13:11,131
And what do Block employees say

356
00:13:11,781 --> 00:13:14,430
Some reports quoted people inside

357
00:13:14,622 --> 00:13:17,070
saying morale might be the lowest in four years

358
00:13:17,342 --> 00:13:19,710
Someone said flat out at an all hands

359
00:13:19,982 --> 00:13:23,950
that being forced to use large language models is insane

360
00:13:24,622 --> 00:13:27,670
Current and laid off employees also told The Guardian

361
00:13:27,901 --> 00:13:32,190
that about 95% of AI generated code still needs human fixes

362
00:13:32,382 --> 00:13:35,590
and in heavily regulated areas like banking and payments

363
00:13:35,661 --> 00:13:37,991
AI simply cannot stand on its own yet

364
00:13:38,440 --> 00:13:41,109
Do these criticisms mean Jack Dorsey is wrong

365
00:13:41,563 --> 00:13:42,364
Not necessarily

366
00:13:42,600 --> 00:13:44,129
What these criticisms show

367
00:13:44,200 --> 00:13:48,048
is that what he is doing sits in an extremely painful middle zone

368
00:13:48,399 --> 00:13:50,209
The old structure is already coming down

369
00:13:50,360 --> 00:13:52,489
the new one is not fully standing yet

370
00:13:52,720 --> 00:13:55,768
and everyone is paying the cost of that transition

371
00:13:56,560 --> 00:13:57,808
But think about it

372
00:13:57,919 --> 00:13:59,989
which paradigm shift was ever any different

373
00:14:00,600 --> 00:14:03,568
The day factories moved the steam engine away from the river

374
00:14:03,759 --> 00:14:06,288
the water wheel repairmen of course felt the sky was falling

375
00:14:06,680 --> 00:14:10,249
but history does not wait for everyone to be ready before it turns

376
00:14:11,034 --> 00:14:12,363
As an investor

377
00:14:12,434 --> 00:14:14,803
I look at this from a very personal angle

378
00:14:15,273 --> 00:14:16,353
Over the past decade or so

379
00:14:16,353 --> 00:14:18,282
I have invested in more than three hundred early stage companies

380
00:14:18,674 --> 00:14:21,002
I have seen the cost structure of too many teams

381
00:14:21,353 --> 00:14:23,073
For a twenty person startup

382
00:14:23,074 --> 00:14:25,603
sixty to seventy percent of costs are people

383
00:14:26,074 --> 00:14:29,162
If AI really lets one person do the work of ten

384
00:14:29,233 --> 00:14:30,063
what does that mean

385
00:14:30,674 --> 00:14:33,032
It means you might only need two or three people

386
00:14:33,034 --> 00:14:35,282
to build what used to take twenty

387
00:14:35,954 --> 00:14:38,083
The barrier to starting a company is dropping fast

388
00:14:38,233 --> 00:14:40,762
but the bar for each individual is rising fast

389
00:14:41,665 --> 00:14:43,674
On the podcast Jack Dorsey brought up

390
00:14:43,825 --> 00:14:45,515
one especially important idea

391
00:14:45,530 --> 00:14:46,329
meditation

392
00:14:46,825 --> 00:14:48,954
He does not mean the mystical kind

393
00:14:49,185 --> 00:14:51,915
He says meditation is essentially training three things

394
00:14:52,106 --> 00:14:55,755
focus observation and the ability to not react immediately

395
00:14:55,986 --> 00:14:57,375
Translated into CEO language

396
00:14:57,746 --> 00:14:59,395
it means inside a complex system

397
00:14:59,425 --> 00:15:00,954
not getting carried off by the noise

398
00:15:01,305 --> 00:15:02,755
That made me think of something

399
00:15:03,025 --> 00:15:04,015
In the AI era

400
00:15:04,386 --> 00:15:06,834
your competitor is not AI itself

401
00:15:07,025 --> 00:15:09,594
it is the people who use AI better than you

402
00:15:09,923 --> 00:15:11,832
And what does using AI better depend on

403
00:15:12,284 --> 00:15:14,133
It depends on knowing what you want

404
00:15:14,443 --> 00:15:15,513
You need a direction first

405
00:15:15,884 --> 00:15:17,092
before tools are any use

406
00:15:17,484 --> 00:15:18,433
For someone with no direction

407
00:15:18,644 --> 00:15:19,883
the more powerful the tool

408
00:15:19,884 --> 00:15:20,893
the more lost they get

409
00:15:21,323 --> 00:15:22,592
This is really a consciousness problem

410
00:15:22,963 --> 00:15:24,092
not a technology problem

411
00:15:24,758 --> 00:15:26,907
Jack Dorsey did what everyone was thinking about

412
00:15:27,278 --> 00:15:28,887
but almost nobody dared to do

413
00:15:29,118 --> 00:15:31,086
He used action to test a hypothesis

414
00:15:31,358 --> 00:15:34,287
that the company of the future is not an organization that hires a lot of people

415
00:15:34,317 --> 00:15:35,907
but an intelligent system with AI at its core

416
00:15:36,278 --> 00:15:39,047
and human judgment at its edges

417
00:15:39,398 --> 00:15:40,427
Is that hypothesis right

418
00:15:40,878 --> 00:15:42,367
Time will answer that

419
00:15:42,837 --> 00:15:44,887
The quarterly reports after the layoffs

420
00:15:44,917 --> 00:15:47,486
will deliver the first batch of real data

421
00:15:47,758 --> 00:15:49,367
But whatever the result

422
00:15:49,518 --> 00:15:51,606
one thing is already irreversible

423
00:15:52,358 --> 00:15:55,086
Starting February 26 2026

424
00:15:55,197 --> 00:15:58,367
the question of what a company in the AI era should look like

425
00:15:58,718 --> 00:16:00,767
went from a vision slide in a deck

426
00:16:00,837 --> 00:16:03,767
to a real experiment that is actually running

427
00:16:04,226 --> 00:16:05,985
Jack Dorsey said it himself

428
00:16:05,985 --> 00:16:07,595
there is no Dorsey playbook

429
00:16:07,905 --> 00:16:09,995
this is not a proven answer

430
00:16:10,025 --> 00:16:11,995
this is a path still being explored

431
00:16:12,265 --> 00:16:13,674
But he said one more thing

432
00:16:13,865 --> 00:16:16,995
if I had to build a company from zero today

433
00:16:17,105 --> 00:16:19,674
I would not design it as an organization

434
00:16:19,706 --> 00:16:22,075
I would design it as an agent

435
00:16:22,426 --> 00:16:24,434
That line is worth chewing on again and again

436
00:16:24,745 --> 00:16:26,195
not because it must be right

437
00:16:26,226 --> 00:16:30,315
but because it might change the whole coordinate system you use to think about work

438
00:16:31,463 --> 00:16:32,932
The company you are in right now

439
00:16:33,344 --> 00:16:34,423
is it an organization

440
00:16:34,423 --> 00:16:35,452
or is it an agent

441
00:16:36,048 --> 00:16:36,847
And you yourself

442
00:16:37,104 --> 00:16:39,352
are you a node that an organization assigns tasks to

443
00:16:39,423 --> 00:16:42,493
or a standalone unit that talks to signals from the world directly

444
00:16:43,104 --> 00:16:44,312
The answer to that question

445
00:16:44,384 --> 00:16:47,352
might decide where you stand over the next five years

446
00:16:47,583 --> 00:16:48,373
I am Leo Wang

447
00:16:48,624 --> 00:16:49,514
see you next time