1
00:00:00,000 --> 00:00:03,340
Hello everybody, welcome to my channel

2
00:00:03,340 --> 00:00:06,640
老田们,今天咱们聊一个劲爆的话题啊

3
00:00:06,640 --> 00:00:10,900
三哥们是怎么把硅谷最强公司给干趴下的

4
00:00:10,900 --> 00:00:12,140
事情是这样啊

5
00:00:12,140 --> 00:00:17,760
就在前几天,AI算力圈的风向标分析发布了一份报告

6
00:00:17,760 --> 00:00:20,360
直接给谷歌下了死亡盘决数了

7
00:00:20,360 --> 00:00:22,340
报告原话是这么说的啊

8
00:00:22,340 --> 00:00:24,940
说谷歌还会继续发布模型

9
00:00:24,940 --> 00:00:28,900
但是重回巅峰的可能性已经降到了零

10
00:00:28,900 --> 00:00:30,680
啥概念啊兄弟们

11
00:00:30,680 --> 00:00:31,960
这就好比啊

12
00:00:31,960 --> 00:00:33,860
博尔特跑着跑着突然说

13
00:00:33,860 --> 00:00:35,080
我不行了

14
00:00:35,080 --> 00:00:37,480
我改行卖烤串吧

15
00:00:37,480 --> 00:00:39,180
谁也没想到吧

16
00:00:39,180 --> 00:00:41,800
AI竞赛里面的第一个掉队的

17
00:00:41,800 --> 00:00:44,040
不是天天画饼的马斯克

18
00:00:44,040 --> 00:00:45,920
也不是心源一马的小诈

19
00:00:45,920 --> 00:00:49,220
既然是AI的开山鼻祖谷歌

20
00:00:49,220 --> 00:00:51,580
Transformer是谷歌发明的啊

21
00:00:51,580 --> 00:00:54,340
AI聊天机器人也是谷歌先捏出来的

22
00:00:54,340 --> 00:00:58,220
结果最后拿着这俩东西颠覆世界的

23
00:00:58,220 --> 00:00:59,380
偏偏是别人

24
00:00:59,380 --> 00:01:01,720
这就跟你自己种了个稀果

25
00:01:01,720 --> 00:01:02,960
结果被别人摘了

26
00:01:02,960 --> 00:01:04,100
还卖了个好价格

27
00:01:04,100 --> 00:01:05,500
你只能在旁边干瞪眼

28
00:01:05,500 --> 00:01:06,740
那是一模一样的

29
00:01:06,740 --> 00:01:09,420
谷歌的柯达诗科又来了

30
00:01:09,420 --> 00:01:10,820
其实这件事

31
00:01:10,820 --> 00:01:12,040
它早就有征兆了

32
00:01:12,040 --> 00:01:15,840
当年柯达首先发明了数码相机技术

33
00:01:15,840 --> 00:01:17,480
结果管理层觉得

34
00:01:17,480 --> 00:01:20,940
这玩意会影响咱们的胶片业务

35
00:01:20,940 --> 00:01:22,800
硬是就按住不发展

36
00:01:22,800 --> 00:01:25,060
结果自己把自己给做死了

37
00:01:25,060 --> 00:01:28,020
谷歌是干了一件一模一样的事

38
00:01:28,020 --> 00:01:32,160
在OpenAI发布ChatGDP之前整整的一年

39
00:01:32,160 --> 00:01:35,760
谷歌就已经研发出了大语言模型

40
00:01:35,760 --> 00:01:36,460
Lamida

41
00:01:36,460 --> 00:01:38,800
现代AI的基础

42
00:01:38,800 --> 00:01:41,740
Transformer的框架也是谷歌创立的

43
00:01:41,740 --> 00:01:45,660
结果当时的谷歌已经换上了一群三哥高果

44
00:01:45,660 --> 00:01:47,340
那天天就是喝咖啡

45
00:01:47,340 --> 00:01:51,900
在这些眼界高远的印度管理层眼中

46
00:01:51,900 --> 00:01:54,700
这玩意可能影响谷歌搜索业务

47
00:01:54,700 --> 00:01:56,360
不发布直接血仓

48
00:01:56,360 --> 00:02:00,500
谷歌的CEO皮查一后来还出来解释

49
00:02:00,500 --> 00:02:02,920
说是因为安全考虑

50
00:02:02,920 --> 00:02:04,520
说质量不行

51
00:02:04,520 --> 00:02:05,960
兄弟们你信吗

52
00:02:05,960 --> 00:02:08,800
一个手握全球最多数据

53
00:02:08,800 --> 00:02:10,140
最多算力

54
00:02:10,140 --> 00:02:12,280
最顶尖科学家的公司

55
00:02:12,280 --> 00:02:14,140
你跟我说质量不行

56
00:02:14,140 --> 00:02:15,120
你糊弄鬼呢

57
00:02:15,120 --> 00:02:17,940
典型的印度式的短视思维

58
00:02:17,940 --> 00:02:20,340
他们只看眼前这三亩丁

59
00:02:20,340 --> 00:02:22,300
看不到外面整个太平洋啊

60
00:02:22,300 --> 00:02:23,040
好

61
00:02:23,040 --> 00:02:24,400
你以为这就完了

62
00:02:24,400 --> 00:02:25,760
突央突三婆

63
00:02:25,760 --> 00:02:29,780
8月5号2024年诺贝尔化学奖得主

64
00:02:29,780 --> 00:02:32,600
AlphaGo和AlphaFold的缔造者

65
00:02:32,600 --> 00:02:37,160
被称为硅谷AI科学灵魂的哈萨比斯

66
00:02:37,160 --> 00:02:39,760
被割除了迪普麦的CEO的职位

67
00:02:39,760 --> 00:02:40,960
就在同一天

68
00:02:40,960 --> 00:02:45,620
效力谷歌27年的传奇工程师杰弗迪恩也走了

69
00:02:45,620 --> 00:02:47,560
这哥们是什么级别呀

70
00:02:47,560 --> 00:02:48,960
谷歌内部有个段子

71
00:02:48,960 --> 00:02:52,880
说他写代码的速度超过别人阅读代码的速度

72
00:02:52,880 --> 00:02:55,260
他每次编译前都不查语法

73
00:02:55,260 --> 00:02:56,900
如果编译器报错

74
00:02:56,900 --> 00:02:58,720
那一定是编译器自己错了

75
00:02:58,720 --> 00:03:00,220
编译器都不敢报错

76
00:03:00,220 --> 00:03:02,020
只敢向他请教语法

77
00:03:02,020 --> 00:03:03,560
就这么个神级人物

78
00:03:03,560 --> 00:03:04,000
走了

79
00:03:04,000 --> 00:03:05,740
两个月前啊

80
00:03:05,740 --> 00:03:08,380
Transformant的论文的另一位核心作者

81
00:03:08,380 --> 00:03:09,800
诺贝尔化学奖得主

82
00:03:09,800 --> 00:03:11,860
No Man转投了OpenAI

83
00:03:11,860 --> 00:03:14,360
AlphaFold的核心领导者

84
00:03:14,360 --> 00:03:16,180
Jumper转投了Ansuric

85
00:03:16,180 --> 00:03:17,820
注意啊兄弟们

86
00:03:17,820 --> 00:03:20,140
这些人可不是一个人走的

87
00:03:20,140 --> 00:03:22,540
每个人都是拉走整整一个队伍

88
00:03:22,540 --> 00:03:23,480
谷歌

89
00:03:23,480 --> 00:03:25,240
Jimny AI模型啊

90
00:03:25,240 --> 00:03:28,780
三代核心研发团队全部离开了

91
00:03:28,780 --> 00:03:30,540
Alpha4的团队啊

92
00:03:30,540 --> 00:03:32,300
几乎是一人都不胜了

93
00:03:32,300 --> 00:03:36,620
要知道谷歌2024年花了27亿美元

94
00:03:36,620 --> 00:03:40,020
把AI研究的科学家杀字给挖过来

95
00:03:40,020 --> 00:03:43,120
不到两年人家反手就加盟了OpenAI

96
00:03:43,120 --> 00:03:45,720
27亿美元啊兄弟们

97
00:03:45,720 --> 00:03:47,760
买了不来两年的忠诚啊

98
00:03:47,760 --> 00:03:49,640
杰弗迪恩走的时候

99
00:03:49,640 --> 00:03:52,280
带走了三名核心研究人员

100
00:03:52,280 --> 00:03:55,480
谷歌技术核心人员那几乎是都跑了呀

101
00:03:55,480 --> 00:03:57,080
那最荒诞的是什么呀

102
00:03:57,080 --> 00:03:59,820
Alphabet自己还投钱入股了

103
00:03:59,820 --> 00:04:00,780
Discourable Lope

104
00:04:00,780 --> 00:04:03,260
成为了他的云合作伙伴

105
00:04:03,260 --> 00:04:05,640
核心班底另立山头

106
00:04:05,640 --> 00:04:07,860
老东家一看拦不住了

107
00:04:07,860 --> 00:04:10,440
只能卑微的掏钱买了个包底

108
00:04:10,440 --> 00:04:12,760
哎您搜了用一下我们的云服务行吗

109
00:04:12,760 --> 00:04:13,940
那关键问题来了

110
00:04:13,940 --> 00:04:16,280
这些核心技术人员为什么跑呢

111
00:04:16,280 --> 00:04:19,240
一句话都是三个的功劳啊

112
00:04:19,240 --> 00:04:21,080
那么这个问题是

113
00:04:21,080 --> 00:04:24,620
这些顶级的人才为啥铁了心要跑呢

114
00:04:24,620 --> 00:04:27,640
作为硅谷的顶级大厂啊

115
00:04:27,640 --> 00:04:29,140
工资肯定不是核心原因

116
00:04:29,140 --> 00:04:31,680
真正的答案就俩字管理

117
00:04:31,680 --> 00:04:33,340
让三个管理了

118
00:04:33,340 --> 00:04:34,600
你说那能有好吗

119
00:04:34,600 --> 00:04:36,380
硅谷普遍认为啊

120
00:04:36,380 --> 00:04:39,660
谷歌正在告别科学家制AI的时代

121
00:04:39,660 --> 00:04:42,940
从追求前沿突破转向拼产品

122
00:04:42,940 --> 00:04:44,940
落地的商业营收

123
00:04:44,940 --> 00:04:46,360
翻译成人话就是

124
00:04:46,360 --> 00:04:48,000
三个管理层啊

125
00:04:48,000 --> 00:04:49,860
要短期见效快的产品

126
00:04:49,860 --> 00:04:52,360
不要AGI这种长远的目标

127
00:04:52,360 --> 00:04:53,260
不给资金

128
00:04:53,260 --> 00:04:54,120
不给算力

129
00:04:54,120 --> 00:04:56,300
顶级人才留在谷歌干嘛呢

130
00:04:56,300 --> 00:04:57,220
当吉祥物吗

131
00:04:57,220 --> 00:04:59,660
那谷歌那么多算力

132
00:04:59,660 --> 00:05:00,780
多出来干嘛呀

133
00:05:00,780 --> 00:05:01,860
出租啊

134
00:05:01,860 --> 00:05:03,520
三哥最喜欢了

135
00:05:03,520 --> 00:05:05,820
提升Google Cloud的营收啊

136
00:05:05,820 --> 00:05:08,740
多做点TPU芯片的生意啊

137
00:05:08,740 --> 00:05:11,020
三哥高管们掰着手指着一算

138
00:05:11,020 --> 00:05:13,460
这OpenAI估值还不到1万亿

139
00:05:13,460 --> 00:05:15,560
英伟达估值5万亿

140
00:05:15,560 --> 00:05:18,360
那搞硬件搞云服务来钱多快

141
00:05:18,360 --> 00:05:19,680
那估值更高啊

142
00:05:19,680 --> 00:05:21,720
这操作简直那么绝了

143
00:05:21,720 --> 00:05:23,720
一个AI公司不搞AI

144
00:05:23,720 --> 00:05:26,040
跑去当二房东出租算力了

145
00:05:26,040 --> 00:05:29,060
就跟他开了一家米其林三星餐厅

146
00:05:29,060 --> 00:05:30,820
结果闲做菜太累

147
00:05:30,820 --> 00:05:33,480
把厨房直接改成仓库收租了

148
00:05:33,480 --> 00:05:35,080
你格局呢兄弟

149
00:05:35,080 --> 00:05:37,340
一位Transformer的论文

150
00:05:37,340 --> 00:05:39,840
联合作者一针见血的指出

151
00:05:39,840 --> 00:05:43,100
谷歌的官僚主义已经病入膏肓

152
00:05:43,100 --> 00:05:46,040
在这地方根本干不成任何的事

153
00:05:46,040 --> 00:05:48,000
所以当一名三哥

154
00:05:48,000 --> 00:05:50,560
当上一家跨国公司CEO的时候

155
00:05:50,560 --> 00:05:53,540
这家公司就注定要被回了

156
00:05:53,540 --> 00:05:57,400
其实这不是三哥第一次在硅谷兴风作浪了

157
00:05:57,400 --> 00:05:59,260
Meta之前就吃了一次大亏

158
00:05:59,260 --> 00:06:02,780
当年Meta团队里面超过一半都是印度人

159
00:06:02,780 --> 00:06:04,820
表面看起来进度很好

160
00:06:04,820 --> 00:06:08,760
推出大模型的自驾的评测里面得分非常高

161
00:06:08,760 --> 00:06:11,160
小扎很开心的赶紧推向市场

162
00:06:11,160 --> 00:06:13,260
结果大家实际一测

163
00:06:13,260 --> 00:06:15,880
发现各种拉跨就是一坨而已

164
00:06:15,880 --> 00:06:17,580
然后小扎懵了

165
00:06:17,580 --> 00:06:19,460
这么多钱这么多人砸去

166
00:06:19,460 --> 00:06:21,020
不是评分很高吗

167
00:06:21,020 --> 00:06:22,640
怎么一上市就拉了拖大的

168
00:06:22,640 --> 00:06:27,760
结果META的前首席科学家杨立坤亲口承认

169
00:06:27,760 --> 00:06:32,740
团队针对不同的基准测试采用了不同的模型变体

170
00:06:32,740 --> 00:06:34,800
目的就是为了优化分数

171
00:06:34,800 --> 00:06:36,320
说白了就是跑分优化

172
00:06:36,320 --> 00:06:39,240
弄出来个高分低能的弱智玩意儿

173
00:06:39,240 --> 00:06:41,940
小渣当时气子擦不眼没背过气去

174
00:06:41,940 --> 00:06:46,380
直接对整个生成式AI团队都失去了信心

175
00:06:46,380 --> 00:06:48,160
把整个部门给边缘化了

176
00:06:48,160 --> 00:06:52,520
随后开启了一场针对三哥的大扫荡大清洗

177
00:06:52,520 --> 00:06:54,640
所以看清楚形势了吗

178
00:06:54,640 --> 00:06:56,720
硅谷的几大AI巨头

179
00:06:56,720 --> 00:06:58,460
谁的华人占比高

180
00:06:58,460 --> 00:06:59,600
谁的模型就强

181
00:06:59,600 --> 00:07:01,320
谁的三哥占比高

182
00:07:01,320 --> 00:07:03,940
谁的模型就一定拉跨呀

183
00:07:03,940 --> 00:07:05,680
你以为只有科技圈

184
00:07:05,680 --> 00:07:07,520
那不要小看三哥的实力啊

185
00:07:07,520 --> 00:07:08,781
波音公司前四年

186
00:07:08,781 --> 00:07:23,781
硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬硬�

187
00:07:23,781 --> 00:07:25,261
到七五七到七八七

188
00:07:25,261 --> 00:07:27,261
各种型号是轮着出身

189
00:07:27,261 --> 00:07:29,421
深度调查指向了

190
00:07:29,421 --> 00:07:31,401
印度化的过度外包

191
00:07:31,401 --> 00:07:33,581
与内部的裙带关系

192
00:07:33,581 --> 00:07:35,061
有内部人士爆料

193
00:07:35,061 --> 00:07:37,581
说波音这几年招聘的工程师

194
00:07:37,581 --> 00:07:39,401
大部分都是印度裔的

195
00:07:39,401 --> 00:07:40,561
责任心不强

196
00:07:40,561 --> 00:07:41,801
质量意识也不强

197
00:07:41,801 --> 00:07:43,501
但是偶尔能催牛逼眼

198
00:07:43,501 --> 00:07:44,861
有报告指出

199
00:07:44,861 --> 00:07:47,781
一位负责多元化的印度裔副总裁

200
00:07:47,781 --> 00:07:49,801
利用职权在几年的时间

201
00:07:49,801 --> 00:07:51,741
把印度员工从一千多人

202
00:07:51,741 --> 00:07:53,601
扩充到了近两万人

203
00:07:53,601 --> 00:07:54,381
好家

204
00:07:54,381 --> 00:07:56,441
这是开新德里分店的

205
00:07:56,441 --> 00:08:00,001
马斯克的推特在完成收购之后

206
00:08:00,001 --> 00:08:02,961
第一件事就是砍掉所有的印度裔高管

207
00:08:02,961 --> 00:08:06,061
推特印度分公司原本200多人

208
00:08:06,061 --> 00:08:07,921
裁员后只剩12人

209
00:08:07,921 --> 00:08:11,681
马斯克甚至不惜赔偿5120万美元

210
00:08:11,681 --> 00:08:13,641
也要裁掉印度裔的高管

211
00:08:13,641 --> 00:08:16,501
据说马斯克还让印度工程师

212
00:08:16,501 --> 00:08:18,161
当场你给我手写代码

213
00:08:18,161 --> 00:08:21,541
结果混日子那些人直接就献出原形

214
00:08:21,541 --> 00:08:24,741
IBM印度裔CEO克里希纳

215
00:08:24,741 --> 00:08:29,081
三分钟就敲定了中国研发中心上千人的去留

216
00:08:29,081 --> 00:08:31,221
他是一边在中国大裁员

217
00:08:31,221 --> 00:08:33,121
一边在印度疯狂扩招

218
00:08:33,121 --> 00:08:35,441
只为把研发中心往自己老家搬

219
00:08:35,441 --> 00:08:38,181
之后IBM就一蹶不振

220
00:08:38,181 --> 00:08:40,701
直接远离了硅谷的科技核心圈

221
00:08:40,701 --> 00:08:42,561
根据不安全统计

222
00:08:42,561 --> 00:08:47,021
2024年至今有超过190名印度裔高管被裁车

223
00:08:47,021 --> 00:08:50,001
其中74位是CEO级别的

224
00:08:50,001 --> 00:08:51,121
这哪是裁员的

225
00:08:51,121 --> 00:08:53,281
这他妈是异常大清算

226
00:08:53,281 --> 00:08:55,521
那么回到五个

227
00:08:55,521 --> 00:08:58,481
那三哥CEO怎么会清洗三哥呢

228
00:08:58,481 --> 00:09:00,041
那个创造了Transformer

229
00:09:00,041 --> 00:09:01,441
曾经人类最强

230
00:09:01,441 --> 00:09:03,681
拥有最好科学家文化的公司

231
00:09:03,681 --> 00:09:06,781
就这么从AI竞赛里面掉队了

232
00:09:06,781 --> 00:09:10,121
如果AI竞赛最终是我们取得胜利

233
00:09:10,121 --> 00:09:14,221
那一定得给三哥颁一个一吨重的大奖章

234
00:09:14,221 --> 00:09:15,441
这不是嘲讽

235
00:09:15,441 --> 00:09:17,061
这是发自内心的感谢

236
00:09:17,061 --> 00:09:19,721
感谢三哥用一己之力

237
00:09:19,721 --> 00:09:22,361
把硅谷最强公司给祸祸死了

238
00:09:22,361 --> 00:09:22,901
哎呀

239
00:09:22,901 --> 00:09:23,141
说

240
00:09:23,141 --> 00:09:24,861
不过话说回来

241
00:09:24,861 --> 00:09:26,941
谷歌这事也给所有人提了个醒

242
00:09:26,941 --> 00:09:30,341
一个公司最可怕的不是神一样的对手

243
00:09:30,341 --> 00:09:31,901
而是猪一样的队友

244
00:09:31,901 --> 00:09:34,881
当一群PPT大师做上决策席

245
00:09:34,881 --> 00:09:37,601
当短期财报压倒长期创新

246
00:09:37,601 --> 00:09:39,981
当裙带关系取代能力导向

247
00:09:39,981 --> 00:09:41,141
再强的公司

248
00:09:41,141 --> 00:09:42,721
你也得量

249
00:09:42,721 --> 00:09:45,481
柯达的故事在谷歌重演了

250
00:09:45,481 --> 00:09:46,441
唯一不同的是

251
00:09:46,441 --> 00:09:49,161
这次毁掉谷歌的不是数码相机

252
00:09:49,161 --> 00:09:52,921
而是一群眼低手低的三哥高管

253
00:09:52,921 --> 00:09:54,461
谢谢啊

254
00:09:54,461 --> 00:09:55,441
好了兄弟们

255
00:09:55,441 --> 00:09:56,501
本期视频就到这了

256
00:09:56,501 --> 00:09:58,161
你们觉得谷歌还能翻身吗

257
00:09:58,161 --> 00:09:59,301
评论区咱们聊聊啊

258
00:09:59,301 --> 00:09:59,881
点赞关注

259
00:09:59,881 --> 00:10:00,721
咱们下期再见

260
00:10:00,721 --> 00:10:01,041
拜拜