【Meet Kevin】美光财报决定生死：芯片股周三的关键变局

1
00:00:01,480 --> 00:00:02,339
好的

2
00:00:02,339 --> 00:00:02,839
嘿

3
00:00:02,839 --> 00:00:04,500
大家欢迎回来

4
00:00:04,500 --> 00:00:06,080
我们必须谈谈美光科技

5
00:00:06,080 --> 00:00:09,169
因为美光科技的财报将于周三公布

6
00:00:09,169 --> 00:00:11,449
说实话，这可能决定市场走势

7
00:00:11,449 --> 00:00:14,089
这里可能并非所有人都持有美光科技股票

8
00:00:14,089 --> 00:00:18,120
或者你可能还没在视频中进行投资

9
00:00:18,120 --> 00:00:21,460
我们将对股票本身保持中立态度

10
00:00:21,460 --> 00:00:22,960
我只是要提供一些事实

11
00:00:22,960 --> 00:00:25,259
我们将用这些事实来理解

12
00:00:25,259 --> 00:00:27,219
整个行业正在发生什么

13
00:00:27,219 --> 00:00:28,719
以及这将如何影响整体

14
00:00:28,719 --> 00:00:29,018
嗯

15
00:00:29,018 --> 00:00:30,939
坦率说硬件叙事

16
00:00:30,939 --> 00:00:33,459
鉴于硬件显然是主导因素

17
00:00:33,459 --> 00:00:36,219
自四月初以来

18
00:00:36,219 --> 00:00:36,880
暂停

19
00:00:36,880 --> 00:00:37,920
如果你想这么称呼的话

20
00:00:37,920 --> 00:00:38,579
然后

21
00:00:38,579 --> 00:00:42,420
当然，此后我们一直在寻找全面反弹

22
00:00:42,420 --> 00:00:45,219
软件领域曾出现过几次起伏

23
00:00:45,219 --> 00:00:46,929
但并未持续发展

24
00:00:46,929 --> 00:00:49,070
上周金融板块表现突出

25
00:00:49,070 --> 00:00:50,149
效果显著

26
00:00:50,149 --> 00:00:51,390
特别是Robinhood

27
00:00:51,390 --> 00:00:54,369
但持续领先的仍是硬件股

28
00:00:54,369 --> 00:00:54,750
是的

29
00:00:54,750 --> 00:00:57,799
包括美光和铠侠

30
00:00:57,799 --> 00:00:58,179
当然

31
00:00:58,179 --> 00:01:01,590
但即使在这些内存股之外

32
00:01:01,590 --> 00:01:03,509
还有像Marvell这样的公司

33
00:01:03,509 --> 00:01:05,549
这是我们课程成员的最爱之一

34
00:01:05,549 --> 00:01:06,750
记得随时加入我们

35
00:01:06,750 --> 00:01:07,650
Kevin频道

36
00:01:07,650 --> 00:01:09,129
使用优惠码Pope

37
00:01:09,129 --> 00:01:13,849
你可以在Instagram或Twitter上看到原因

38
00:01:13,849 --> 00:01:14,329
不过

39
00:01:14,329 --> 00:01:14,829
嗯

40
00:01:14,829 --> 00:01:16,469
Marvell表现惊人

41
00:01:16,469 --> 00:01:17,549
我指的是历史高点

42
00:01:17,549 --> 00:01:18,829
股价现报30美元

43
00:01:18,829 --> 00:01:22,239
并且前一天盘中一度更高

44
00:01:22,239 --> 00:01:22,920
市场开盘时

45
00:01:22,920 --> 00:01:27,120
但暂且聚焦美光，讨论其财报预期

46
00:01:27,120 --> 00:01:29,030
这里为何表现强劲

47
00:01:29,030 --> 00:01:32,260
首先我们预计将超出指引

48
00:01:32,260 --> 00:01:34,739
超出美光预期约18%

49
00:01:34,739 --> 00:01:39,469
市场共识仅为13%的增幅

50
00:01:39,469 --> 00:01:41,629
但他们大概率会超出

51
00:01:41,629 --> 00:01:42,370
这很好

52
00:01:42,370 --> 00:01:46,590
你希望他们完全不遗漏任何组件

53
00:01:46,590 --> 00:01:49,289
否则你可能会拆掉博通

54
00:01:49,289 --> 00:01:52,769
博通整体表现都很出色

55
00:01:52,769 --> 00:01:57,540
但他们的定制芯片收入时间安排有些延迟，而博通大量生产asic芯片

56
00:01:57,540 --> 00:01:58,859
比如美光

57
00:01:58,859 --> 00:02:02,939
他们的定制芯片收入时间安排有所延迟

58
00:02:02,939 --> 00:02:04,200
而他们的

59
00:02:04,200 --> 00:02:09,219
下个季度的营收预期略低于市场预期

60
00:02:09,219 --> 00:02:10,900
股价仍未恢复

61
00:02:10,900 --> 00:02:13,219
财报后下跌了约20%

62
00:02:13,219 --> 00:02:16,969
如今仍比财报公布时下跌约14%

63
00:02:16,969 --> 00:02:19,129
尽管他们的财报相当

64
00:02:19,129 --> 00:02:19,770
惊人

65
00:02:19,770 --> 00:02:21,449
公司业务蓬勃发展

66
00:02:21,449 --> 00:02:23,050
对于长期投资者而言

67
00:02:23,050 --> 00:02:25,819
这为博通创造了买入机会

68
00:02:26,780 --> 00:02:32,800
即使是小幅不及预期也可能重创这些强势股

69
00:02:32,800 --> 00:02:33,419
如果这么说的话

70
00:02:33,419 --> 00:02:35,159
这些股票并不难推动

71
00:02:35,159 --> 00:02:35,990
言外之意

72
00:02:35,990 --> 00:02:40,129
我们预计会超出预期

73
00:02:40,129 --> 00:02:41,778
而市场本就预期会超出

74
00:02:41,778 --> 00:02:48,090
目前美光的最大驱动力是价格实现

75
00:02:48,090 --> 00:02:50,530
这一点非常重要

76
00:02:50,530 --> 00:02:53,060
当你考虑投资美光时

77
00:02:53,060 --> 00:02:57,840
我将直接引用他们上季度报告中的内容

78
00:02:57,840 --> 00:03:00,740
他们实现了

79
00:03:00,740 --> 00:03:01,460
呃

80
00:03:01,460 --> 00:03:07,740
同比DRAM供应增长40%，NAND供应增长30%或出货量

81
00:03:07,740 --> 00:03:08,240
应该说

82
00:03:08,240 --> 00:03:09,269
而非供应

83
00:03:09,269 --> 00:03:11,949
出货量同比大幅增长

84
00:03:11,949 --> 00:03:12,560
没错

85
00:03:12,560 --> 00:03:19,599
DRAM价格同比上涨约115%

86
00:03:19,599 --> 00:03:21,620
NAND价格同比上涨约100%

87
00:03:21,620 --> 00:03:27,789
换句话说，价格上涨快于出货量增长

88
00:03:27,789 --> 00:03:31,530
这在短缺环境中很常见

89
00:03:31,530 --> 00:03:32,909
这一点值得考虑

90
00:03:32,909 --> 00:03:35,110
如果你听到NAND和DRAM

91
00:03:35,110 --> 00:03:37,050
我不会深入细节

92
00:03:37,050 --> 00:03:39,050
我会给出一个非常基础的

93
00:03:39,050 --> 00:03:40,530
两者区别

94
00:03:40,530 --> 00:03:42,530
我认为两者的关键区别是

95
00:03:42,530 --> 00:03:43,689
我喜欢这样区分

96
00:03:43,689 --> 00:03:46,250
NAND在断电时仍能保持记忆

97
00:03:46,250 --> 00:03:48,789
当芯片关闭时

98
00:03:48,789 --> 00:03:50,550
电源关闭时

99
00:03:50,550 --> 00:03:52,039
这在什么情况下有用

100
00:03:52,039 --> 00:03:54,379
许多AI应用

101
00:03:54,379 --> 00:03:56,099
我不想不得不重新加载

102
00:03:56,099 --> 00:03:59,419
每次都要联系我的所有AI

103
00:03:59,419 --> 00:04:01,159
该死的东西又重生了

104
00:04:01,159 --> 00:04:03,060
我应该说重生或重启

105
00:04:03,060 --> 00:04:04,719
我就像在玩视频游戏

106
00:04:04,719 --> 00:04:06,319
尽管我身处现实世界

107
00:04:06,319 --> 00:04:06,959
大型多人在线游戏

108
00:04:06,959 --> 00:04:08,558
这里有很多NPC

109
00:04:08,558 --> 00:04:09,899
我应该这么说

110
00:04:09,899 --> 00:04:10,558
呃

111
00:04:10,558 --> 00:04:11,538
不过话说回来

112
00:04:11,538 --> 00:04:12,299
戏剧

113
00:04:12,299 --> 00:04:16,629
嗯，它一直非常强大

114
00:04:16,629 --> 00:04:19,769
但关闭时会清除你的记忆

115
00:04:19,769 --> 00:04:23,730
这非常简化地解释了区别

116
00:04:23,730 --> 00:04:26,629
当然他们有四个核心业务部门

117
00:04:26,629 --> 00:04:31,019
核心内存业务部门针对超大规模云

118
00:04:31,019 --> 00:04:32,920
显然这就是我们增长的地方

119
00:04:32,920 --> 00:04:35,100
目前在数据中心花费微缩技术

120
00:04:35,100 --> 00:04:38,339
当然还有传统数据中心业务

121
00:04:38,339 --> 00:04:39,199
想想存储

122
00:04:39,199 --> 00:04:45,230
我认为是戴尔机架内的原始组件

123
00:04:45,230 --> 00:04:47,449
谷歌网盘亚马逊云

124
00:04:47,449 --> 00:04:47,769
没错

125
00:04:47,769 --> 00:04:51,209
这种核心传统数据中心业务

126
00:04:51,209 --> 00:04:52,800
然后是移动业务

127
00:04:52,800 --> 00:04:55,839
汽车、工业和消费领域

128
00:04:55,839 --> 00:04:58,098
所以我认为移动电话

129
00:04:58,098 --> 00:05:00,699
内存芯片用于洗碗机或

130
00:05:00,699 --> 00:05:01,139
呃

131
00:05:01,139 --> 00:05:04,699
你的冰箱、洗衣机或汽车

132
00:05:04,699 --> 00:05:09,459
比如座椅设置等基础芯片

133
00:05:09,459 --> 00:05:11,259
这些领域

134
00:05:11,259 --> 00:05:14,000
最后两个是汽车、工业、消费

135
00:05:14,000 --> 00:05:15,500
这些被合并为两类

136
00:05:15,500 --> 00:05:17,180
我正在更详细地列举

137
00:05:17,180 --> 00:05:18,220
和移动业务

138
00:05:18,220 --> 00:05:20,819
这些实际上在衰退

139
00:05:20,819 --> 00:05:22,180
这非常有趣

140
00:05:22,180 --> 00:05:23,199
所以呃

141
00:05:23,199 --> 00:05:27,129
尽管各领域价格都在上涨

142
00:05:28,329 --> 00:05:31,988
移动和企业领域出货量下降

143
00:05:31,988 --> 00:05:33,418
这些其他类别

144
00:05:33,418 --> 00:05:35,619
所以真正推动芯片价格的是AI叙事

145
00:05:35,619 --> 00:05:38,418
现在正是AI推动芯片价格上涨

146
00:05:38,418 --> 00:05:41,288
而任何可能削弱

147
00:05:41,288 --> 00:05:47,670
希望美光股价曲线的指数增长持续

148
00:05:47,670 --> 00:05:50,339
将成为股票的重大风险因素

149
00:05:50,339 --> 00:05:53,259
尤其是近期股价已大幅上涨

150
00:05:53,259 --> 00:05:54,889
现在问题是

151
00:05:54,889 --> 00:05:57,310
尽管价格高达一千一百美元

152
00:05:57,310 --> 00:06:00,428
看起来已经疯涨到离谱的地步

153
00:06:00,428 --> 00:06:03,408
由于华尔街对这家公司的预测

154
00:06:03,408 --> 00:06:05,028
其实价格很便宜

155
00:06:05,028 --> 00:06:09,000
像瑞银将其估值为潜在翻倍

156
00:06:09,319 --> 00:06:11,259
我不认为他们错了

157
00:06:11,259 --> 00:06:13,139
这简直难以置信

158
00:06:13,139 --> 00:06:17,629
但目前我的KPEG估值是1.03

159
00:06:17,629 --> 00:06:20,509
这意味着可以轻松合理估值

160
00:06:20,509 --> 00:06:23,899
这可能更接近两倍于当前

161
00:06:23,899 --> 00:06:26,339
每股大概在两千五百美元左右

162
00:06:26,339 --> 00:06:30,569
但在这两者之间会有很大波动

163
00:06:30,569 --> 00:06:34,168
涨跌都有可能发生

164
00:06:34,168 --> 00:06:34,720
现在

165
00:06:34,783 --> 00:06:36,379
这简直太惊人了

166
00:06:36,379 --> 00:06:37,879
我来给你分解一下

167
00:06:37,879 --> 00:06:39,759
这就是构成它的因素

168
00:06:39,759 --> 00:06:42,220
所以你能理解为什么

169
00:06:42,220 --> 00:06:42,699
呃

170
00:06:42,699 --> 00:06:45,220
我们才会得出这样的估值

171
00:06:45,220 --> 00:06:49,480
这与今年末每股收益预期五

172
00:06:49,480 --> 00:06:50,908
九美元每股有关

173
00:06:50,908 --> 00:06:54,749
当我们看到这个并用当前股价除以五九

174
00:06:54,749 --> 00:06:55,608
我们可能会说

175
00:06:55,608 --> 00:06:56,389
等等

176
00:06:56,389 --> 00:06:56,788
哦

177
00:06:56,788 --> 00:06:57,829
其实这并不算太离谱

178
00:06:57,829 --> 00:07:00,178
这仅是十九点二倍市盈率

179
00:07:00,178 --> 00:07:00,798
是的

180
00:07:00,798 --> 00:07:03,350
这些人简直在印钞票

181
00:07:03,350 --> 00:07:04,689
尽管这个数据

182
00:07:04,689 --> 00:07:05,089
你知道的

183
00:07:05,089 --> 00:07:06,089
图表呈现火箭发射形态

184
00:07:06,089 --> 00:07:07,870
这完全是因为在印钞

185
00:07:07,870 --> 00:07:09,170
他们印钞速度极快

186
00:07:09,170 --> 00:07:10,689
他们仅以十九点

187
00:07:10,689 --> 00:07:12,839
二倍市盈率出售

188
00:07:12,839 --> 00:07:18,930
未来四年预计年增长十八%

189
00:07:18,930 --> 00:07:22,209
这接近一倍市盈率增长比

190
00:07:22,209 --> 00:07:23,329
现在

191
00:07:23,329 --> 00:07:24,550
如果看历史数据

192
00:07:24,550 --> 00:07:25,730
数字会有所不同

193
00:07:25,730 --> 00:07:27,788
但看这些前瞻预测

194
00:07:27,788 --> 00:07:28,269
是的

195
00:07:28,269 --> 00:07:30,488
预测实际上显示股票被低估

196
00:07:30,488 --> 00:07:31,309
这就是原因

197
00:07:31,309 --> 00:07:35,730
最敏感的因素是可能影响这些预测的变量

198
00:07:35,730 --> 00:07:37,529
现在我要快速强调一点

199
00:07:37,529 --> 00:07:39,199
这是完全免费的功能

200
00:07:39,199 --> 00:07:41,100
名为我们的Alpha电报服务

201
00:07:41,100 --> 00:07:42,839
如果你还没实际查看过

202
00:07:42,839 --> 00:07:44,600
我们上周刚发布

203
00:07:44,600 --> 00:07:45,660
而且非常惊人

204
00:07:45,660 --> 00:07:47,139
基本上来说

205
00:07:47,139 --> 00:07:47,579
它就是

206
00:07:47,579 --> 00:07:48,000
呃

207
00:07:48,000 --> 00:07:52,449
可以想象成CNBC或彭博社的锚点

208
00:07:52,449 --> 00:07:53,269
你知道他们是怎么样的

209
00:07:53,269 --> 00:07:53,428
哦

210
00:07:53,428 --> 00:07:54,869
这信息正在实时传输

211
00:07:54,869 --> 00:07:56,209
这信息正在通过电报传来

212
00:07:56,209 --> 00:07:58,029
现在你可以免费获取

213
00:07:58,029 --> 00:07:59,730
在Meet Kevin应用内

214
00:07:59,730 --> 00:08:03,550
所以如果你在苹果或安卓应用商店下载Meet Kevin应用

215
00:08:03,550 --> 00:08:04,810
就可以使用这个功能

216
00:08:04,810 --> 00:08:05,889
注册后

217
00:08:05,889 --> 00:08:08,329
之后也可以使用桌面版

218
00:08:08,329 --> 00:08:08,970
如果你想的话

219
00:08:08,970 --> 00:08:09,430
呃

220
00:08:09,430 --> 00:08:09,730
你知道的

221
00:08:09,730 --> 00:08:11,149
你首先要应用内注册

222
00:08:11,149 --> 00:08:14,279
Just App MeetKevin.com是桌面版

223
00:08:14,279 --> 00:08:16,879
但必须先在应用内创建账户

224
00:08:16,879 --> 00:08:20,160
可以在桌面、安卓、苹果设备使用电报服务

225
00:08:20,160 --> 00:08:21,839
目前完全免费

226
00:08:21,839 --> 00:08:23,319
但挺酷的因为

227
00:08:23,319 --> 00:08:24,639
你会收到这些即时更新

228
00:08:24,639 --> 00:08:25,579
比如特朗普前往营地

229
00:08:25,579 --> 00:08:28,000
大卫在伊朗谈判中陷入僵局之类的

230
00:08:28,000 --> 00:08:29,199
这些都会及时推送

231
00:08:29,199 --> 00:08:30,240
这刚推出不久

232
00:08:30,240 --> 00:08:31,120
处于超级测试阶段

233
00:08:31,120 --> 00:08:32,600
我们刚发布这个并提到Kevin

234
00:08:32,600 --> 00:08:36,178
但我认为真的很酷因为我也能

235
00:08:36,178 --> 00:08:36,938
呃

236
00:08:36,938 --> 00:08:38,399
接收电报警报

237
00:08:38,399 --> 00:08:41,399
然后我会在下方添加Meet Kevin的评论

238
00:08:41,399 --> 00:08:42,960
这清晰明了

239
00:08:42,960 --> 00:08:43,600
可以查看一下

240
00:08:43,600 --> 00:08:45,500
就在Alpha Wire标签下方

241
00:08:45,500 --> 00:08:47,440
在Meet Kevin应用内

242
00:08:47,440 --> 00:08:51,750
我们还有一个终端按钮可切换房产和股票

243
00:08:51,750 --> 00:08:54,190
因为我们有股票AI工具

244
00:08:54,190 --> 00:08:55,509
现在已自定义开发

245
00:08:55,509 --> 00:08:57,339
还有房地产AI工具

246
00:08:57,339 --> 00:09:00,058
估值AI服务月底推出

247
00:09:00,058 --> 00:09:01,239
非常令人兴奋

248
00:09:01,239 --> 00:09:04,089
我们正在全力开发软件

249
00:09:04,089 --> 00:09:06,399
我们兴奋不已

250
00:09:06,399 --> 00:09:08,000
目前免费提供

251
00:09:08,000 --> 00:09:09,080
所以我们对此感到非常兴奋

252
00:09:09,080 --> 00:09:09,759
你可以尝试使用它

253
00:09:09,759 --> 00:09:12,710
至少在未来几个月内

254
00:09:12,710 --> 00:09:13,990
完全免费

255
00:09:13,990 --> 00:09:15,909
所以我认为你应该充分利用它

256
00:09:15,909 --> 00:09:17,580
为什么不试试看呢

257
00:09:17,580 --> 00:09:18,419
给我们一些反馈

258
00:09:18,419 --> 00:09:19,100
告诉我们你的看法

259
00:09:19,100 --> 00:09:19,639
不过暂且不提

260
00:09:19,639 --> 00:09:21,940
让我们更多关注微米级产品

261
00:09:21,940 --> 00:09:26,279
这些是每股收益和增长预测

262
00:09:26,279 --> 00:09:28,379
这里有没有什么警示信号

263
00:09:28,379 --> 00:09:31,460
可能影响股价的因素是平均售价下滑

264
00:09:31,460 --> 00:09:32,720
订单增长放缓

265
00:09:32,720 --> 00:09:34,509
显然未能达成目标

266
00:09:34,509 --> 00:09:35,350
但对我来说

267
00:09:35,350 --> 00:09:37,690
这是一家新冠时代的卫生纸供应商

268
00:09:37,690 --> 00:09:38,929
他们正在销售卫生纸

269
00:09:38,929 --> 00:09:40,210
当所有人都需要卫生纸时

270
00:09:40,210 --> 00:09:41,610
卫生纸供应严重不足

271
00:09:41,610 --> 00:09:44,269
这不是对公司的侮辱

272
00:09:44,269 --> 00:09:46,089
公司自己也清楚这一点

273
00:09:46,089 --> 00:09:48,448
他们在文件中明确警告

274
00:09:48,448 --> 00:09:52,389
目前大部分现金流来自价格提升

275
00:09:52,389 --> 00:09:54,450
并非因为增加了产品交付

276
00:09:54,450 --> 00:09:56,129
他们意识到这一点

277
00:09:56,129 --> 00:09:58,789
如果你仔细看季度环比数据

278
00:09:58,789 --> 00:10:00,169
就能看到数据有多极端

279
00:10:00,169 --> 00:10:02,659
所以我切换到季度环比分析

280
00:10:02,659 --> 00:10:06,179
我希望能在演播室共享屏幕展示给你们

281
00:10:06,179 --> 00:10:08,179
如果你是课程会员

282
00:10:08,179 --> 00:10:08,799
你可以查看

283
00:10:08,799 --> 00:10:09,919
在股票标签下

284
00:10:09,919 --> 00:10:13,240
所有数据都位于股票标签下

285
00:10:13,240 --> 00:10:16,460
但我想让你思考的是下一季度

286
00:10:16,460 --> 00:10:18,639
我甚至不确定是否截图了

287
00:10:18,639 --> 00:10:20,460
让我截图确认内容

288
00:10:20,460 --> 00:10:20,940
不过暂且不提

289
00:10:20,940 --> 00:10:22,139
这是季度报告中的内容

290
00:10:22,139 --> 00:10:22,580
哦对啊

291
00:10:22,580 --> 00:10:22,659
在这里

292
00:10:22,659 --> 00:10:23,559
好的太好了

293
00:10:23,559 --> 00:10:24,259
哦好的

294
00:10:24,259 --> 00:10:24,899
完美没错

295
00:10:24,899 --> 00:10:25,429
就在这里

296
00:10:25,429 --> 00:10:30,700
我们的DRAM产品销售额增长了74%

297
00:10:30,700 --> 00:10:33,799
但价格提升了约65%

298
00:10:33,799 --> 00:10:35,639
这是季度环比数据

299
00:10:35,639 --> 00:10:37,139
在三个月内

300
00:10:37,139 --> 00:10:41,440
他们在三个月内连续提价65%

301
00:10:41,440 --> 00:10:43,519
这简直是严重的短缺

302
00:10:43,519 --> 00:10:47,519
这其中部分是我们看到的claude之后的情况

303
00:10:48,210 --> 00:10:49,509
协作真的变得疯狂了

304
00:10:49,509 --> 00:10:50,649
我不该说协作

305
00:10:50,649 --> 00:10:52,240
嗯随便你怎么看

306
00:10:52,240 --> 00:10:52,639
我的意思是

307
00:10:52,639 --> 00:10:53,159
claude代码

308
00:10:53,159 --> 00:10:53,980
就是这样在之后

309
00:10:53,980 --> 00:10:55,639
claude代码真的失控了

310
00:10:55,639 --> 00:10:57,639
在q1季度

311
00:10:57,679 --> 00:10:58,720
所有产品的价格

312
00:10:58,720 --> 00:10:59,840
HBM内存

313
00:10:59,840 --> 00:11:00,559
呃

314
00:11:00,559 --> 00:11:00,879
你知道的

315
00:11:00,879 --> 00:11:01,659
高带宽内存

316
00:11:01,659 --> 00:11:06,700
但英伟达的h100在q1也出现了大幅上涨

317
00:11:06,700 --> 00:11:09,559
但这些价格已经开始回落

318
00:11:09,559 --> 00:11:11,820
这也是我们最担心的问题之一

319
00:11:11,820 --> 00:11:12,580
对于美光来说

320
00:11:12,580 --> 00:11:15,029
我们正在追踪h100价格

321
00:11:15,029 --> 00:11:16,809
但价格仍居高不下

322
00:11:16,809 --> 00:11:20,470
不过已大幅低于第一季度峰值

323
00:11:20,470 --> 00:11:25,090
因此他们提到的价格增长可能不会持续

324
00:11:25,090 --> 00:11:25,509
但我说

325
00:11:25,509 --> 00:11:26,039
听这个

326
00:11:26,039 --> 00:11:28,080
他们提高了DRAM价格

327
00:11:28,080 --> 00:11:31,620
记得那种一直正常工作但断电后会丢失数据的

328
00:11:31,620 --> 00:11:33,740
价格涨了65%

329
00:11:33,740 --> 00:11:36,460
而实际出货量仅增长约7-9%

330
00:11:36,460 --> 00:11:41,450
所以这太好了

331
00:11:41,450 --> 00:11:43,629
你知道的

332
00:11:43,629 --> 00:11:44,009
出货量仍在增长

333
00:11:44,009 --> 00:11:45,909
你知道的

334
00:11:45,909 --> 00:11:46,429
大约三二到三六%

335
00:11:46,429 --> 00:11:48,230
但大部分收入来自涨价

336
00:11:48,230 --> 00:11:51,110
因为他们可以

337
00:11:51,110 --> 00:11:52,389
有点像

338
00:11:52,389 --> 00:11:53,210
哇

339
00:11:53,210 --> 00:11:53,529
好吧

340
00:11:53,529 --> 00:11:53,830
行

341
00:11:53,830 --> 00:11:54,730
我们试着提价

342
00:11:54,730 --> 00:11:55,009
哇

343
00:11:55,009 --> 00:11:56,090
人们仍在买单

344
00:11:56,090 --> 00:11:56,909
我们再试一次提价

345
00:11:56,909 --> 00:11:57,190
哇

346
00:11:57,190 --> 00:11:58,429
人们依然在支付

347
00:11:58,429 --> 00:11:59,789
这就是他们在做的

348
00:11:59,789 --> 00:12:01,269
他们正在榨取利润

349
00:12:01,269 --> 00:12:03,409
但这不会永远持续

350
00:12:03,409 --> 00:12:04,990
我们确实知道

351
00:12:04,990 --> 00:12:07,730
如果你知道peepee使用过量过久

352
00:12:07,730 --> 00:12:09,009
你知道有人说是四个小时

353
00:12:09,009 --> 00:12:10,399
你需要去看医生

354
00:12:10,399 --> 00:12:10,879
好的

355
00:12:10,879 --> 00:12:11,259
总之

356
00:12:11,259 --> 00:12:11,639
呃

357
00:12:11,639 --> 00:12:15,759
实际上同样的情况发生在纳米产品和产品上八十个百分点

358
00:12:15,759 --> 00:12:19,070
但其中百分之七十五是因为价格问题

359
00:12:19,070 --> 00:12:20,169
数量在增加

360
00:12:20,169 --> 00:12:23,409
但大部分增长其实是价格驱动

361
00:12:23,409 --> 00:12:24,049
好的

362
00:12:24,049 --> 00:12:24,210
是的

363
00:12:24,210 --> 00:12:26,539
所以我确实有这个修正版本

364
00:12:26,539 --> 00:12:28,220
已经在股票上更新了

365
00:12:28,220 --> 00:12:28,620
呃

366
00:12:28,620 --> 00:12:29,419
终端端

367
00:12:29,419 --> 00:12:29,799
好的

368
00:12:29,799 --> 00:12:30,120
很好

369
00:12:30,120 --> 00:12:30,840
所以呃

370
00:12:30,840 --> 00:12:32,820
现在让我们继续讨论

371
00:12:32,820 --> 00:12:34,740
更广泛的潜在风险因素

372
00:12:34,740 --> 00:12:37,100
公司任何成本削减

373
00:12:37,100 --> 00:12:41,500
或供应增加都是此类公司的主要风险

374
00:12:41,500 --> 00:12:43,779
潜在的不利因素包括

375
00:12:43,779 --> 00:12:46,299
现在meta微软等公司

376
00:12:46,299 --> 00:12:51,169
和沃尔玛等企业正在减少这种token消耗文化

377
00:12:51,169 --> 00:12:53,389
我们上周刚分析了meta

378
00:12:53,389 --> 00:12:56,220
告知员工即将限制AI使用

379
00:12:56,220 --> 00:12:59,080
在观察到token消耗激增后

380
00:12:59,080 --> 00:12:59,779
公平地说

381
00:12:59,779 --> 00:13:01,279
他们有点疯狂

382
00:13:01,279 --> 00:13:02,500
他们提出

383
00:13:02,500 --> 00:13:05,899
设立排行榜看谁消耗的token最多

384
00:13:05,899 --> 00:13:06,779
和呃

385
00:13:06,779 --> 00:13:07,240
你知道的

386
00:13:07,240 --> 00:13:09,039
基于token消耗

387
00:13:09,039 --> 00:13:11,179
人们实际上在循环使用代理

388
00:13:11,179 --> 00:13:14,600
几乎不做实际工作

389
00:13:14,600 --> 00:13:14,980
呃

390
00:13:14,980 --> 00:13:16,460
但最大化token使用量

391
00:13:16,460 --> 00:13:17,279
这很愚蠢

392
00:13:17,279 --> 00:13:20,360
meta正在终止这种做法沃尔玛也在行动微软

393
00:13:20,360 --> 00:13:23,710
正考虑从open ai和claude切换

394
00:13:23,710 --> 00:13:24,950
主要是聊天功能

395
00:13:24,950 --> 00:13:27,970
微软团队采用GPT-4o级别

396
00:13:27,970 --> 00:13:31,110
并考虑转向DeepSeek V4

397
00:13:31,110 --> 00:13:32,669
这简直是侮辱

398
00:13:32,669 --> 00:13:35,429
因为这意味着从美国LLM供应商转向

399
00:13:35,429 --> 00:13:38,850
类似模仿且效率更高的版本

400
00:13:38,850 --> 00:13:39,750
深海

401
00:13:39,750 --> 00:13:41,809
你知道我们非常期待深海探索

402
00:13:41,809 --> 00:13:46,379
基本上通过训练开放AI的输出来训练他们的版本

403
00:13:46,379 --> 00:13:49,659
因此绕过了OpenAI的所有训练过程

404
00:13:49,659 --> 00:13:51,000
这相当聪明

405
00:13:51,000 --> 00:13:52,589
但也非常邪恶

406
00:13:52,589 --> 00:13:53,729
但当你切换模型时

407
00:13:53,729 --> 00:13:54,869
成本会骤降

408
00:13:54,869 --> 00:13:56,048
如果你使用Sonnet

409
00:13:56,048 --> 00:14:03,059
实际上能节省约95%的Opus在Claude上的成本

410
00:14:03,059 --> 00:14:05,649
所以你使用Claude Opus

411
00:14:05,649 --> 00:14:07,870
你知道这需要20x的成本

412
00:14:07,870 --> 00:14:08,980
而Sonic的成本是多少

413
00:14:08,980 --> 00:14:10,279
如果你使用Kimi

414
00:14:10,279 --> 00:14:12,700
这是Moonshot AI的产品

415
00:14:12,700 --> 00:14:14,120
这是另一家中国初创公司

416
00:14:14,120 --> 00:14:15,019
仅需花费1美元

417
00:14:15,019 --> 00:14:16,750
另一个20美分

418
00:14:16,750 --> 00:14:18,789
因此实现95%的成本降低

419
00:14:18,789 --> 00:14:21,129
成本降低确实是关键所在

420
00:14:21,129 --> 00:14:22,639
随着我们看到的越来越多

421
00:14:22,639 --> 00:14:27,179
芯片制造商的定价能力反而减弱

422
00:14:27,179 --> 00:14:29,860
但这都是下游效应

423
00:14:29,860 --> 00:14:31,899
这就是我们关注他们指引的原因

424
00:14:31,899 --> 00:14:34,480
因为我们认为他们能给出强劲指引

425
00:14:34,480 --> 00:14:36,610
至少持续一年以上

426
00:14:36,610 --> 00:14:37,970
彭博情报指出

427
00:14:37,970 --> 00:14:40,419
他们将在本十年末保持定价权

428
00:14:40,419 --> 00:14:44,419
我怀疑他们声称在2029至2030年间供应受限

429
00:14:44,419 --> 00:14:46,309
从现在到2029-2030年

430
00:14:46,309 --> 00:14:49,090
他们预计2027年价格将上涨

431
00:14:49,090 --> 00:14:51,509
我同意2027年的情况

432
00:14:51,509 --> 00:14:52,549
进入明年后

433
00:14:52,549 --> 00:14:53,490
我赞同这一点

434
00:14:53,490 --> 00:14:54,730
直至本十年末

435
00:14:54,730 --> 00:14:55,929
我不认同这个观点

436
00:14:55,929 --> 00:14:57,710
我将解释为何不认同

437
00:14:57,710 --> 00:15:01,419
这与即将讨论的供应扩张有关

438
00:15:01,419 --> 00:15:02,539
但呃

439
00:15:02,539 --> 00:15:04,559
模型效率正在缓慢提升

440
00:15:04,559 --> 00:15:06,879
尽管模型使用量激增

441
00:15:06,879 --> 00:15:08,840
二级导数的令牌使用

442
00:15:08,840 --> 00:15:10,690
我们已知正在下降

443
00:15:10,690 --> 00:15:12,490
令牌使用量

444
00:15:12,490 --> 00:15:14,149
以谷歌为例

445
00:15:14,149 --> 00:15:17,789
I/O事件显示从2024到2025年

446
00:15:17,789 --> 00:15:19,720
令牌使用量增长50倍

447
00:15:19,720 --> 00:15:23,460
现在增长至约6.9倍

448
00:15:23,460 --> 00:15:26,559
增长率可能继续放缓

449
00:15:26,559 --> 00:15:29,529
随着令牌使用接近峰值

450
00:15:29,529 --> 00:15:31,169
我们将看到更多下降

451
00:15:31,169 --> 00:15:35,980
最终，这些定价能力将在芯片公司中逐渐减弱

452
00:15:35,980 --> 00:15:37,440
这对所有芯片公司都是警示

453
00:15:37,440 --> 00:15:38,240
我们很清楚这一点

454
00:15:38,240 --> 00:15:39,080
但目前来看

455
00:15:39,080 --> 00:15:40,399
形势看起来非常乐观

456
00:15:40,399 --> 00:15:43,899
公司正在努力扩大产能以

457
00:15:43,899 --> 00:15:45,839
因为他们是自家船只的主要制造商

458
00:15:45,839 --> 00:15:46,678
同时他们使用

459
00:15:46,678 --> 00:15:47,119
呃

460
00:15:47,119 --> 00:15:47,578
第三方供应商

461
00:15:47,578 --> 00:15:49,450
目前他们正在大量生产

462
00:15:49,450 --> 00:15:50,730
他们正在偿还债务

463
00:15:50,730 --> 00:15:52,210
他们的资产负债表状况良好

464
00:15:52,210 --> 00:15:55,110
现在他们有足够的现金偿还所有债务

465
00:15:55,110 --> 00:15:57,230
并且正在积极偿还债务

466
00:15:57,230 --> 00:15:58,690
我认为这是一个非常明智的举措

467
00:15:58,690 --> 00:16:02,470
他们显然清楚这一点，尽管在偿还债务

468
00:16:02,470 --> 00:16:03,970
他们仍在大量投入

469
00:16:03,970 --> 00:16:05,620
预计资本支出将达两

470
00:16:05,620 --> 00:16:06,860
五亿美元

471
00:16:06,860 --> 00:16:10,539
但第一季度已支出超过十一亿美元

472
00:16:10,539 --> 00:16:11,240
所以你知道

473
00:16:11,240 --> 00:16:13,460
我认为他们已经花了很多

474
00:16:13,460 --> 00:16:17,120
但第一季度已耗去近半年度预算

475
00:16:17,120 --> 00:16:19,299
但也许这只是时间安排问题

476
00:16:19,299 --> 00:16:20,019
不过无论如何

477
00:16:20,019 --> 00:16:22,340
他们大幅扩产的主要原因之一

478
00:16:22,340 --> 00:16:24,460
这也是我认为他们无法撑到十年末期

479
00:16:24,460 --> 00:16:26,158
这与定价能力有关

480
00:16:26,158 --> 00:16:29,578
他们从纽约州获得约五十五亿美元补贴

481
00:16:29,578 --> 00:16:32,119
再加上拜登时代的芯片补贴三

482
00:16:32,119 --> 00:16:34,549
五%

483
00:16:34,549 --> 00:16:35,750
芯片法案现在

484
00:16:35,750 --> 00:16:37,970
特朗普完全批评了芯片补贴

485
00:16:37,970 --> 00:16:38,610
芯片补贴

486
00:16:38,610 --> 00:16:43,190
但背后他们正积极运营办公室管理

487
00:16:43,190 --> 00:16:47,110
如何通过芯片法案分配税收抵免

488
00:16:47,110 --> 00:16:50,078
特朗普的反对只是表面文章

489
00:16:50,078 --> 00:16:54,379
但背后他正以特朗普式风格选择赢家

490
00:16:54,379 --> 00:16:54,999
我们

491
00:16:54,999 --> 00:16:56,979
我们大致了解特朗普的运作方式

492
00:16:56,979 --> 00:16:57,678
这不是什么秘密

493
00:16:57,678 --> 00:17:00,460
只是如实陈述事实

494
00:17:00,460 --> 00:17:02,000
你知道的

495
00:17:02,080 --> 00:17:02,500
这

496
00:17:02,500 --> 00:17:03,879
这几乎像是在说

497
00:17:03,879 --> 00:17:05,920
我们能否在梅凯文视频中避免

498
00:17:05,920 --> 00:17:07,539
这里有个球传给B队

499
00:17:07,539 --> 00:17:09,029
公平地说

500
00:17:09,029 --> 00:17:10,930
价格会随时间上涨

501
00:17:10,930 --> 00:17:12,839
我们再次提高了价格

502
00:17:12,839 --> 00:17:15,759
月底我们还将迎来新一轮大幅涨价

503
00:17:15,759 --> 00:17:17,920
因为我们即将推出更多新产品

504
00:17:17,920 --> 00:17:20,160
我们在研发上投入了很多资源

505
00:17:20,160 --> 00:17:22,240
所以我们正在努力回馈社区

506
00:17:22,240 --> 00:17:22,960
嗯

507
00:17:22,960 --> 00:17:25,519
对于不想支付一分钱的你们

508
00:17:25,519 --> 00:17:27,619
使用这个Alpha Wire服务

509
00:17:27,619 --> 00:17:29,380
我强烈推荐你们试试

510
00:17:29,380 --> 00:17:33,579
Alpha Wire服务非常强大，评论区告诉我们你的看法

511
00:17:33,579 --> 00:17:34,299
你觉得怎么样

512
00:17:34,299 --> 00:17:35,079
如果你不喜欢

513
00:17:35,079 --> 00:17:36,500
卸载应用即可

514
00:17:36,500 --> 00:17:37,849
但说实话

515
00:17:37,849 --> 00:17:40,190
我不会放弃这个免费资源

516
00:17:40,190 --> 00:17:41,388
就像这样

517
00:17:41,388 --> 00:17:42,929
如果我不觉得它很棒

518
00:17:42,929 --> 00:17:43,868
你会说

519
00:17:43,868 --> 00:17:44,788
天啊凯文

520
00:17:44,788 --> 00:17:45,108
是的

521
00:17:45,108 --> 00:17:46,588
这确实不错

522
00:17:46,588 --> 00:17:48,709
真的太感谢了

523
00:17:48,709 --> 00:17:49,469
嗯好的

524
00:17:49,469 --> 00:17:52,069
供应扩张的问题在于

525
00:17:52,069 --> 00:17:53,529
我这里会快速说明

526
00:17:53,529 --> 00:17:55,509
新加坡高带宽内存扩产

527
00:17:55,509 --> 00:17:58,359
预计2027年投产

528
00:17:58,359 --> 00:17:59,799
这基本上是他们在说

529
00:17:59,799 --> 00:18:02,960
他们正在大幅扩展芯片封装产能

530
00:18:02,960 --> 00:18:05,920
2027年高带宽内存芯片组装

531
00:18:05,920 --> 00:18:09,160
我们的新加坡工厂将大幅增加

532
00:18:09,160 --> 00:18:09,839
HBM

533
00:18:09,839 --> 00:18:12,180
这将提升市场供应量

534
00:18:12,180 --> 00:18:12,579
哦太棒了

535
00:18:12,579 --> 00:18:13,220
我刚放大了

536
00:18:13,220 --> 00:18:14,839
我没想到自己缩放到了最远

537
00:18:14,839 --> 00:18:15,660
哦这挺有意思的

538
00:18:15,660 --> 00:18:16,019
我可以

539
00:18:16,019 --> 00:18:18,380
其实可以用手机控制缩放，挺酷的

540
00:18:18,380 --> 00:18:20,740
或者用手表控制，挺酷的

541
00:18:20,740 --> 00:18:23,240
但缺点是不小心碰到

542
00:18:23,240 --> 00:18:24,839
会搞乱你的视角角度

543
00:18:24,839 --> 00:18:26,319
希望你没注意到啊

544
00:18:26,319 --> 00:18:27,480
你看我的裤子

545
00:18:28,680 --> 00:18:30,529
什么裤子啊，算了

546
00:18:30,529 --> 00:18:31,609
所以啊

547
00:18:31,609 --> 00:18:33,130
结果还行

548
00:18:33,130 --> 00:18:33,779
嗯

549
00:18:33,779 --> 00:18:35,759
2022年9月

550
00:18:35,759 --> 00:18:38,339
他们启动了爱达荷州博伊西工厂

551
00:18:38,339 --> 00:18:41,160
他们实际上在2023年破土动工

552
00:18:41,160 --> 00:18:43,099
猜什么时候能上线

553
00:18:43,099 --> 00:18:46,180
预计2027年中完成开发，耗时四年

554
00:18:46,180 --> 00:18:49,099
这意味着博伊西爱达荷工厂即将投产

555
00:18:49,099 --> 00:18:52,720
同时他们的新加坡工厂正在扩大产能

556
00:18:52,720 --> 00:18:54,849
供应量大幅增加，值得注意

557
00:18:54,849 --> 00:18:56,390
这只是美光科技

558
00:18:56,390 --> 00:18:57,670
作为内存行业的领头羊

559
00:18:57,670 --> 00:18:59,309
实际生产由三星负责

560
00:18:59,309 --> 00:19:00,990
然后还有SK海力士

561
00:19:00,990 --> 00:19:03,849
三星市占率约38%

562
00:19:03,849 --> 00:19:05,660
SK海力士29%

563
00:19:05,660 --> 00:19:07,799
美光目前仅约22%

564
00:19:07,799 --> 00:19:09,200
这点需要记住

565
00:19:09,200 --> 00:19:11,480
他们现在是行业巨头

566
00:19:11,480 --> 00:19:13,059
如果你看

567
00:19:13,059 --> 00:19:15,730
还有新的工厂在建设

568
00:19:15,730 --> 00:19:20,309
他们今年在爱达荷州启动了第二座工厂建设

569
00:19:20,309 --> 00:19:21,089
目前来说

570
00:19:21,089 --> 00:19:23,849
预计2028年底投产

571
00:19:23,849 --> 00:19:27,890
从2027年中到之后十八个月

572
00:19:27,890 --> 00:19:29,930
2027年中到2028年底

573
00:19:29,930 --> 00:19:31,430
将有三座工厂

574
00:19:31,430 --> 00:19:35,750
仅美光投产就试图实现这一目标

575
00:19:35,750 --> 00:19:37,609
现在价格正在上涨

576
00:19:37,609 --> 00:19:41,369
看来他们开始加速建设这些工厂

577
00:19:41,369 --> 00:19:43,049
第一个用了四年时间

578
00:19:43,049 --> 00:19:45,269
下一个开发周期两年半

579
00:19:45,269 --> 00:19:46,390
可能规模更小

580
00:19:46,390 --> 00:19:47,329
但在我看来

581
00:19:47,329 --> 00:19:48,009
如果我是他们

582
00:19:48,009 --> 00:19:48,920
我会这么做

583
00:19:48,920 --> 00:19:49,240
嘿

584
00:19:49,240 --> 00:19:50,539
现在价格非常高

585
00:19:50,539 --> 00:19:52,119
尽快建设这些工厂

586
00:19:52,119 --> 00:19:53,720
趁价格有利可图

587
00:19:53,720 --> 00:19:54,400
太阳当头时抢收

588
00:19:54,400 --> 00:19:55,659
这就是他们还债的原因

589
00:19:55,659 --> 00:19:57,338
他们同时非常精明

590
00:19:57,338 --> 00:19:57,979
他们正在还债

591
00:19:57,979 --> 00:20:00,459
用现金流购置设备

592
00:20:00,459 --> 00:20:00,939
非常

593
00:20:00,939 --> 00:20:01,699
非常聪明

594
00:20:01,699 --> 00:20:02,138
现在

595
00:20:02,138 --> 00:20:02,898
短期内

596
00:20:02,898 --> 00:20:08,338
苹果警告内存价格上涨将推高产品成本

597
00:20:08,338 --> 00:20:09,950
这不可避免

598
00:20:09,950 --> 00:20:11,349
我同意明年

599
00:20:11,349 --> 00:20:12,509
这大概无法避免

600
00:20:12,509 --> 00:20:15,690
可能导致苹果销量下滑

601
00:20:15,690 --> 00:20:18,349
在我看来，这将导致股价下跌

602
00:20:18,349 --> 00:20:21,089
实际上这将成为苹果的买入机会

603
00:20:21,089 --> 00:20:23,750
我认为苹果目前估值大致合理

604
00:20:23,750 --> 00:20:25,019
上涨空间并不大

605
00:20:25,019 --> 00:20:26,619
我希望它们能更便宜一些

606
00:20:26,619 --> 00:20:27,059
说实话

607
00:20:27,059 --> 00:20:32,549
我希望能趁苹果推出AI产品前买入

608
00:20:32,549 --> 00:20:34,650
我想在他们搞清楚之前就持有

609
00:20:34,650 --> 00:20:35,869
也许他们永远搞不懂

610
00:20:35,869 --> 00:20:36,750
这确实存在风险

611
00:20:36,750 --> 00:20:37,750
但我现在并不持有

612
00:20:37,750 --> 00:20:38,569
不过话说回来

613
00:20:38,569 --> 00:20:41,839
我们需要关注这些设施的扩建计划

614
00:20:41,839 --> 00:20:43,759
他们的DNA并没有明显变动

615
00:20:43,759 --> 00:20:44,339
这说明

616
00:20:44,339 --> 00:20:44,799
你知道的

617
00:20:44,799 --> 00:20:47,700
这表明他们能很好扩展现有业务

618
00:20:47,700 --> 00:20:52,839
他们警告现金流高度依赖售价

619
00:20:52,839 --> 00:20:54,470
而售价极其波动

620
00:20:54,470 --> 00:20:57,450
这就是他们现在处于月球周期的原因

621
00:20:57,450 --> 00:20:59,159
因为一切都在飙升

622
00:20:59,159 --> 00:21:02,680
但价格涨幅的持续性肯定会减弱

623
00:21:02,680 --> 00:21:03,880
这取决于

624
00:21:03,880 --> 00:21:05,440
产能是否先提升

625
00:21:05,440 --> 00:21:06,779
以降低价格

626
00:21:06,779 --> 00:21:08,299
还是需求先放缓

627
00:21:08,299 --> 00:21:09,599
先发生还是更糟

628
00:21:09,599 --> 00:21:10,319
两者都发生

629
00:21:10,319 --> 00:21:13,599
当供需同时变化时

630
00:21:13,599 --> 00:21:17,878
这只股票将暴跌

631
00:21:17,878 --> 00:21:19,098
这会很糟糕

632
00:21:19,098 --> 00:21:21,809
我不想持有十年

633
00:21:21,809 --> 00:21:25,230
或许可以持有明年

634
00:21:25,230 --> 00:21:27,349
尤其是财报后暴跌时

635
00:21:27,349 --> 00:21:29,569
他们有53000名员工

636
00:21:29,569 --> 00:21:31,549
这确实很 impressive，比我预期的多

637
00:21:31,549 --> 00:21:35,638
核心结论是硬件端尚未接近终点

638
00:21:35,638 --> 00:21:37,838
财报后可能出现波动

639
00:21:37,838 --> 00:21:39,359
就像博通的情况一样

640
00:21:39,359 --> 00:21:41,519
但说实话大概会是买入时机

641
00:21:41,519 --> 00:21:43,720
现在到明年之间的回调期

642
00:21:43,720 --> 00:21:46,279
随着更多芯片产能上线

643
00:21:46,279 --> 00:21:48,839
只要超大规模数据中心

644
00:21:48,839 --> 00:21:50,259
还没启动资本开支

645
00:21:50,259 --> 00:21:52,089
目前确实没有迹象

646
00:21:52,089 --> 00:21:56,170
但我认为2027年可能是转折点

647
00:21:56,170 --> 00:21:58,079
换句话说仍有时间窗口

648
00:21:58,079 --> 00:22:00,299
公司当前估值非常合理

649
00:22:00,299 --> 00:22:01,859
基于华尔街预测

650
00:22:01,859 --> 00:22:04,460
资产负债表实力非常强劲

651
00:22:04,460 --> 00:22:06,140
这家公司是个金矿

652
00:22:06,140 --> 00:22:08,779
它也是一个非常耐用的公司

653
00:22:08,779 --> 00:22:10,909
我认为他们破产的可能性微乎其微

654
00:22:10,909 --> 00:22:12,368
我知道这是最后的忠告

655
00:22:12,368 --> 00:22:14,489
但他们的资产负债表强得惊人

656
00:22:14,489 --> 00:22:15,868
而且正在偿还债务

657
00:22:15,868 --> 00:22:17,980
明知危机即将来临

658
00:22:17,980 --> 00:22:20,019
却只想尽可能榨取更多资金

659
00:22:20,019 --> 00:22:20,740
在这段时间里

660
00:22:20,740 --> 00:22:24,980
并构建最坚固的基础设施和抢占最大市场份额

661
00:22:24,980 --> 00:22:27,319
他们确实很聪明

662
00:22:27,319 --> 00:22:29,920
我真的很敬佩这些高管

663
00:22:29,920 --> 00:22:31,859
他们正在做正确的事情

664
00:22:31,859 --> 00:22:33,940
他们非常自律

665
00:22:33,940 --> 00:22:34,940
非常强大的公司

666
00:22:34,940 --> 00:22:37,460
我对这家公司几乎挑不出毛病

667
00:22:37,460 --> 00:22:39,900
我其实非常期待

668
00:22:39,900 --> 00:22:42,849
周三将会是重要的一天

669
00:22:42,849 --> 00:22:43,970
嗯，是的

670
00:22:43,970 --> 00:22:44,809
就是我回来的日子

671
00:22:46,009 --> 00:22:46,609
所以啊

672
00:22:46,609 --> 00:22:46,869
我们

673
00:22:46,869 --> 00:22:48,150
之后会分析财报

674
00:22:48,150 --> 00:22:48,819
不过总之

675
00:22:48,819 --> 00:22:49,539
如果你喜欢这个

676
00:22:49,539 --> 00:22:50,380
请考虑订阅

677
00:22:50,380 --> 00:22:51,839
我们下次视频再见

678
00:22:51,839 --> 00:22:53,400
再见，再见

共 4 条，按点赞数排序。

采集完成，该视频无弹幕