万亿营收喂不饱华尔街：一场比抢芯片更疯狂的资本围猎，已经打响!丨这个板块迎来最强背书，错过后悔一辈子!MU,SNDK,P#trading #投資 #美股

1
00:00:00,000 --> 00:00:01,600
刚刚落幕的英伟达财报业

2
00:00:01,600 --> 00:00:05,009
再次上演了一场救美股于水火的史诗级戏码

3
00:00:05,009 --> 00:00:06,030
炸裂的营收

4
00:00:06,030 --> 00:00:06,950
逆天的毛利

5
00:00:06,950 --> 00:00:09,630
黄仁秀用一份几乎没有瑕疵的成绩单

6
00:00:09,630 --> 00:00:11,560
让空头们彻底闭了嘴

7
00:00:11,560 --> 00:00:13,000
但如果你真的以为

8
00:00:13,000 --> 00:00:15,199
这场狂欢的最大赢家是英伟达

9
00:00:15,199 --> 00:00:17,699
或者是那些抢到芯片的科技巨头

10
00:00:17,699 --> 00:00:20,579
那你就太小看资本市场的借力打力了

11
00:00:20,579 --> 00:00:22,579
在这个万物皆可AI的时代

12
00:00:22,579 --> 00:00:24,500
真正悄悄把利润吃干抹净

13
00:00:24,500 --> 00:00:28,059
甚至在未来几年拥有绝对定价权的隐藏大赢家

14
00:00:28,059 --> 00:00:29,359
其实根本不是他

15
00:00:29,359 --> 00:00:30,379
今天这期内容

16
00:00:30,379 --> 00:00:32,439
我们就撕开财报的表面繁荣

17
00:00:32,439 --> 00:00:34,619
带你看清这个躺在英伟达背后

18
00:00:34,619 --> 00:00:37,500
正在闷声发大财的超级巨头

19
00:00:39,380 --> 00:00:40,299
hi朋友们

20
00:00:40,299 --> 00:00:41,600
欢迎来到ruby投资笔记

21
00:00:41,600 --> 00:00:42,700
我是你们的老朋友

22
00:00:42,700 --> 00:00:43,100
Ruby

23
00:00:43,100 --> 00:00:43,899
一家公司

24
00:00:43,899 --> 00:00:46,500
单季度营收直接砸出816亿美元

25
00:00:46,500 --> 00:00:47,799
这是个什么概念

26
00:00:47,799 --> 00:00:50,560
这不仅是远超华尔街预期的神级答卷

27
00:00:50,560 --> 00:00:51,460
更恐怖的是

28
00:00:51,460 --> 00:00:53,179
在这么庞大的体量下

29
00:00:53,179 --> 00:00:54,280
它的环比增速

30
00:00:54,280 --> 00:00:56,729
竟然还能硬生生的增长出20%

31
00:00:56,729 --> 00:01:00,039
现在全网都在为英伟达为black vivo架构疯狂

32
00:01:00,039 --> 00:01:01,320
他们疯狂的是什么

33
00:01:01,320 --> 00:01:02,280
是英伟达的财报

34
00:01:02,280 --> 00:01:03,859
证明了AI绝不是泡沫

35
00:01:03,859 --> 00:01:06,180
从而支撑了美股的强劲反弹

36
00:01:06,180 --> 00:01:08,739
但在我仔细的去研究了他的财报后

37
00:01:08,739 --> 00:01:10,459
我的心里是咯噔了一下

38
00:01:10,459 --> 00:01:11,519
因为我发现

39
00:01:11,519 --> 00:01:14,430
绝大多数的普通投资者都看错了方向

40
00:01:14,430 --> 00:01:16,209
英伟达的财报如此炸裂

41
00:01:16,209 --> 00:01:17,870
但最受益的并不是他

42
00:01:17,870 --> 00:01:19,469
那最受益的是谁呢

43
00:01:19,469 --> 00:01:22,269
答案其实就在我接下来讲的这两个方向

44
00:01:22,269 --> 00:01:23,469
一是新于算力

45
00:01:23,469 --> 00:01:24,459
二是存储

46
00:01:24,459 --> 00:01:27,519
那关于新运算力以及新云算力蕴含的机会

47
00:01:27,519 --> 00:01:29,939
我会放到下周专门出一期视频来讲

48
00:01:29,939 --> 00:01:31,000
内容会比较硬核

49
00:01:31,000 --> 00:01:32,170
千万不能错过

50
00:01:32,170 --> 00:01:32,849
同时呢

51
00:01:32,849 --> 00:01:35,670
我也会在ruby交流圈里提前上传一部分资料

52
00:01:35,670 --> 00:01:37,530
方便大家在下周开盘前

53
00:01:37,530 --> 00:01:40,180
提前去了解这个行业和它的底层逻辑

54
00:01:40,180 --> 00:01:41,640
那今天这期视频呢

55
00:01:41,640 --> 00:01:43,980
我们就先来拆解一下NVIDIA的财报

56
00:01:43,980 --> 00:01:46,040
给镁光西部数据细节科技

57
00:01:46,040 --> 00:01:48,370
这些存储厂商带来的发展机遇

58
00:01:48,370 --> 00:01:50,569
我们光看英伟达数据中心的业务

59
00:01:50,569 --> 00:01:53,750
就能够发现内存正在迎来史诗级的增长速度

60
00:01:53,750 --> 00:01:55,469
我们直接用数据说话

61
00:01:55,469 --> 00:01:57,030
优惠达这个板块的营收

62
00:01:57,030 --> 00:01:58,989
达到了惊人的752亿美元

63
00:01:58,989 --> 00:02:00,269
同比翻了将近一倍

64
00:02:00,269 --> 00:02:01,659
增长了92%

65
00:02:01,659 --> 00:02:04,859
但是我强烈建议大家把这个数据再往下细

66
00:02:04,859 --> 00:02:05,549
拆一层

67
00:02:05,549 --> 00:02:06,510
你仔细看

68
00:02:06,510 --> 00:02:08,110
负责算力的计算板块

69
00:02:08,110 --> 00:02:10,560
是创下了604亿美元的历史记录

70
00:02:10,560 --> 00:02:12,819
而负责连接的网络板块更夸张

71
00:02:12,819 --> 00:02:15,460
他的单季度直接逼近150亿美元

72
00:02:15,460 --> 00:02:18,329
同比增幅拿下了更恐怖的200%

73
00:02:18,329 --> 00:02:19,449
那听到这里

74
00:02:19,449 --> 00:02:20,688
想必你已经明白了

75
00:02:20,688 --> 00:02:24,400
这604亿美元的计算GPU想要真正运转起来

76
00:02:24,400 --> 00:02:26,639
每一块背后都必须一车车的未尽

77
00:02:26,639 --> 00:02:29,289
海量高带宽的内存和底层的存储

78
00:02:29,289 --> 00:02:31,370
没有这些底层的存储军火库

79
00:02:31,370 --> 00:02:33,229
英伟达的算力就算拉的再高

80
00:02:33,229 --> 00:02:35,199
也就是一堆发热的废铁

81
00:02:35,199 --> 00:02:37,919
正如黄仁勋在新闻发布会上所说的

82
00:02:37,919 --> 00:02:39,080
AI工厂的建设

83
00:02:39,080 --> 00:02:41,959
是人类历史上最大规模的基础设施扩张

84
00:02:41,959 --> 00:02:44,849
而且正在以惊人的速度席卷全球

85
00:02:44,849 --> 00:02:46,090
数据不会说谎

86
00:02:46,090 --> 00:02:48,750
AI计算的物理极限正在被不断的打破

87
00:02:48,750 --> 00:02:51,210
而存储层作为算力的亲密伴侣

88
00:02:51,210 --> 00:02:52,959
已经验来最强烈的需求

89
00:02:52,959 --> 00:02:54,878
但这里面更深的底气在于

90
00:02:54,878 --> 00:02:57,598
整个存储板块能有今天这么硬的估值

91
00:02:57,598 --> 00:02:58,118
背后

92
00:02:58,118 --> 00:03:00,519
其实还有一个被物理定律死死卡住的

93
00:03:00,519 --> 00:03:02,599
长达数年的供应缺口

94
00:03:04,680 --> 00:03:06,240
为了验证这个大逻辑

95
00:03:06,240 --> 00:03:08,000
我专门去看了美光的高管

96
00:03:08,000 --> 00:03:10,969
在摩根大通全球科技大会上的最新发言

97
00:03:10,969 --> 00:03:12,490
在那场炉边谈话里

98
00:03:12,490 --> 00:03:14,129
美光的全球运营副总裁

99
00:03:14,129 --> 00:03:16,139
表现出了少有的极度自信

100
00:03:16,139 --> 00:03:19,340
他直接甩出了一个让我非常兴奋的定性结论

101
00:03:19,340 --> 00:03:19,860
他说

102
00:03:19,860 --> 00:03:23,319
美国的资产负债表从来没有像今天这么强劲过

103
00:03:23,319 --> 00:03:25,580
紧接着他抛出了三个硬核利好

104
00:03:25,580 --> 00:03:26,319
大家注意

105
00:03:26,319 --> 00:03:27,039
这三点

106
00:03:27,039 --> 00:03:30,400
正在重塑我们过去对于内存行业的估值逻辑

107
00:03:30,400 --> 00:03:32,319
第一就是整个AI市场

108
00:03:32,319 --> 00:03:35,849
正在经历一场从人机互动到机器互动的质变

109
00:03:35,849 --> 00:03:39,110
现在的AI早就不是简单的陪你聊两句天

110
00:03:39,110 --> 00:03:41,729
他正在疯狂转向人与智能体A进的

111
00:03:41,729 --> 00:03:45,039
甚至是机器与机器M2M之间的深度交互

112
00:03:45,039 --> 00:03:46,400
这种演变意味着

113
00:03:46,400 --> 00:03:49,789
AI模型的上下文窗口正在呈指数级的拉长

114
00:03:49,789 --> 00:03:51,490
而内存的容量和速度

115
00:03:51,490 --> 00:03:54,639
直接决定这些智能模型的准确性和性能

116
00:03:54,639 --> 00:03:57,120
所以现在那些大型的云服务厂商

117
00:03:57,120 --> 00:04:00,240
已经不再把内存当成一种普通的商品来看待

118
00:04:00,240 --> 00:04:02,060
而是把它当作不可或缺

119
00:04:02,060 --> 00:04:03,680
高度重视的战略资产

120
00:04:03,680 --> 00:04:06,539
他们甚至开始通过签署战略客户协议

121
00:04:06,539 --> 00:04:08,539
提前好几年去锁定产能

122
00:04:08,539 --> 00:04:11,930
这在过去整个半导体行业里是根本无法想象的

123
00:04:11,930 --> 00:04:12,610
第二点呢

124
00:04:12,610 --> 00:04:15,250
也就是大家作为投资者都必须看到的一个内容

125
00:04:15,250 --> 00:04:17,629
是一个致命的HBM交易比率

126
00:04:17,629 --> 00:04:18,670
很多人问我

127
00:04:18,670 --> 00:04:19,790
既然需求这么旺

128
00:04:19,790 --> 00:04:22,339
厂商加班加点多生产点不就行了

129
00:04:22,339 --> 00:04:23,639
真没有这么简单

130
00:04:23,639 --> 00:04:25,160
HBM高带宽内存

131
00:04:25,160 --> 00:04:27,060
因为它极端的性能架构

132
00:04:27,060 --> 00:04:30,569
它的实际芯片尺寸比传统标准的D2M大得多

133
00:04:30,569 --> 00:04:32,970
这就产生了一个物理霸王条款

134
00:04:32,970 --> 00:04:34,769
晶圆厂的产能就那么大

135
00:04:34,769 --> 00:04:36,370
一旦切产HBM

136
00:04:36,370 --> 00:04:38,170
就会导致每片晶圆上

137
00:04:38,170 --> 00:04:40,668
切出来的总比特数就大幅度减少

138
00:04:40,668 --> 00:04:42,468
再加上HBM的需求

139
00:04:42,468 --> 00:04:43,908
现在直接大塞车

140
00:04:43,908 --> 00:04:46,588
原有的标准的DM产能被严重挤压

141
00:04:46,588 --> 00:04:48,040
两个缺口同时撕开

142
00:04:48,040 --> 00:04:50,079
全行业面临大面积的缺货

143
00:04:50,079 --> 00:04:53,569
而第三个利好则是绿色无尘车间的物理限制

144
00:04:53,569 --> 00:04:56,750
行业的工艺开发现在已经碰到了物理瓶颈

145
00:04:56,750 --> 00:04:59,470
美国现在重点攻克的下一代先进工艺

146
00:04:59,470 --> 00:05:00,970
也就是一纳米的节点

147
00:05:00,970 --> 00:05:03,069
已经不是靠现有的老产线

148
00:05:03,069 --> 00:05:05,529
家电设备改改参数就能够搞定的

149
00:05:05,529 --> 00:05:07,290
他们必须得从零开始

150
00:05:07,290 --> 00:05:11,000
砸重金去建全新的绿色无尘车间而建

151
00:05:11,000 --> 00:05:13,720
这些高端基础设施动辄需要好几年

152
00:05:13,720 --> 00:05:15,279
所以美国直接放话

153
00:05:15,279 --> 00:05:15,959
短期内

154
00:05:15,959 --> 00:05:19,129
全球供应根本不可能跟得上AI需求的爆发

155
00:05:19,129 --> 00:05:21,610
当一个全球运营负责人公开承认

156
00:05:21,610 --> 00:05:23,129
物理定律和建设周期

157
00:05:23,129 --> 00:05:25,500
注定会导致多年的供应缺口时

158
00:05:25,500 --> 00:05:28,790
聪明的资本其实早就已经开始绕道重仓了

159
00:05:28,790 --> 00:05:31,709
而且存储产业有一个我不得不说的优点

160
00:05:31,709 --> 00:05:34,709
就是各大存储厂商希望面临的行业竞争局面

161
00:05:34,709 --> 00:05:35,829
是百花齐放

162
00:05:35,829 --> 00:05:36,790
百家争鸣的

163
00:05:36,790 --> 00:05:38,399
他们不希望一家垄断

164
00:05:38,399 --> 00:05:39,959
所以从这个角度来说

165
00:05:39,959 --> 00:05:42,420
存储行业的竞争它是非常健康

166
00:05:42,420 --> 00:05:45,360
它是有着良心向上生长土壤的一个行业

167
00:05:45,360 --> 00:05:48,740
而这也是我非常喜欢存储行业的一点

168
00:05:50,459 --> 00:05:54,019
那如果说美光是代表了大厂们供不上货的焦虑

169
00:05:54,019 --> 00:05:56,259
那另外一家存储巨头闪迪

170
00:05:56,259 --> 00:05:59,519
则是直接颠覆了商业世界最残酷的游戏规则

171
00:05:59,620 --> 00:06:00,560
众所周知

172
00:06:00,560 --> 00:06:03,399
存储行业几十年来都逃不过繁荣过剩

173
00:06:03,399 --> 00:06:05,319
暴跌亏损的死亡周期

174
00:06:05,319 --> 00:06:06,660
但就在几天前

175
00:06:06,660 --> 00:06:09,420
闪迪的CEO突然丢出了一篇重磅文章

176
00:06:09,420 --> 00:06:11,319
高调唱红自家的产品

177
00:06:11,319 --> 00:06:13,540
他给出的判断比镁光还要激进

178
00:06:13,540 --> 00:06:15,959
他宣布闪存市场至少会供不应求

179
00:06:15,959 --> 00:06:17,158
到2027年底

180
00:06:17,158 --> 00:06:19,199
这个让无数厂商破产的魔咒

181
00:06:19,199 --> 00:06:21,500
已经被AI永久性的粉碎了

182
00:06:21,500 --> 00:06:22,500
回顾过去

183
00:06:22,500 --> 00:06:23,459
闪存和内存

184
00:06:23,459 --> 00:06:25,300
那历来就是一个残酷的周期

185
00:06:25,300 --> 00:06:27,019
生意行情好的时候

186
00:06:27,019 --> 00:06:28,019
厂商疯狂建厂

187
00:06:28,019 --> 00:06:28,759
过度生产

188
00:06:28,759 --> 00:06:29,879
随后价格暴跌

189
00:06:29,879 --> 00:06:30,759
全行业亏钱

190
00:06:30,759 --> 00:06:31,740
屡试不爽

191
00:06:31,740 --> 00:06:34,990
但是今天AI彻底打破了这个死亡循环

192
00:06:34,990 --> 00:06:37,550
这些大学云服务商和企业数据中心

193
00:06:37,550 --> 00:06:38,949
为了保住算力优势

194
00:06:38,949 --> 00:06:42,139
急需稳定且有保障的高密度的闪存供应

195
00:06:42,139 --> 00:06:44,279
于是闪迪正在做一件大事

196
00:06:44,279 --> 00:06:47,819
成功地把整个商业模式从过去的现货市场交易

197
00:06:47,819 --> 00:06:50,660
转向了长期有约束力的多年级合约

198
00:06:50,660 --> 00:06:52,259
就像是期货一样

199
00:06:52,259 --> 00:06:54,600
这彻底改变了公司的风险状况

200
00:06:54,600 --> 00:06:56,459
以前做城主是看天吃饭

201
00:06:56,459 --> 00:06:58,569
价格波动大的像是坐过山车

202
00:06:58,569 --> 00:07:00,730
现在变成了类似SARS软件股

203
00:07:00,730 --> 00:07:03,589
或者是电力基础设施那样稳定收租的生意

204
00:07:03,589 --> 00:07:05,730
它极大的降低了下行的风险

205
00:07:05,730 --> 00:07:07,680
带来了极其可测的现金流

206
00:07:07,680 --> 00:07:09,439
这也完美解释了为什么

207
00:07:09,439 --> 00:07:11,259
花旗在最新发布的报告中

208
00:07:11,259 --> 00:07:12,100
直接指出

209
00:07:12,100 --> 00:07:14,759
闪迪的股票还有52%的上涨空间

210
00:07:14,759 --> 00:07:17,939
并把目标价定在了一个让人惊掉下巴的高位

211
00:07:17,939 --> 00:07:19,620
不过大家先别急着激动

212
00:07:19,620 --> 00:07:22,540
在替大家深度挖掘整个底层产业链的时候

213
00:07:22,540 --> 00:07:23,540
我顺藤摸瓜

214
00:07:23,540 --> 00:07:24,379
抓到了两个

215
00:07:24,379 --> 00:07:27,819
被绝大多数市场分析师完全忽略的财富暴击点

216
00:07:27,819 --> 00:07:29,319
而这两个隐藏杀招

217
00:07:29,319 --> 00:07:32,279
或许是接下来内存板块能够持续超预期

218
00:07:32,279 --> 00:07:33,610
爆发的真正底牌

219
00:07:33,610 --> 00:07:34,730
那顺便说一句

220
00:07:34,730 --> 00:07:38,139
如果你对于闪迪这种高波动的标的的操作策略

221
00:07:38,139 --> 00:07:39,139
没有把握的话

222
00:07:39,139 --> 00:07:41,560
可以点击评论区的置顶链接加入我们

223
00:07:41,560 --> 00:07:43,480
获取为您量身定制的方案

224
00:07:43,480 --> 00:07:45,100
那接下来我们就来继续

225
00:07:45,100 --> 00:07:47,649
拆解存储行业的隐藏杀招

226
00:07:49,620 --> 00:07:51,220
这第一个隐藏杀招呢

227
00:07:51,220 --> 00:07:53,660
是关乎大模型落地的一个致命痛点

228
00:07:53,660 --> 00:07:54,408
电费

229
00:07:54,408 --> 00:07:56,809
当所有人都在讨论算力荒的时候

230
00:07:56,809 --> 00:08:00,408
AI巨头们其实在被昂贵的电费折磨得夜不能寐

231
00:08:00,408 --> 00:08:02,189
马斯克甚至公开预测过

232
00:08:02,189 --> 00:08:04,389
未来限制AI发展的核心瓶颈

233
00:08:04,389 --> 00:08:05,209
除了芯片

234
00:08:05,209 --> 00:08:06,199
就是电力

235
00:08:06,199 --> 00:08:08,180
而在服务器的整体能耗中

236
00:08:08,180 --> 00:08:09,600
可能有很多人不知道

237
00:08:09,600 --> 00:08:11,720
存储系统和内存的功耗占比

238
00:08:11,720 --> 00:08:14,220
竟然高达30%到40%

239
00:08:14,220 --> 00:08:17,220
一个拥有10万个H100节点的高规格

240
00:08:17,220 --> 00:08:18,160
AI数据中心

241
00:08:18,160 --> 00:08:21,220
一年的电费支出轻松突破2亿美元

242
00:08:21,220 --> 00:08:23,500
如果继续使用传统的存储架构

243
00:08:23,500 --> 00:08:25,288
随着大模型参数的暴涨

244
00:08:25,288 --> 00:08:27,369
光是维存内存数据不丢失

245
00:08:27,369 --> 00:08:28,389
所消耗的电量

246
00:08:28,389 --> 00:08:31,279
就能把这些科技巨头们利润直接烧光

247
00:08:31,279 --> 00:08:34,360
这就逼得各大云厂商不得不向更高的工艺

248
00:08:34,360 --> 00:08:36,828
更低功耗的下一代内存架构升级

249
00:08:36,828 --> 00:08:40,408
比如镁光和闪迪正在全力推进的全新高容量

250
00:08:40,408 --> 00:08:41,960
低功耗的内存产品

251
00:08:41,960 --> 00:08:45,179
它们能在提供数倍于传统内存带宽的同时

252
00:08:45,179 --> 00:08:47,700
将功耗直接砍掉30%以上

253
00:08:47,700 --> 00:08:51,460
在绿色数据中心和全球电力紧张的双重高压下

254
00:08:51,460 --> 00:08:54,039
这种能够省下天文数字电费的存储

255
00:08:54,039 --> 00:08:56,460
已经不再是可有可无的升级选择

256
00:08:56,460 --> 00:08:59,970
而是大厂们为了省电必须买单的救命稻草

257
00:08:59,970 --> 00:09:03,029
谁手里的先进工艺能够把功耗压得更低

258
00:09:03,029 --> 00:09:05,840
谁就能独占这个万亿级的高毛利市场

259
00:09:05,840 --> 00:09:09,019
如果说功耗是倒逼云端大厂升级的鞭子

260
00:09:09,019 --> 00:09:12,320
那最后一个看好逻辑则是属于我们每一个人

261
00:09:12,320 --> 00:09:14,320
手里都在用的设备

262
00:09:15,899 --> 00:09:16,980
过去两年呢

263
00:09:16,980 --> 00:09:18,940
AI的狂欢一直停留在云端

264
00:09:18,940 --> 00:09:21,600
停留在那些离我们很远的数据中心里

265
00:09:21,600 --> 00:09:24,519
但是硅谷的巨头们绝对不会满足于

266
00:09:24,519 --> 00:09:25,840
只赚大厂的钱

267
00:09:25,840 --> 00:09:27,440
一个正在悄悄成型

268
00:09:27,440 --> 00:09:29,879
即将卷走所有人钱包的终端大洗牌

269
00:09:29,879 --> 00:09:31,340
已经拉开了序幕

270
00:09:31,340 --> 00:09:33,980
这就是资本市场目前最期待的破局

271
00:09:33,980 --> 00:09:36,129
点端侧AI的全面爆发

272
00:09:36,129 --> 00:09:37,289
从AI手机

273
00:09:37,289 --> 00:09:40,850
AIPC端到接下来可能普及的各类终端智能设备

274
00:09:40,850 --> 00:09:41,809
比如说手表

275
00:09:41,809 --> 00:09:43,470
比如说车载AI等等

276
00:09:43,470 --> 00:09:46,509
如果你想要在手机或者电脑本地流畅运行

277
00:09:46,509 --> 00:09:49,860
一个70亿甚至更高参数的AI智能体模型

278
00:09:49,860 --> 00:09:52,059
有一个无法逾越的硬件关卡

279
00:09:52,059 --> 00:09:53,980
那就是内存必须足够大

280
00:09:53,980 --> 00:09:56,000
在过去一个普通的旗舰手机

281
00:09:56,000 --> 00:09:58,179
8GB或者是12GB的内存

282
00:09:58,179 --> 00:10:00,200
就可以让手机跑得很流畅

283
00:10:00,200 --> 00:10:03,899
因为所有的app都只是在做简单的本地计算

284
00:10:03,899 --> 00:10:06,720
但是端侧AI要求模型常驻内存

285
00:10:06,720 --> 00:10:08,620
随时准备响应你的指令

286
00:10:08,620 --> 00:10:11,159
那根据我拿到的行业测算数据

287
00:10:11,159 --> 00:10:14,299
一部能够流畅运行本地AI大模型的手机

288
00:10:14,299 --> 00:10:17,960
起步内存至少要达到16GB甚至是24GB

289
00:10:17,960 --> 00:10:21,080
而AIPC的门槛更是直接被拉高到了

290
00:10:21,080 --> 00:10:23,370
32GB甚至是64GB

291
00:10:23,370 --> 00:10:24,509
这意味着什么

292
00:10:24,509 --> 00:10:25,210
这意味着

293
00:10:25,210 --> 00:10:25,929
过去几年

294
00:10:25,929 --> 00:10:28,289
因为消费电子饱和而陷入沉寂的

295
00:10:28,289 --> 00:10:31,409
手机和PC供应链即将迎来一场恐怖的

296
00:10:31,409 --> 00:10:34,179
有物理底层驱动的内存容量翻倍超

297
00:10:34,179 --> 00:10:37,279
即使全球手机和电脑的总出货量不增长

298
00:10:37,279 --> 00:10:39,460
光是单机内存容量的翻倍

299
00:10:39,460 --> 00:10:42,600
就足以让镁光闪迪这些底层的存储巨头

300
00:10:42,600 --> 00:10:46,379
在消费端重新迎来一轮巨大的业绩暴增

301
00:10:47,139 --> 00:10:50,100
那在确定了整个存储行业

302
00:10:50,100 --> 00:10:52,159
供不应求的长期大前景之后

303
00:10:52,159 --> 00:10:52,860
接下来呢

304
00:10:52,860 --> 00:10:54,779
我必须和大家紧急更新一下

305
00:10:54,779 --> 00:10:56,960
我的AVP2之后的操作重点

306
00:10:57,009 --> 00:10:59,450
那半个月前我拆解MVP2的时候

307
00:10:59,450 --> 00:11:01,190
他刚突破前高的阻力位

308
00:11:01,190 --> 00:11:03,570
我原本的打算是他会稳扎稳打

309
00:11:03,570 --> 00:11:04,590
以时间换空间

310
00:11:04,590 --> 00:11:05,818
逐步的走上去

311
00:11:05,818 --> 00:11:07,639
但是计划赶不上变化

312
00:11:07,639 --> 00:11:10,019
市场直接用一根爆大量的阳线

313
00:11:10,019 --> 00:11:12,009
彻底打乱了我原本的节奏

314
00:11:12,009 --> 00:11:14,769
所以呢我的投资策略现在有了改变

315
00:11:14,769 --> 00:11:16,490
在聊后面的新策略之前

316
00:11:16,490 --> 00:11:18,330
我先带大家快速复盘一下

317
00:11:18,330 --> 00:11:21,490
这家公司到底是靠什么在市场上立足的

318
00:11:21,490 --> 00:11:22,490
简单来说呢

319
00:11:22,490 --> 00:11:25,169
AVP2是一家专门做企业级存储架构的

320
00:11:25,169 --> 00:11:26,029
硬核玩家

321
00:11:26,029 --> 00:11:29,429
他们最擅长的就是通过定制化的全闪存阵列

322
00:11:29,429 --> 00:11:33,299
去帮助那些大厂用最快的速度把AI大模型落地

323
00:11:33,299 --> 00:11:36,080
所以站在现阶段的产业视角来看

324
00:11:36,080 --> 00:11:37,980
虽然他的短期业绩爆发力

325
00:11:37,980 --> 00:11:40,809
可能没有直接卖NETFLASH的闪迪那么猛

326
00:11:40,809 --> 00:11:44,169
但它也是和整个存储板块的大行情共振的

327
00:11:44,169 --> 00:11:46,169
而随着存储行情的反转

328
00:11:46,169 --> 00:11:48,529
它的技术形态也已经筑底成功

329
00:11:48,529 --> 00:11:51,210
那接下来呢最需要大家关注的时间点

330
00:11:51,210 --> 00:11:54,509
就是他5月27号盘后的这份最新财报

331
00:11:54,509 --> 00:11:57,049
我复盘了他过去几年的这个周线图

332
00:11:57,049 --> 00:11:59,250
抓到一个非常反常的波动规律

333
00:11:59,250 --> 00:12:01,169
只要到了财报发布的当周

334
00:12:01,169 --> 00:12:04,309
IVP2的股价就会迎来一段成交量最剧烈

335
00:12:04,309 --> 00:12:04,769
波顿

336
00:12:04,769 --> 00:12:06,109
最惊人的过山车

337
00:12:06,109 --> 00:12:08,168
我们直接用历史数据说话

338
00:12:08,168 --> 00:12:10,249
2025年Q2财报的当中

339
00:12:10,249 --> 00:12:12,328
它是一周飙升了33%

340
00:12:12,328 --> 00:12:13,609
到了Q3季度

341
00:12:13,609 --> 00:12:14,989
它的财报是直接砸盘

342
00:12:14,989 --> 00:12:16,249
暴跌了20%

343
00:12:16,249 --> 00:12:19,740
那Q4季度呢是继续顺势回调了13%

344
00:12:19,740 --> 00:12:22,879
而即将到来的2026年第一季度的财报

345
00:12:22,879 --> 00:12:26,240
毫无疑问将成为他接下来能否打破周线颓势

346
00:12:26,240 --> 00:12:28,559
走出新一轮主升浪的转折点

347
00:12:28,559 --> 00:12:31,179
那针对接下来即将放榜的这次财报

348
00:12:31,179 --> 00:12:34,649
我主要的关注方向就是下面三个最硬核的指标

349
00:12:34,649 --> 00:12:37,690
第一个指标呢是订阅这年经常收入收入

350
00:12:37,690 --> 00:12:39,230
A arr i vp

351
00:12:39,230 --> 00:12:42,750
二呢和卖硬件的SNDK最大的区别就在于

352
00:12:42,750 --> 00:12:45,730
它成功的把存储做成类似SARS软件一样

353
00:12:45,730 --> 00:12:47,230
稳定性收租的生意

354
00:12:47,230 --> 00:12:49,230
他之前几个季度的arr

355
00:12:49,230 --> 00:12:51,210
都稳定在17亿美元以上

356
00:12:51,210 --> 00:12:54,409
那这一次呢我们就要重点看他的AI的增速

357
00:12:54,409 --> 00:12:56,669
能否在AI需求倒逼的情况下

358
00:12:56,669 --> 00:12:58,009
突破同比18%

359
00:12:58,009 --> 00:13:00,179
甚至是20%的加速拐点

360
00:13:00,179 --> 00:13:02,179
如果这个数据能够超预期

361
00:13:02,179 --> 00:13:04,820
就说明那些拿到IVP2产品的公司

362
00:13:04,820 --> 00:13:07,559
在持续的为IVP2提供现金流

363
00:13:07,559 --> 00:13:11,200
这是他商业模型能够享受估值溢价的绝对底气

364
00:13:11,200 --> 00:13:14,450
那第二个指标呢是超大规模的云厂商订单

365
00:13:14,450 --> 00:13:16,250
GUCIVP2的主要客户

366
00:13:16,250 --> 00:13:18,200
大多都是传统的企业级市场

367
00:13:18,200 --> 00:13:19,519
所以对于现在的它

368
00:13:19,519 --> 00:13:21,000
市场上质疑最多的就是

369
00:13:21,000 --> 00:13:22,779
除了metal用了它的存储方案

370
00:13:22,779 --> 00:13:25,190
还有没有哪家大厂在用他的方案

371
00:13:25,190 --> 00:13:27,309
而在上个季度的电话会议里面

372
00:13:27,309 --> 00:13:29,250
管理层其实已经明确透露过

373
00:13:29,250 --> 00:13:32,509
他们正在和几家全球最顶级的超大规模云厂商

374
00:13:32,509 --> 00:13:33,940
进行深度的探讨

375
00:13:33,940 --> 00:13:36,539
所以这一次的财报会和电话会议

376
00:13:36,539 --> 00:13:37,620
我最关心的就是

377
00:13:37,620 --> 00:13:40,339
大客户落地的靴子能不能真正落实

378
00:13:40,339 --> 00:13:42,999
哪怕只是管理层在会里面透露一句

379
00:13:42,999 --> 00:13:44,859
已经和某家头部云厂商

380
00:13:44,859 --> 00:13:46,980
达成了全闪存架构的战略合作

381
00:13:46,980 --> 00:13:48,259
不说名字也可以

382
00:13:48,259 --> 00:13:50,799
都会坚定我对IVPI的投资看法

383
00:13:50,799 --> 00:13:53,759
最后一个指标也是最考验管理层操盘能力的

384
00:13:53,759 --> 00:13:55,849
就是他的非GAP毛利率

385
00:13:55,849 --> 00:13:57,548
正如我们前面所说的

386
00:13:57,548 --> 00:14:00,528
现在上游的HBM和NETFLASH的产能大

387
00:14:00,528 --> 00:14:03,068
赛车原材料的价格正在疯涨

388
00:14:03,068 --> 00:14:06,369
作为一家做架构集成和全闪存阵列的厂商

389
00:14:06,369 --> 00:14:07,149
Avp2

390
00:14:07,149 --> 00:14:09,068
能不能把上涨的原材料成本

391
00:14:09,068 --> 00:14:11,899
完美的转嫁给急需算力的下游客户呢

392
00:14:11,899 --> 00:14:13,700
他的非GAP的毛利率

393
00:14:13,700 --> 00:14:15,940
能否继续死死守在70%

394
00:14:15,940 --> 00:14:17,399
这一道基础底线上呢

395
00:14:17,399 --> 00:14:20,029
将会是我们这次财报关注的重点

396
00:14:20,029 --> 00:14:22,190
如果他的毛利率继续往上走

397
00:14:22,190 --> 00:14:24,470
就说明他们公司能够完美的应对

398
00:14:24,470 --> 00:14:26,220
NETFLASH的涨价影响

399
00:14:26,220 --> 00:14:29,700
那总结来说就是IVPI的逻辑其实非常纯粹

400
00:14:29,700 --> 00:14:31,539
他不需要去和英伟达抢风头

401
00:14:31,539 --> 00:14:34,940
他只需要默默的在幕后把算力带来的海量数据

402
00:14:34,940 --> 00:14:37,578
用最低的功耗和最快的速度存下来

403
00:14:37,578 --> 00:14:39,158
然后提供给企业

404
00:14:39,158 --> 00:14:40,658
5月27日盘后

405
00:14:40,658 --> 00:14:43,589
就是见证它是真金还是概念的决战之夜

406
00:14:43,589 --> 00:14:45,109
那如果有任何消息的话

407
00:14:45,109 --> 00:14:46,308
我会第一时间更新

408
00:14:46,308 --> 00:14:47,068
在焦虑圈

409
00:14:47,068 --> 00:14:49,028
可以点击评论区的置顶链接

410
00:14:49,028 --> 00:14:50,159
我们一起等待

411
00:14:50,159 --> 00:14:52,240
那本期的视频到这里就结束了

412
00:14:52,240 --> 00:14:55,259
如果你对于存储行业反弹有什么不一样的看法

413
00:14:55,259 --> 00:14:57,240
欢迎你把观点表达在评论区

414
00:14:57,240 --> 00:14:58,929
我们可以互相探讨一下

415
00:14:58,929 --> 00:15:01,490
同时也欢迎您点赞订阅ruby投资笔记

416
00:15:01,490 --> 00:15:03,070
不错过下一期精彩视频

417
00:15:03,070 --> 00:15:03,889
我是ruby

418
00:15:03,889 --> 00:15:05,860
咱们一期一会下期再见拜

共 1 条，按点赞数排序。

采集完成，该视频无弹幕