450亿天价订单落地，Neocloud算力大洗牌！四家明星公司，谁才是笑到最后的隐形赢家？CRWV,NBIS,IREN,CIFR#trading #投資 #美股

1
00:00:00,080 --> 00:00:02,879
如果说你没有买到翻了11倍的LITE

2
00:00:02,879 --> 00:00:05,759
有没有追上年初至今涨了五倍的SNDK

3
00:00:05,759 --> 00:00:07,160
那么在2026年

4
00:00:07,160 --> 00:00:08,759
还有没有什么办法能够

5
00:00:08,759 --> 00:00:10,960
在这场万亿级的AI军备竞赛里

6
00:00:10,960 --> 00:00:12,580
分走最后一块蛋糕呢

7
00:00:12,580 --> 00:00:14,699
答案就在SPSPACEX刚刚砸下来

8
00:00:14,699 --> 00:00:17,359
那张450亿美元的算力合同里

9
00:00:17,359 --> 00:00:19,800
这张合同瞬间把两家不碰芯片

10
00:00:19,800 --> 00:00:22,379
只当电力房东的公司推上了神坛

11
00:00:22,379 --> 00:00:23,260
而同时呢

12
00:00:23,260 --> 00:00:25,480
也把另外两个纯算力的云玩家

13
00:00:25,480 --> 00:00:27,339
逼上了截然不同的命运

14
00:00:27,339 --> 00:00:29,920
究竟谁是跟着巨兽起飞的技术流

15
00:00:29,920 --> 00:00:32,359
谁又是随时可能暴雷的公司呢

16
00:00:32,359 --> 00:00:34,759
本期视频我将深入拆解new cloud

17
00:00:34,759 --> 00:00:36,899
赛道里最瞩目的四家明星公司

18
00:00:36,899 --> 00:00:38,759
在space x AI的冲击下

19
00:00:38,759 --> 00:00:40,179
各自的发展路径

20
00:00:40,179 --> 00:00:43,619
并结合四个核心维度给出我心目中的number one

21
00:00:43,619 --> 00:00:45,420
同时我也会在视频的最后

22
00:00:45,420 --> 00:00:47,820
给出这四家公司的合理内在价值

23
00:00:47,820 --> 00:00:51,500
所以本期精彩视频你千万不能错过

24
00:00:53,140 --> 00:00:54,100
hi朋友们

25
00:00:54,100 --> 00:00:55,380
欢迎来到ruby投资笔记

26
00:00:55,380 --> 00:00:56,539
我是你们的老朋友

27
00:00:56,539 --> 00:00:56,979
Ruby

28
00:00:56,979 --> 00:01:00,500
过去市场只是知道斯space x和XAI合并

29
00:01:00,500 --> 00:01:01,920
但是谁也说不清楚

30
00:01:01,920 --> 00:01:04,379
马斯克的这一步棋到底是为了什么

31
00:01:04,379 --> 00:01:08,650
而最新的IPO文件把这场万亿布局掀开惊天一角

32
00:01:08,650 --> 00:01:10,450
最新的资产重组显示

33
00:01:10,450 --> 00:01:13,450
马斯克正式把XAI并入了space x

34
00:01:13,450 --> 00:01:16,370
组成了全新的space x AI分支

35
00:01:16,370 --> 00:01:20,030
这个动作看起来只是一次简单的组织架构调整

36
00:01:20,030 --> 00:01:21,010
但实际上

37
00:01:21,010 --> 00:01:23,530
它背后隐藏的是一次足以改变AI

38
00:01:23,530 --> 00:01:25,599
基础设施格局的超级整合

39
00:01:25,599 --> 00:01:27,200
因为这一合并意味着

40
00:01:27,200 --> 00:01:30,159
原本属于XAI的庞大地面算力资产

41
00:01:30,159 --> 00:01:32,890
现在全部进入了space x体系

42
00:01:32,890 --> 00:01:34,829
也就是说从这一刻开始

43
00:01:34,829 --> 00:01:37,689
斯space x已经不再只是那个造火箭

44
00:01:37,689 --> 00:01:38,290
发卫星

45
00:01:38,290 --> 00:01:39,980
卖星链的航天公司了

46
00:01:39,980 --> 00:01:43,340
他正式变成了一家拥有全球顶级算力集群的

47
00:01:43,340 --> 00:01:44,489
new cloud巨头

48
00:01:44,489 --> 00:01:46,150
那怎样判断一家公司

49
00:01:46,150 --> 00:01:48,659
到底是不是真正的new cl的玩家呢

50
00:01:48,659 --> 00:01:50,040
标准其实很简单

51
00:01:50,040 --> 00:01:52,680
就是看你能不能把裸金属GPU算力

52
00:01:52,680 --> 00:01:54,379
大规模的租给外部的客户

53
00:01:54,379 --> 00:01:56,750
甚至是租给你的潜在竞争对手

54
00:01:56,750 --> 00:01:58,590
而SPSPACE的X刚刚下场

55
00:01:58,590 --> 00:02:02,009
就直接打出来整个行业最夸张的一张王牌

56
00:02:02,009 --> 00:02:02,929
AI巨头

57
00:02:02,929 --> 00:02:03,668
Anthropic

58
00:02:03,668 --> 00:02:05,629
正式和space x签下了一份

59
00:02:05,629 --> 00:02:07,909
多年期的超级算力采购协议

60
00:02:07,909 --> 00:02:10,459
这个协议的规模大到什么程度呢

61
00:02:10,459 --> 00:02:13,618
ATHROPIC以每月12.5亿美元的价格

62
00:02:13,618 --> 00:02:15,239
直接抛下了space x

63
00:02:15,239 --> 00:02:18,939
位于孟菲斯的close one数据中心全部算力资源

64
00:02:18,939 --> 00:02:20,599
注意这不是部分租赁

65
00:02:20,599 --> 00:02:21,789
是整个集群

66
00:02:21,789 --> 00:02:25,069
而这个集权背后的硬件规格更是夸张的离谱

67
00:02:25,069 --> 00:02:27,750
它拥有超过22万颗NVIDIAGPU

68
00:02:27,750 --> 00:02:28,909
涵盖了H100

69
00:02:28,909 --> 00:02:31,750
H200以及下一代GB200的加速器

70
00:02:31,750 --> 00:02:35,250
整个数据中心的供电规模已经超过了300兆瓦

71
00:02:35,250 --> 00:02:36,270
更关键的是

72
00:02:36,270 --> 00:02:37,500
这还只是开始

73
00:02:37,500 --> 00:02:41,159
双方的合作已经推进到了close to的后续扩建

74
00:02:41,159 --> 00:02:43,960
整个合同周期直接签到了2029年

75
00:02:43,960 --> 00:02:46,409
总价值是接近450亿美元

76
00:02:46,409 --> 00:02:49,090
这张space x瞬间成为AI算力

77
00:02:49,090 --> 00:02:51,590
再让订单规模最夸张的玩家之一

78
00:02:51,590 --> 00:02:54,349
尤其是他给出的7.8美元一小时

79
00:02:54,349 --> 00:02:56,069
GPU的算力租赁价格

80
00:02:56,069 --> 00:02:59,229
更是直接在二级市场掀起了巨大的冲击

81
00:02:59,229 --> 00:03:02,189
因为这个价格体系很可能会重新锚定AI

82
00:03:02,189 --> 00:03:04,139
算力行业未来的估值逻辑

83
00:03:04,139 --> 00:03:06,939
而当space x这种级别的超级巨头

84
00:03:06,939 --> 00:03:08,939
亲自下场做算力租赁之后

85
00:03:08,939 --> 00:03:11,520
那些已经上市的纯牛克劳的公司

86
00:03:11,520 --> 00:03:13,379
到底是会迎来估值重估呢

87
00:03:13,379 --> 00:03:15,539
还是会被马斯克降维打击呢

88
00:03:15,539 --> 00:03:18,379
尤其是那些本身现金流已经非常紧张

89
00:03:18,379 --> 00:03:20,400
严重依赖融资输血的玩家

90
00:03:20,400 --> 00:03:24,120
space x的出现会不会直接把他们逼上绝路呢

91
00:03:24,120 --> 00:03:26,080
接下来咱们就来聊聊两只

92
00:03:26,080 --> 00:03:30,960
现在市场上最瞩目的算力独角兽NBS和KV夫

93
00:03:32,599 --> 00:03:36,280
NBS和COVIV的商业模式是直接向客户出租

94
00:03:36,280 --> 00:03:38,180
带有软件生态的GPU算力

95
00:03:38,180 --> 00:03:40,990
是space x AI的直接竞争对手

96
00:03:40,990 --> 00:03:42,750
先说NBAS这家公司

97
00:03:42,750 --> 00:03:45,639
那NEBAS最大的受益点其实是价格

98
00:03:45,639 --> 00:03:49,360
因为space x这次是直接把GPU的租赁价格锚定了

99
00:03:49,360 --> 00:03:50,680
7.8美元一小时

100
00:03:50,680 --> 00:03:53,439
这相当于是一脚把整个行业的价格

101
00:03:53,439 --> 00:03:55,090
天花板重新抬高了

102
00:03:55,090 --> 00:03:57,169
而NBS的反应也非常快

103
00:03:57,169 --> 00:04:00,909
在市场突然意识到AI算力依旧极度短缺之后

104
00:04:00,909 --> 00:04:04,528
NBS已经开始上调了自己的GPU租赁报价

105
00:04:04,528 --> 00:04:05,769
这意味着什么呢

106
00:04:05,769 --> 00:04:08,990
意味着它短期的盈利能力会被明显强化

107
00:04:08,990 --> 00:04:11,710
尤其是2026年Q1季度的毛利率

108
00:04:11,710 --> 00:04:14,189
以及年底冲刺70到90亿美元

109
00:04:14,189 --> 00:04:16,290
a arr年金长期收入的目标

110
00:04:16,290 --> 00:04:18,189
现在都获得了更强的支撑

111
00:04:18,189 --> 00:04:21,160
但真正让市场开始重新提供NBS的

112
00:04:21,160 --> 00:04:22,740
并不是他的收入增长

113
00:04:22,740 --> 00:04:24,000
而是资产负债表

114
00:04:24,000 --> 00:04:26,439
因为相比很多同行疯狂举债扩张

115
00:04:26,439 --> 00:04:28,788
NEVS的财务结构要健康得多

116
00:04:28,788 --> 00:04:31,569
目前他的债务规模大约40亿美元

117
00:04:31,569 --> 00:04:33,709
而且第一季度已经打出了

118
00:04:33,709 --> 00:04:36,100
接近45%的ABT的利润率

119
00:04:36,100 --> 00:04:38,720
这个数字在整个纽可乐的赛道里

120
00:04:38,720 --> 00:04:40,740
已经算是非常能打的数据

121
00:04:40,740 --> 00:04:41,779
更关键的是

122
00:04:41,779 --> 00:04:44,970
他现在还有能力靠自身的现金流继续扩张

123
00:04:44,970 --> 00:04:47,230
包括之前签下的26亿美元

124
00:04:47,230 --> 00:04:48,850
B1的燃料电池协议

125
00:04:48,850 --> 00:04:52,209
本质上也是在为他后面4G瓦的电力扩张计划

126
00:04:52,209 --> 00:04:53,209
提前铺路

127
00:04:53,209 --> 00:04:54,410
所以短期来看

128
00:04:54,410 --> 00:04:56,250
space x抬高行业价格

129
00:04:56,250 --> 00:04:58,540
对于NE贝尔来说其实是利好

130
00:04:58,540 --> 00:05:00,800
但问题在于长期发展呢

131
00:05:00,800 --> 00:05:02,720
NEIGHBIS长期最大的隐患

132
00:05:02,720 --> 00:05:04,350
其实是他的全站能力

133
00:05:04,350 --> 00:05:07,730
因为space x最可怕的地方不只是它有GPU

134
00:05:07,730 --> 00:05:10,709
而是它同时拥有火箭新链网络能源系统

135
00:05:10,709 --> 00:05:12,459
数据中轨道运输能力

136
00:05:12,459 --> 00:05:15,939
他几乎把整个AI基础设施链条全包了

137
00:05:15,939 --> 00:05:16,980
那相比之下

138
00:05:16,980 --> 00:05:19,639
NBS虽然第三方的身份足够独立

139
00:05:19,639 --> 00:05:21,720
也更容易获得中立客户的信任

140
00:05:21,720 --> 00:05:24,790
但他缺少是真正底层的全站网络能力

141
00:05:24,790 --> 00:05:25,810
这就意味着

142
00:05:25,810 --> 00:05:28,629
未来在争夺顶级客户超大订单的时候

143
00:05:28,629 --> 00:05:31,779
他可能会越来越难和space x正面对抗

144
00:05:31,779 --> 00:05:34,579
到最后他很可能会被市场逼向两个方向

145
00:05:34,579 --> 00:05:36,120
一个是中型企业客户

146
00:05:36,120 --> 00:05:38,970
另一个是更偏推理型AI的利基市场

147
00:05:38,970 --> 00:05:39,870
也就是说

148
00:05:39,870 --> 00:05:42,189
NBS的问题不是不能赚钱

149
00:05:42,189 --> 00:05:43,769
而是未来还能不能继续

150
00:05:43,769 --> 00:05:45,678
站在产业链最核心的位置

151
00:05:45,678 --> 00:05:48,418
再来看另一个更夸张的玩家call wave

152
00:05:48,418 --> 00:05:49,499
从短期来看

153
00:05:49,499 --> 00:05:53,019
SPSPACEX这次反而帮call wave打开了估值天花板

154
00:05:53,019 --> 00:05:54,040
那为什么呢

155
00:05:54,040 --> 00:05:57,980
是因为coffee是现在市场公认的new可乐的龙头之一

156
00:05:57,980 --> 00:06:00,129
他手里的订单规模非常夸张

157
00:06:00,129 --> 00:06:03,550
目前在手的IPO已经超过了990亿美元

158
00:06:03,550 --> 00:06:04,769
最值得一提的是

159
00:06:04,769 --> 00:06:07,680
他拥有meta21000000000美元的长期合同

160
00:06:07,680 --> 00:06:08,459
而现在

161
00:06:08,459 --> 00:06:12,620
space x正准备以AI基础设施平台的身份冲刺IPO

162
00:06:12,620 --> 00:06:15,560
这就为new cloud赛道带来了一个新的变化

163
00:06:15,560 --> 00:06:17,220
整个市场突然意识到

164
00:06:17,220 --> 00:06:19,060
原来new cloud公司不一定

165
00:06:19,060 --> 00:06:21,870
只能按照传统的数据中心的定位来估值

166
00:06:21,870 --> 00:06:25,149
它甚至是可能获得超级科技平台的溢价定位

167
00:06:25,149 --> 00:06:28,689
这会直接抬高coo wave在二级市场的估值宝典

168
00:06:28,689 --> 00:06:29,970
所以短期来看

169
00:06:29,970 --> 00:06:33,319
call wave的情绪和估值反而可能继续受益

170
00:06:33,319 --> 00:06:35,279
但问题也恰恰出在这里

171
00:06:35,279 --> 00:06:36,899
因为space x越强

172
00:06:36,899 --> 00:06:39,540
coffee实际上的软肋也会被放的越大

173
00:06:39,540 --> 00:06:42,199
最核心的问题就是它的融资成本

174
00:06:42,199 --> 00:06:44,519
因为space x最大的优势之一就是

175
00:06:44,519 --> 00:06:46,120
它拥有极强的主体信用

176
00:06:46,120 --> 00:06:47,699
它的融资会非常便宜

177
00:06:47,699 --> 00:06:49,079
但是call wave不一样

178
00:06:49,079 --> 00:06:52,569
它的底层逻辑是一个典型的高杠杆扩张模型

179
00:06:52,569 --> 00:06:56,209
他目前账面上的债务已经超过了200亿美元

180
00:06:56,209 --> 00:06:58,129
仅仅是Q一一个季度

181
00:06:58,129 --> 00:07:00,418
他的利息支出就已经超过5亿美元

182
00:07:00,418 --> 00:07:01,619
这意味着什么呢

183
00:07:01,619 --> 00:07:03,418
意味着它整个商业模式

184
00:07:03,418 --> 00:07:06,639
对于高估值和持续融资的依赖性极强

185
00:07:06,639 --> 00:07:08,920
一旦资本市场的风险偏好下降

186
00:07:08,920 --> 00:07:10,908
他的压力会瞬间暴雷

187
00:07:10,908 --> 00:07:13,488
而更致命的问题其实还在后面

188
00:07:13,488 --> 00:07:14,988
因为现在整个行业

189
00:07:14,988 --> 00:07:17,189
还有一个大家都不愿意明说的风险

190
00:07:17,189 --> 00:07:19,519
那就是对于NVIDIA的绝对依赖

191
00:07:19,519 --> 00:07:21,139
而市场已经开始传出

192
00:07:21,139 --> 00:07:23,879
space x正在联合特斯拉和英特尔

193
00:07:23,879 --> 00:07:26,620
推荐所谓的泰瑞发布芯片制造项目

194
00:07:26,620 --> 00:07:28,600
如果这件事情未来真的落地

195
00:07:28,600 --> 00:07:30,480
那事情就会变得非常恐怖

196
00:07:30,480 --> 00:07:31,519
因为它意味着

197
00:07:31,519 --> 00:07:35,430
未来AI算力芯片可能不再完全由NVIDIA1家垄断

198
00:07:35,430 --> 00:07:37,670
而康威夫恰恰是整个行业里面

199
00:07:37,670 --> 00:07:40,209
对于NVIDIA绑定最深的公司之一

200
00:07:40,209 --> 00:07:41,449
那更麻烦的是

201
00:07:41,449 --> 00:07:43,810
他自己其实也并不掌握太多的底层

202
00:07:43,810 --> 00:07:44,949
数据中心资产

203
00:07:44,949 --> 00:07:47,939
很多基础设施它是属于一个租赁状态

204
00:07:47,939 --> 00:07:48,759
也就是说

205
00:07:48,759 --> 00:07:50,459
一旦供应链融资环境

206
00:07:50,459 --> 00:07:52,660
或者是芯片获取能力出现问题

207
00:07:52,660 --> 00:07:55,678
它底层结构的脆弱性会被瞬间放大

208
00:07:55,678 --> 00:07:58,439
所以康wave现在最尴尬的地方就在这里

209
00:07:58,439 --> 00:08:00,819
它可能是这轮AI算力牛市里面

210
00:08:00,819 --> 00:08:02,370
弹性最大的股票之一

211
00:08:02,370 --> 00:08:05,810
但同时也可能是系统性风险最高的玩家之一

212
00:08:05,810 --> 00:08:08,110
那看完这些卖算力的公司之后

213
00:08:08,110 --> 00:08:10,309
咱们再把自己的视角拔高一点

214
00:08:10,309 --> 00:08:12,189
也有很多人忽略了一件事

215
00:08:12,189 --> 00:08:15,428
这些算力云公司哪怕是拿了再多的GPU

216
00:08:15,428 --> 00:08:16,329
如果没有电

217
00:08:16,329 --> 00:08:18,339
他们也只是一堆昂贵的废铁

218
00:08:18,339 --> 00:08:21,879
而马斯克这次住给AERROPIC的300兆瓦的电力

219
00:08:21,879 --> 00:08:25,408
实际上已经把美国部分地区的电网逼到了极限

220
00:08:25,408 --> 00:08:26,449
这也意味着

221
00:08:26,449 --> 00:08:28,809
AI下一阶段真正最暴力的方向

222
00:08:28,809 --> 00:08:30,309
可能根本不是芯片

223
00:08:30,309 --> 00:08:31,459
而是能源

224
00:08:31,459 --> 00:08:34,820
甚至有一家公司他们连一颗GPU都不用买

225
00:08:34,820 --> 00:08:37,769
却能在这场AI军备竞赛里躺着赚钱

226
00:08:37,769 --> 00:08:38,929
那么问题来了

227
00:08:38,929 --> 00:08:40,330
到底是什么样的公司

228
00:08:40,330 --> 00:08:42,090
能够在这场算力战争里

229
00:08:42,090 --> 00:08:44,769
成为真正的幕后收租王呢

230
00:08:46,620 --> 00:08:48,059
在纽克拉的赛道上

231
00:08:48,059 --> 00:08:50,849
有一类玩家比很多人想象的更重要

232
00:08:50,849 --> 00:08:53,408
他们就是从比特币挖矿转型过来的

233
00:08:53,408 --> 00:08:55,408
数据中心与电力容量的巨头

234
00:08:55,408 --> 00:08:57,399
AREN和CNFR

235
00:08:57,399 --> 00:08:59,759
拥有的核心资产是能源和土地

236
00:08:59,759 --> 00:09:01,240
而不是单纯的芯片

237
00:09:01,240 --> 00:09:02,600
先说AREN

238
00:09:02,600 --> 00:09:05,849
AREN现在最大的逻辑其实是资产重估

239
00:09:05,849 --> 00:09:07,688
过去市场一直把它当成一个

240
00:09:07,688 --> 00:09:10,599
况且转型的普通数据中心公司来定价

241
00:09:10,599 --> 00:09:12,038
但是现在不一样了

242
00:09:12,038 --> 00:09:13,499
正如我们刚才所说的

243
00:09:13,499 --> 00:09:15,558
未来真正稀缺的不是GPU

244
00:09:15,558 --> 00:09:17,389
而是低成本的电力资源

245
00:09:17,389 --> 00:09:18,570
这也就意味着

246
00:09:18,570 --> 00:09:21,389
AIEN手里那些极瓦级别的清洁能源

247
00:09:21,389 --> 00:09:23,440
储备的价值正在疯狂上升

248
00:09:23,440 --> 00:09:25,840
包括他已经建好了高密度的机房

249
00:09:25,840 --> 00:09:27,419
以及配套的土地资源

250
00:09:27,419 --> 00:09:30,059
现在都变成了市场眼里的硬通货

251
00:09:30,059 --> 00:09:31,879
所以你又会看到这段时间

252
00:09:31,879 --> 00:09:34,578
整个市场开始重新给这类公司定价

253
00:09:34,578 --> 00:09:37,859
但是相比我接下来要讲的CFRIREN

254
00:09:37,859 --> 00:09:38,818
存在的问题是

255
00:09:38,818 --> 00:09:41,899
他走的是购买芯片的传统new克劳的路线

256
00:09:41,899 --> 00:09:44,659
也就是说它不仅提供数据中心和电力

257
00:09:44,659 --> 00:09:45,799
还要自己买芯片

258
00:09:45,799 --> 00:09:47,299
自己扩GPU集群

259
00:09:47,299 --> 00:09:49,860
它会在融资方面存在一定的压力

260
00:09:49,860 --> 00:09:53,009
因为我们刚才说的这些都是要花费巨额的资金

261
00:09:53,009 --> 00:09:56,649
而这件事情在space x这种资本怪兽入场之后

262
00:09:56,649 --> 00:09:57,919
会变得越来越难

263
00:09:57,919 --> 00:09:59,440
因为钱都是流动性的

264
00:09:59,440 --> 00:10:02,970
热钱以及一些活跃的钱可能都会融入到space x

265
00:10:02,970 --> 00:10:04,730
所以对于IREA来说

266
00:10:04,730 --> 00:10:05,730
他必须尽快的

267
00:10:05,730 --> 00:10:08,389
把手里的电力资源和基础设施变现

268
00:10:08,389 --> 00:10:12,009
否则一旦整个行业进入更疯狂的GPU军备竞赛

269
00:10:12,009 --> 00:10:13,549
他会越来越吃力

270
00:10:13,549 --> 00:10:14,570
那换句话说

271
00:10:14,570 --> 00:10:17,450
AREN的核心价值正在从算力运营商

272
00:10:17,450 --> 00:10:19,860
慢慢转向能源资产的持有者

273
00:10:19,860 --> 00:10:23,340
而另一边ca f2的打法就更有意思了

274
00:10:23,340 --> 00:10:24,259
甚至可以说

275
00:10:24,259 --> 00:10:27,700
他可能是整个new cloud公司里面最聪明的一家

276
00:10:27,700 --> 00:10:29,690
因为他干脆不碰芯片

277
00:10:29,690 --> 00:10:32,750
赛博money今年正式更名为赛博数据

278
00:10:32,750 --> 00:10:34,710
就是在向市场明确一件事

279
00:10:34,710 --> 00:10:36,230
他不做GPU的玩家

280
00:10:36,230 --> 00:10:38,639
他想做的是AI时代的超级房东

281
00:10:38,639 --> 00:10:39,759
那什么意思呢

282
00:10:39,759 --> 00:10:41,840
就是他只负责提供土地

283
00:10:41,840 --> 00:10:42,299
电力

284
00:10:42,299 --> 00:10:43,600
机房和基础设施

285
00:10:43,600 --> 00:10:44,559
至于芯片

286
00:10:44,559 --> 00:10:45,600
那客户自己带

287
00:10:45,600 --> 00:10:48,589
也就是典型的BYOC自带芯片模式

288
00:10:48,589 --> 00:10:50,808
这个模式最大的好处是什么呢

289
00:10:50,808 --> 00:10:54,269
就是它几乎完美规避了GPU快速迭代带来的

290
00:10:54,269 --> 00:10:55,509
硬件贬值风险

291
00:10:55,509 --> 00:10:58,710
因为现在AI芯片的更新速度已经快到离谱

292
00:10:58,710 --> 00:11:00,350
今天还值钱的H100

293
00:11:00,350 --> 00:11:03,149
可能两三年后就会被新的架构迅速替代

294
00:11:03,149 --> 00:11:04,549
很多牛克劳的公司

295
00:11:04,549 --> 00:11:07,440
其实都在承担着巨大的设备折旧风险

296
00:11:07,440 --> 00:11:10,159
但是CIFR基本不碰这个问题

297
00:11:10,159 --> 00:11:13,820
他占的是地产加电力加基础设施的长期租金

298
00:11:13,820 --> 00:11:15,179
所以当space x

299
00:11:15,179 --> 00:11:16,860
Ai aerotic matter

300
00:11:16,860 --> 00:11:20,200
这些巨头开始用整座数据中心打包租赁的方式

301
00:11:20,200 --> 00:11:21,639
疯狂扩张的时候

302
00:11:21,639 --> 00:11:24,559
cf2这种模式反而变得特别吃香

303
00:11:24,559 --> 00:11:26,899
因为市场现在最缺的不是芯片

304
00:11:26,899 --> 00:11:28,279
而是现成合规

305
00:11:28,279 --> 00:11:31,289
能够快速接入电网的大型数据中心基地

306
00:11:31,289 --> 00:11:33,529
尤其是那些具备了几瓦级别

307
00:11:33,529 --> 00:11:35,129
扩张潜力的能源站点

308
00:11:35,129 --> 00:11:38,490
而这恰恰就是CFR手里最核心的资产

309
00:11:38,490 --> 00:11:39,509
更关键的是

310
00:11:39,509 --> 00:11:42,490
它的财务结构会比很多同行要轻松得多

311
00:11:42,490 --> 00:11:45,909
因为它不像ARENNEVS或者是coffee f那样

312
00:11:45,909 --> 00:11:49,599
动不动去砸几10亿美元去追新一代的NVIDIA的GPU

313
00:11:49,599 --> 00:11:52,379
这就意味着在当前高利率的环境下

314
00:11:52,379 --> 00:11:55,279
CIF2的抗风险能力会明显更强

315
00:11:55,279 --> 00:11:57,679
虽然它单个的订单金额可能没有

316
00:11:57,679 --> 00:12:00,078
CVV或者是NEBELS那么夸张

317
00:12:00,078 --> 00:12:01,759
但是它利润的确定性

318
00:12:01,759 --> 00:12:04,220
部署的速度以及现金流的稳定性

319
00:12:04,220 --> 00:12:07,179
在现在这个被space x彻底炒热的AI

320
00:12:07,179 --> 00:12:08,419
基础设施周期里

321
00:12:08,419 --> 00:12:10,690
反而会被市场重新重视

322
00:12:10,690 --> 00:12:11,889
所以你会发现

323
00:12:11,889 --> 00:12:15,309
现在整个new克劳德赛道的规则已经彻底变了

324
00:12:15,309 --> 00:12:17,490
过去大家拼的是谁芯片更多

325
00:12:17,490 --> 00:12:18,389
谁模型更强

326
00:12:18,389 --> 00:12:19,379
谁参数更大

327
00:12:19,379 --> 00:12:21,580
但现在真正决定胜负的开始变成

328
00:12:21,580 --> 00:12:22,179
谁有电

329
00:12:22,179 --> 00:12:22,779
谁有地

330
00:12:22,779 --> 00:12:25,049
谁能最快把大型数据中心落地

331
00:12:25,049 --> 00:12:28,409
这也是为什么IREN和CIF2从年前至今

332
00:12:28,409 --> 00:12:31,429
不断有市场上的利好消息传来的重大原因

333
00:12:31,429 --> 00:12:34,649
而在space x这个超级巨兽下场之后

334
00:12:34,649 --> 00:12:38,210
整个行业的竞争维度又被硬生生的拔高了一层

335
00:12:38,210 --> 00:12:39,669
那问题也来了

336
00:12:39,669 --> 00:12:42,090
面对这四家打法完全不同的公司

337
00:12:42,090 --> 00:12:44,570
在2026年这个新的AI周期里面

338
00:12:44,570 --> 00:12:47,078
我们到底该怎样去分配权重呢

339
00:12:47,078 --> 00:12:47,859
那别急

340
00:12:47,859 --> 00:12:49,999
我已经把这四家公司的核心定位

341
00:12:49,999 --> 00:12:50,678
成长性

342
00:12:50,678 --> 00:12:52,379
c space x带来的冲击

343
00:12:52,379 --> 00:12:53,958
以及他们的估值逻辑

344
00:12:53,958 --> 00:12:56,328
做成了一张完整的终极复盘表

345
00:12:56,328 --> 00:12:58,869
到底谁是真正new克劳D王者呢

346
00:12:58,869 --> 00:13:01,089
咱们来用数据说话

347
00:13:02,840 --> 00:13:06,220
在space x掀起了这场new可乐的大洗牌里

348
00:13:06,220 --> 00:13:09,909
市场的天平正在以前所未有的速度重新倾斜

349
00:13:09,909 --> 00:13:10,990
那这一部分呢

350
00:13:10,990 --> 00:13:12,710
我会从四个核心维度出发

351
00:13:12,710 --> 00:13:14,990
来横向对比这几家公司的基本面

352
00:13:14,990 --> 00:13:17,289
并给出我心目中的排名第一

353
00:13:17,289 --> 00:13:19,450
首先我们从核心定位来说

354
00:13:19,450 --> 00:13:20,389
这四家公司

355
00:13:20,389 --> 00:13:23,609
分别代表了new cloud的两条底层进化路线

356
00:13:23,609 --> 00:13:26,860
奈贝尔和COVIVE属于纯算力的云服务商

357
00:13:26,860 --> 00:13:28,480
NBS主打技术流

358
00:13:28,480 --> 00:13:30,960
主攻推理和全站的软件优化

359
00:13:30,960 --> 00:13:33,360
依靠技术提升GPU的运行效率

360
00:13:33,360 --> 00:13:35,019
COVIVE则是规模狂

361
00:13:35,019 --> 00:13:37,100
他深度绑定英伟达主攻训练

362
00:13:37,100 --> 00:13:39,320
靠大规模举债买入最新的芯片

363
00:13:39,320 --> 00:13:40,450
再打包长租

364
00:13:40,450 --> 00:13:44,070
AREN和CNFR则属于能源基础设商

365
00:13:44,070 --> 00:13:46,110
NIEN走的是自购芯片

366
00:13:46,110 --> 00:13:49,419
与自营绿电并重的硬核重资产路线

367
00:13:49,419 --> 00:13:51,779
那CFR则是纯粹的AI房东

368
00:13:51,779 --> 00:13:52,620
他不买芯片

369
00:13:52,620 --> 00:13:54,259
不承担硬件贬值的风险

370
00:13:54,259 --> 00:13:56,269
只租高容量的电力机房

371
00:13:56,269 --> 00:13:57,110
两个赛道

372
00:13:57,110 --> 00:14:00,679
两种打法定位决定他们的风险承受力完全不同

373
00:14:00,679 --> 00:14:02,120
其次是成长性

374
00:14:02,120 --> 00:14:03,799
康威V的统治力最强

375
00:14:03,799 --> 00:14:06,059
在手订单超过990亿美元

376
00:14:06,059 --> 00:14:07,240
预期营收翻倍

377
00:14:07,240 --> 00:14:10,019
但他紧绷的资产负债表是潜在的阿雷

378
00:14:10,019 --> 00:14:10,860
相比之下

379
00:14:10,860 --> 00:14:12,659
NEVERS有着极强的确定性

380
00:14:12,659 --> 00:14:14,000
它几乎没有烂账

381
00:14:14,000 --> 00:14:17,179
年底a arr将从此70到90亿美元

382
00:14:17,179 --> 00:14:19,700
IREN的成长取决于搞电的速度

383
00:14:19,700 --> 00:14:21,568
它的滤电是一个硬通货

384
00:14:21,568 --> 00:14:24,469
而CFR是速度与防守的结合体

385
00:14:24,469 --> 00:14:27,759
靠着114亿美元的合同锁定远期收入

386
00:14:27,759 --> 00:14:29,240
无需等待芯片排产

387
00:14:29,240 --> 00:14:30,320
高利润的环境下

388
00:14:30,320 --> 00:14:31,879
它的部署速度极快

389
00:14:31,879 --> 00:14:34,720
第三个核心维度是space x的冲击

390
00:14:34,720 --> 00:14:38,240
space x以1.75万亿美元科技平台的定位

391
00:14:38,240 --> 00:14:39,669
递交招股说明书

392
00:14:39,669 --> 00:14:43,110
并以7.8美元一小时高价锁定的ATHAPIC

393
00:14:43,110 --> 00:14:44,490
450亿的长约

394
00:14:44,490 --> 00:14:46,379
彻底分化了现有格局

395
00:14:46,379 --> 00:14:48,639
NEVS短期是涨价的红利

396
00:14:48,639 --> 00:14:51,519
但是长期面临着超级集群的客户挤压

397
00:14:51,519 --> 00:14:53,779
COVIV虽然打开了估值天花板

398
00:14:53,779 --> 00:14:56,559
但是高昂的利息和对英伟达单一供应链

399
00:14:56,559 --> 00:15:00,828
的脆弱性在space x主体信誉的压制下被彻底暴露

400
00:15:00,828 --> 00:15:03,328
IREN的几瓦级电力迎来重估

401
00:15:03,328 --> 00:15:05,789
但是面临着自购芯片的资本压力

402
00:15:05,789 --> 00:15:08,090
CRFR则成为最大的受益者

403
00:15:08,090 --> 00:15:09,009
巨头为了抗洪

404
00:15:09,009 --> 00:15:12,220
space x必须疯狂的抢购其现有的电力

405
00:15:12,220 --> 00:15:13,559
作为纯电力房东

406
00:15:13,559 --> 00:15:15,730
他的资产价值被瞬间放大

407
00:15:15,730 --> 00:15:16,929
最后呢是估值

408
00:15:16,929 --> 00:15:18,990
从投资赔率与性价比来看

409
00:15:18,990 --> 00:15:20,370
四者分化严重

410
00:15:20,370 --> 00:15:22,649
NBAS的债务仅约40亿美元

411
00:15:22,649 --> 00:15:24,789
且他的远期PEE是41倍

412
00:15:24,789 --> 00:15:25,950
成长确定性高

413
00:15:25,950 --> 00:15:27,529
资产负债表极其健康

414
00:15:27,529 --> 00:15:28,750
属于高安全性

415
00:15:28,750 --> 00:15:30,120
高性价比的标的

416
00:15:30,120 --> 00:15:32,179
coffee虽然河东储备恐怖

417
00:15:32,179 --> 00:15:33,700
但是他的PB高企

418
00:15:33,700 --> 00:15:36,179
隐藏着一个高杠杆与贬值的风险

419
00:15:36,179 --> 00:15:37,320
利息开支巨大

420
00:15:37,320 --> 00:15:39,840
当前市场给予其极高的AI溢价

421
00:15:39,840 --> 00:15:41,039
容错率极低

422
00:15:41,039 --> 00:15:43,759
IREA远期的EPS持续亏损

423
00:15:43,759 --> 00:15:47,340
处于从矿股向AI基础设施估值过渡的尴尬期

424
00:15:47,340 --> 00:15:48,419
等待重估

425
00:15:48,419 --> 00:15:51,000
CRFR远期的PEE仅为16倍

426
00:15:51,000 --> 00:15:53,659
凭借独特的只租店不买芯片的模式

427
00:15:53,659 --> 00:15:56,360
在当前高利率与硬件贬值的风险下

428
00:15:56,360 --> 00:15:57,539
防守属性最强

429
00:15:57,539 --> 00:15:59,429
但并不适合重仓死守

430
00:15:59,429 --> 00:16:01,190
那综合前面所有的分析

431
00:16:01,190 --> 00:16:01,990
对我来说

432
00:16:01,990 --> 00:16:05,000
这四家公司排在第一位的依然是NBS

433
00:16:05,000 --> 00:16:06,539
它是new可乐的赛道里面

434
00:16:06,539 --> 00:16:10,240
少数兼具成长性和财务质量的高质量成长标的

435
00:16:10,240 --> 00:16:11,159
第二名的话

436
00:16:11,159 --> 00:16:12,480
我会给到CFR

437
00:16:12,480 --> 00:16:16,240
在高利率的环境下做AI时代的房东的商业模式

438
00:16:16,240 --> 00:16:17,470
防守能力极强

439
00:16:17,470 --> 00:16:20,049
但我并不指望它能带来多大的爆发

440
00:16:20,049 --> 00:16:21,830
第三名我会放call威夫

441
00:16:21,830 --> 00:16:23,850
这是一家典型的顺风起飞

442
00:16:23,850 --> 00:16:25,330
逆风危险的公司

443
00:16:25,330 --> 00:16:27,129
最后一名则是AREN

444
00:16:27,129 --> 00:16:28,590
但这里我要强调一下

445
00:16:28,590 --> 00:16:30,789
排在最后不代表他没有机会

446
00:16:30,789 --> 00:16:31,570
恰恰相反

447
00:16:31,570 --> 00:16:34,779
我觉得AREN可能是里面最硬核的一家公司

448
00:16:34,779 --> 00:16:36,899
但是我对于他的未来发展权益

449
00:16:36,899 --> 00:16:38,860
比如说比特币对于他的影响

450
00:16:38,860 --> 00:16:40,779
再比如说他加入了罗素3000

451
00:16:40,779 --> 00:16:42,559
对于他后续的一个影响

452
00:16:42,559 --> 00:16:44,789
所以我暂时只给他一个第四名

453
00:16:44,789 --> 00:16:46,909
那我们本期视频到这里就结束了

454
00:16:46,909 --> 00:16:48,789
今天的信息量着实是比较大

455
00:16:48,789 --> 00:16:51,470
那面对space x砸板后的新局势

456
00:16:51,470 --> 00:16:53,399
你们觉得谁能笑到最后呢

457
00:16:53,399 --> 00:16:56,240
是手里握着英伟达大单的规模款COVIF

458
00:16:56,240 --> 00:16:58,710
还是不买芯片的房东CRF2呢

459
00:16:58,710 --> 00:17:00,769
欢迎在评论区留下你们的看法

460
00:17:00,769 --> 00:17:02,409
我们可以一起探讨交流

461
00:17:02,409 --> 00:17:03,389
那顺便提一下

462
00:17:03,389 --> 00:17:05,890
我已经把这四家公司的合理内在价值

463
00:17:05,890 --> 00:17:07,130
全部整理好了

464
00:17:07,130 --> 00:17:08,430
那视频时长有限

465
00:17:08,430 --> 00:17:11,318
可以点击我评论区的置顶链接自行领取

466
00:17:11,318 --> 00:17:13,798
而如果你觉得本期视频电影有启发的

467
00:17:13,798 --> 00:17:14,199
朋友们

468
00:17:14,199 --> 00:17:16,459
别忘了点赞订阅并打开小铃铛

469
00:17:16,459 --> 00:17:18,200
不错过下一期精彩视频

470
00:17:18,200 --> 00:17:19,000
我是ruby

471
00:17:19,000 --> 00:17:20,539
咱们一起也会下期再见

472
00:17:20,539 --> 00:17:21,000
拜

共 1 条，按点赞数排序。

采集完成，该视频无弹幕