数据中心是高速收费站

1
00:00:00,440 --> 00:00:04,439
甲骨文用了3~4天的时间涨了30%多

2
00:00:04,719 --> 00:00:05,918
从底部130多

3
00:00:05,918 --> 00:00:07,078
现在涨180了

4
00:00:07,078 --> 00:00:11,019
各位我讲数据中心讲了很长时间

5
00:00:11,019 --> 00:00:13,460
我说从GPU英伟达开始啊

6
00:00:13,460 --> 00:00:16,780
到CPUAMD到液冷为基础

7
00:00:16,780 --> 00:00:17,920
到光模块

8
00:00:17,920 --> 00:00:20,859
再到往下内存是吧

9
00:00:20,859 --> 00:00:22,260
就是镁光啊

10
00:00:22,260 --> 00:00:23,820
西部数据啊

11
00:00:23,820 --> 00:00:27,199
然后还有这个闪迪三星海力士

12
00:00:27,199 --> 00:00:28,679
我接下来一定会到数据中心

13
00:00:28,679 --> 00:00:29,559
我说了很多次

14
00:00:29,559 --> 00:00:32,079
你们可以看我从2月份3月份所有的视频

15
00:00:32,079 --> 00:00:32,640
你可以看看

16
00:00:32,640 --> 00:00:33,479
网上都有

17
00:00:33,479 --> 00:00:35,579
每次直播都讲过数据中心商业模式不好

18
00:00:35,579 --> 00:00:37,799
但是最后一公里一定是它

19
00:00:37,799 --> 00:00:38,640
就我经常讲

20
00:00:38,640 --> 00:00:39,479
我说一个高速公路

21
00:00:39,479 --> 00:00:41,229
公路在高速公路呢

22
00:00:41,229 --> 00:00:42,850
他是一个什么逻辑呢

23
00:00:42,850 --> 00:00:45,109
就是高速公路上卖沙子

24
00:00:45,109 --> 00:00:45,829
卖石肘的

25
00:00:45,829 --> 00:00:49,429
就像哎英伟达AMD啊

26
00:00:49,429 --> 00:00:50,289
英特尔

27
00:00:50,289 --> 00:00:52,390
然后再往后呢就是维帝技术啊

28
00:00:52,390 --> 00:00:53,429
易中天啊

29
00:00:53,429 --> 00:00:54,469
再接下来啊

30
00:00:54,469 --> 00:00:56,329
就是这内存公式了

31
00:00:56,329 --> 00:00:58,070
那最后呢收费站是谁啊

32
00:00:58,070 --> 00:00:59,469
就是数据中心啊

33
00:00:59,469 --> 00:01:02,130
你最后所有的所有的token

34
00:01:02,130 --> 00:01:03,310
所有的数据训练

35
00:01:03,310 --> 00:01:04,930
你最后还是回到数据中心

36
00:01:04,930 --> 00:01:05,569
他有提价能力

37
00:01:05,569 --> 00:01:07,370
他只是说所以东这个商业模式

38
00:01:07,370 --> 00:01:09,209
它是它是折旧快对吧

39
00:01:09,209 --> 00:01:09,849
五到6年

40
00:01:09,849 --> 00:01:11,519
但是呢你要投数据中时

41
00:01:11,519 --> 00:01:12,239
你的逻辑是什么呢

42
00:01:12,239 --> 00:01:13,599
你看我是把数据中心分类啊

43
00:01:13,599 --> 00:01:15,338
比如说第一梯队啊

44
00:01:15,338 --> 00:01:16,838
就是微软啊

45
00:01:16,838 --> 00:01:19,599
亚马逊谷歌是吧

46
00:01:19,599 --> 00:01:21,540
然后第二梯队呢是甲骨文

47
00:01:21,540 --> 00:01:22,439
是第三梯呢

48
00:01:22,439 --> 00:01:23,920
就是啊

49
00:01:23,920 --> 00:01:28,079
诶诶就是那个CIWVINNBIS对吧

50
00:01:28,079 --> 00:01:31,000
那那甲骨文比起来CWNINNBI

51
00:01:31,000 --> 00:01:31,599
有个优势是什么

52
00:01:31,599 --> 00:01:32,420
准业务赚钱

53
00:01:32,420 --> 00:01:33,280
数据库赚钱

54
00:01:33,280 --> 00:01:36,209
甲骨文的数据库40多万个高端用户啊

55
00:01:36,209 --> 00:01:37,370
他的迁移成本很高

56
00:01:37,370 --> 00:01:38,730
准翼也不错对吧

57
00:01:38,730 --> 00:01:40,769
甲骨文其实今年走势不好啊

58
00:01:40,769 --> 00:01:41,769
就是无非两个点

59
00:01:41,769 --> 00:01:43,170
第一个呢负债率太高了啊

60
00:01:43,170 --> 00:01:43,890
有些负债率很高

61
00:01:43,890 --> 00:01:44,980
1000多亿啊

62
00:01:44,980 --> 00:01:45,819
第二个是什么呢

63
00:01:45,819 --> 00:01:48,140
就是open i的订单能不能兑现

64
00:01:48,140 --> 00:01:49,739
实际上在上个季度财报的时候

65
00:01:49,739 --> 00:01:50,980
甲骨文已经有一个苗头啊

66
00:01:50,980 --> 00:01:52,390
就是他已经证明了啊

67
00:01:52,390 --> 00:01:55,799
就他的数据中心呢可以让客户去出这个

68
00:01:55,799 --> 00:01:56,939
出这个所有的费用啊

69
00:01:56,939 --> 00:01:59,739
比如说啊让客户出这个啊硬件啊

70
00:01:59,739 --> 00:02:00,780
GPUCPU啊

71
00:02:00,780 --> 00:02:02,180
他自己只是做仓库管理啊

72
00:02:02,180 --> 00:02:03,049
做运营中心

73
00:02:03,049 --> 00:02:04,930
实际上他已经变成平台模式了啊

74
00:02:04,930 --> 00:02:06,250
第二个呢就是资金流的问题

75
00:02:06,250 --> 00:02:07,730
但融资也也没差是吧

76
00:02:07,730 --> 00:02:09,009
四五百个亿很快融上了

77
00:02:09,009 --> 00:02:10,209
其实这就是一个逻辑

78
00:02:10,209 --> 00:02:12,408
只不过呢这个情绪恐慌的时候呢

79
00:02:12,408 --> 00:02:14,068
就是哐哐的掉啊

80
00:02:14,068 --> 00:02:18,058
从这个我记得是从300多掉到了130

81
00:02:18,899 --> 00:02:20,859
打了个3~4次哈哈哈

82
00:02:20,859 --> 00:02:22,699
实际上实际上我们分享的时候

83
00:02:22,699 --> 00:02:23,870
大概是一百七八啊

84
00:02:23,870 --> 00:02:25,750
中间也就是该做期权嘛

85
00:02:25,750 --> 00:02:26,789
实际上你最后算下来

86
00:02:26,789 --> 00:02:28,280
其实今天肯定是盈利了的

87
00:02:28,280 --> 00:02:29,680
对你应该不会亏的

88
00:02:29,680 --> 00:02:30,199
因数据中心

89
00:02:30,199 --> 00:02:33,199
最后啊其实大家发现那随着脱困的量越来越大

90
00:02:33,199 --> 00:02:34,080
数据中开始涨价

91
00:02:34,080 --> 00:02:34,759
所有都涨价啊

92
00:02:34,759 --> 00:02:36,110
谷歌确实去年涨价了

93
00:02:36,110 --> 00:02:38,050
实际上涨完价之后原因涨价啊

94
00:02:38,050 --> 00:02:39,310
他所有的增速都涨价

95
00:02:39,310 --> 00:02:41,129
就是你发现它利润增速更高了

96
00:02:41,129 --> 00:02:44,050
其实甲骨文在他上个季度财报的时候

97
00:02:44,050 --> 00:02:45,770
他的利润增速大概70%多

98
00:02:45,770 --> 00:02:47,338
就云这块70%多

99
00:02:47,338 --> 00:02:48,639
整体的话四五十%

100
00:02:48,639 --> 00:02:51,658
其实这个这个增速对应个二十五三十倍

101
00:02:51,658 --> 00:02:52,419
其实还可以的

102
00:02:52,419 --> 00:02:54,939
实际上啊就按照这个AI领域哈

103
00:02:54,939 --> 00:02:55,539
你在国内的话

104
00:02:55,539 --> 00:02:57,000
那能干到六七十倍了

105
00:02:57,000 --> 00:02:58,599
所以这个呢也是啊

106
00:02:58,599 --> 00:03:00,120
我们分享这个数据中心的原因啊

107
00:03:00,120 --> 00:03:02,659
其实啊并不是说谁的啊

108
00:03:02,659 --> 00:03:03,379
预测能力强

109
00:03:03,379 --> 00:03:04,240
他是逻辑的问题

110
00:03:04,240 --> 00:03:06,340
就是究竟谁时有提价能力

111
00:03:06,340 --> 00:03:07,860
这个是我觉得在AI的时候

112
00:03:07,860 --> 00:03:08,219
最主要的

113
00:03:08,219 --> 00:03:09,139
因为你的业绩预估

114
00:03:09,139 --> 00:03:11,818
说实话你说估估5年之后啊

115
00:03:11,818 --> 00:03:12,859
谁还会更牛逼啊

116
00:03:12,859 --> 00:03:14,139
比如说有的人说啊

117
00:03:14,139 --> 00:03:15,199
光模块会怎么样啊

118
00:03:15,199 --> 00:03:15,900
存储会怎么样

119
00:03:15,900 --> 00:03:16,460
不知道

120
00:03:16,460 --> 00:03:17,360
真的不知道啊

121
00:03:17,360 --> 00:03:18,579
所以东西会怎么样不知道

122
00:03:18,579 --> 00:03:21,239
但是呢按照高速公路收费站的逻辑

123
00:03:21,239 --> 00:03:23,718
其实最后的高速公路收费站肯定是能赚钱的

124
00:03:23,718 --> 00:03:27,258
因为他你最后所有通过他的他都可以是吧啊

125
00:03:27,258 --> 00:03:28,538
就雁过拔毛是吧

126
00:03:28,538 --> 00:03:29,378
坐坐地起价

127
00:03:29,378 --> 00:03:30,338
你最后一定是这样的

128
00:03:30,338 --> 00:03:31,139
你在供不应求的时候

129
00:03:31,139 --> 00:03:31,959
你也只能接受

130
00:03:31,959 --> 00:03:33,618
所以他这个游戏的门槛很高

131
00:03:33,618 --> 00:03:35,039
实际上所有数据资金

132
00:03:35,039 --> 00:03:36,620
我觉得最终的美卡一定是千亿美金起

133
00:03:36,620 --> 00:03:38,000
就千亿美金啊

134
00:03:38,000 --> 00:03:40,199
你不管谁来这个做这个事

135
00:03:40,199 --> 00:03:41,199
你没有1000亿美金

136
00:03:41,199 --> 00:03:42,599
你连最基本的门槛过不去

137
00:03:42,599 --> 00:03:46,780
所以我觉得这个里面呢只要是啊能够赶砸的

138
00:03:46,780 --> 00:03:48,060
拿到优质订单的啊

139
00:03:48,060 --> 00:03:51,379
然后第一波呢能够拿到这种低价的客户啊

140
00:03:51,379 --> 00:03:53,479
连这优质的客户和低价硬件了

141
00:03:53,479 --> 00:03:54,379
其实他就已经赢了

142
00:03:54,379 --> 00:03:55,419
拿过来赚钱对吧

143
00:03:55,419 --> 00:03:57,219
那谁最先解决电力问题啊

144
00:03:57,219 --> 00:03:58,400
赢得赢得面更大了

145
00:03:58,400 --> 00:03:59,860
那这种叫趋势投资啊

146
00:03:59,860 --> 00:04:01,110
你说现金流是吧

147
00:04:01,110 --> 00:04:01,710
那个没办法

148
00:04:01,710 --> 00:04:02,569
你说这个啊

149
00:04:02,569 --> 00:04:04,229
这个折旧折旧速度

150
00:04:04,229 --> 00:04:05,629
这个说实话实说

151
00:04:05,629 --> 00:04:07,370
那你怎么算呢

152
00:04:07,370 --> 00:04:09,930
那微软他说他折旧速度他分摊的快是吧

153
00:04:09,930 --> 00:04:11,189
他他用最快的速度分摊成本

154
00:04:11,189 --> 00:04:12,870
那微软是跌35%对吧

155
00:04:12,870 --> 00:04:13,909
那至于说怎么算折旧

156
00:04:13,909 --> 00:04:14,509
那不知道的啊

157
00:04:14,509 --> 00:04:15,039
比如说

158
00:04:15,039 --> 00:04:18,000
那那你你你今天买的时候花了500个亿硬件

159
00:04:18,000 --> 00:04:19,680
那你现在内存翻倍了

160
00:04:19,680 --> 00:04:21,180
你是不是零件可以变成1000亿了

161
00:04:21,180 --> 00:04:22,509
那你折旧你怎么折呀

162
00:04:22,509 --> 00:04:23,970
你那1100折还是外一折

163
00:04:23,970 --> 00:04:25,610
这种事情都可以用财务来测算

164
00:04:25,610 --> 00:04:26,009
你知道吗

165
00:04:26,009 --> 00:04:28,089
我觉得中大还是估值对吧

166
00:04:28,089 --> 00:04:29,149
就这个行业增速多少

167
00:04:29,149 --> 00:04:30,189
应该给多少估值

168
00:04:30,189 --> 00:04:31,930
那只要这个估值以内有安全变异

169
00:04:31,930 --> 00:04:33,009
那我觉得这是对的

170
00:04:33,009 --> 00:04:34,970
所以实际上这也恭喜啊

171
00:04:34,970 --> 00:04:36,910
就是拿到现在的人其实也不容易

172
00:04:37,810 --> 00:04:41,399
毕竟拿个三折四折这个这么大市值公司啊

173
00:04:41,399 --> 00:04:43,480
这万亿市值跌到了5000亿

174
00:04:43,480 --> 00:04:44,839
这个还是很凶残的

175
00:04:44,839 --> 00:04:47,680
呵呵没有的心理素质拿不到啊

176
00:04:47,680 --> 00:04:48,180
恭喜

177
00:04:48,180 --> 00:04:49,500
但是我就数据中心心

178
00:04:49,500 --> 00:04:50,839
这绝对是最后一公里

179
00:04:50,839 --> 00:04:53,420
绝对是高速公路收费站的逻辑

共 1 条，按点赞数排序。

采集完成，该视频无弹幕