【Financial Education】浑水做空SOFI，是安然第二还是史诗级黄金坑？

1
00:00:00,160 --> 00:00:01,159
天啊，这些烟民

2
00:00:01,159 --> 00:00:01,840
你简直疯了

3
00:00:01,840 --> 00:00:02,520
我的煎饼果子

4
00:00:02,520 --> 00:00:03,259
女士们先生们

5
00:00:03,259 --> 00:00:07,379
今天早上我在我的前任页面发布了几条动态

6
00:00:07,379 --> 00:00:08,099
呃，然后

7
00:00:08,099 --> 00:00:08,740
这太有趣了

8
00:00:08,740 --> 00:00:12,300
因为在这些帖子里我确实提到了一些细节

9
00:00:12,300 --> 00:00:14,398
我收到了很多网友的评论

10
00:00:14,398 --> 00:00:15,159
杰里米

11
00:00:15,159 --> 00:00:19,589
你看到今天关于sofi的长空报告了吗

12
00:00:19,589 --> 00:00:21,929
此时股价正在暴跌

13
00:00:21,929 --> 00:00:23,629
跌到了十六美元区间

14
00:00:23,629 --> 00:00:25,230
我这是怎么了

15
00:00:25,230 --> 00:00:28,170
于是我开始在浑水里摸鱼

16
00:00:28,170 --> 00:00:30,250
好的浑水做空报告

17
00:00:30,250 --> 00:00:32,738
所以我们发布了这份长空报告

18
00:00:32,738 --> 00:00:34,639
所以在本视频开头

19
00:00:34,639 --> 00:00:37,158
我们先来看看这份做空报告

20
00:00:37,158 --> 00:00:38,289
让我给你展示

21
00:00:38,289 --> 00:00:38,890
你知道的

22
00:00:38,890 --> 00:00:42,409
我觉得这可以称为另一方的辩论

23
00:00:42,409 --> 00:00:43,369
关于此事

24
00:00:43,369 --> 00:00:45,210
谈谈哪些现实哪些不现实

25
00:00:45,210 --> 00:00:46,070
这对意味着什么

26
00:00:46,070 --> 00:00:48,829
所以当前股价走势如何

27
00:00:48,829 --> 00:00:50,590
接下来我要对此做出反应

28
00:00:50,590 --> 00:00:51,630
英伟达CEO

29
00:00:51,630 --> 00:00:52,229
黄仁勋

30
00:00:52,229 --> 00:00:55,450
他最近两天成了热议话题

31
00:00:55,450 --> 00:00:57,390
因为刚开了大型发布会

32
00:00:57,390 --> 00:01:02,630
他似乎竭尽全力想让英伟达股价上涨

33
00:01:02,630 --> 00:01:04,329
但股价仍未上涨

34
00:01:04,329 --> 00:01:05,349
简直难以置信

35
00:01:05,349 --> 00:01:06,349
真的太棒了

36
00:01:06,349 --> 00:01:07,969
一切都这么积极

37
00:01:07,969 --> 00:01:09,469
但股价就是涨不上去

38
00:01:09,469 --> 00:01:11,189
所以我们来聊聊英伟达

39
00:01:11,189 --> 00:01:11,969
到底怎么回事

40
00:01:11,969 --> 00:01:14,060
以及它与AMD的关系

41
00:01:14,060 --> 00:01:16,780
现在有很多AMD股东在看这个视频

42
00:01:16,780 --> 00:01:20,310
我也要谈谈这对AMD的影响

43
00:01:20,310 --> 00:01:23,769
如果你最近在买入SaaS股票

44
00:01:23,769 --> 00:01:25,569
又经历了SaaS崩盘

45
00:01:25,569 --> 00:01:26,409
你可能会想

46
00:01:26,409 --> 00:01:27,609
我觉得这些股票还不错

47
00:01:27,609 --> 00:01:28,510
可能是Salesforce

48
00:01:28,510 --> 00:01:29,590
可能是Adobe

49
00:01:29,590 --> 00:01:30,689
可能是ServiceNow

50
00:01:30,689 --> 00:01:32,009
或者别的什么股票

51
00:01:32,009 --> 00:01:36,989
我认为今天埃米特必须向詹森提出的问题非常重要

52
00:01:36,989 --> 00:01:38,810
并倾听詹森如何回答这个问题

53
00:01:38,810 --> 00:01:40,239
我想你们知道

54
00:01:40,239 --> 00:01:41,400
我要为你们剖析一下

55
00:01:41,400 --> 00:01:43,260
并在此分享我的观点和看法

56
00:01:43,260 --> 00:01:44,280
然后这里最后一个

57
00:01:44,280 --> 00:01:47,359
英伟达将从推理市场增长中获益良多

58
00:01:47,359 --> 00:01:48,750
这将非常重要

59
00:01:48,750 --> 00:01:51,189
尤其是如果你是AMD的股东

60
00:01:51,189 --> 00:01:52,750
感谢大家一如既往地加入我

61
00:01:52,750 --> 00:01:55,189
我只有一件事需要你们

62
00:01:55,189 --> 00:01:58,259
请务必点击点赞按钮

63
00:01:58,259 --> 00:02:00,519
这对我来说意义重大

64
00:02:00,519 --> 00:02:03,618
希望你们能抽出一秒时间

65
00:02:03,618 --> 00:02:06,338
我录制这些视频并点击那个点赞图标

66
00:02:06,338 --> 00:02:10,620
此外我每周通常在这个频道发布2-3个视频

67
00:02:10,620 --> 00:02:13,020
请务必订阅并开启通知

68
00:02:13,020 --> 00:02:14,759
所以当我在这里发布视频时

69
00:02:14,759 --> 00:02:16,759
你们就能随时观看并准备好了

70
00:02:16,759 --> 00:02:17,080
好的

71
00:02:17,080 --> 00:02:19,889
感谢你们的到来和所有支持

72
00:02:19,889 --> 00:02:20,129
凯特

73
00:02:20,129 --> 00:02:20,930
顺便说一句

74
00:02:20,930 --> 00:02:21,789
我的主页

75
00:02:21,789 --> 00:02:23,069
如果你想在X上关注我

76
00:02:23,069 --> 00:02:25,580
我总是将链接放在下方描述区域

77
00:02:25,580 --> 00:02:28,560
今天我也会在下方添加第二个提示评论

78
00:02:28,560 --> 00:02:30,020
就在应用下方

79
00:02:30,020 --> 00:02:32,300
如果你想申请加入我的私有股票群组

80
00:02:32,300 --> 00:02:32,680
好的

81
00:02:32,680 --> 00:02:36,620
所以我将在那个浑水短评下方

82
00:02:36,620 --> 00:02:37,360
所以我

83
00:02:37,360 --> 00:02:38,460
哦宝贝

84
00:02:38,460 --> 00:02:40,110
我们出发吧

85
00:02:40,229 --> 00:02:41,509
浑水是短评

86
00:02:41,509 --> 00:02:46,840
我们认为SOI是一个金融工程 treadmill

87
00:02:46,840 --> 00:02:48,500
不是健康的起始模式

88
00:02:48,500 --> 00:02:51,219
像GE资本那样的业务

89
00:02:51,219 --> 00:02:52,919
马克和罗恩

90
00:02:52,919 --> 00:02:54,379
风格的离场

91
00:02:54,379 --> 00:02:58,840
资产负债结构持续稀释

92
00:02:58,840 --> 00:02:59,620
听着

93
00:02:59,620 --> 00:03:01,819
你将发布一份做空报告

94
00:03:01,819 --> 00:03:02,800
你得聊聊

95
00:03:02,800 --> 00:03:03,500
GPT

96
00:03:03,500 --> 00:03:04,099
你得

97
00:03:04,099 --> 00:03:05,199
谷歌Gemini

98
00:03:05,199 --> 00:03:06,840
如何最佳表述

99
00:03:06,840 --> 00:03:07,740
这正是正确的

100
00:03:07,740 --> 00:03:09,340
这几乎感觉像他们

101
00:03:09,340 --> 00:03:10,699
他就是这样做的

102
00:03:10,699 --> 00:03:14,060
我们怎么能让这个听起来最恐怖

103
00:03:14,060 --> 00:03:15,099
好的

104
00:03:15,099 --> 00:03:15,949
嗯，然后

105
00:03:15,949 --> 00:03:16,849
这就是你要做的

106
00:03:16,849 --> 00:03:18,210
你要做一个简短的报告

107
00:03:18,210 --> 00:03:18,569
好的

108
00:03:18,569 --> 00:03:19,270
首先

109
00:03:19,270 --> 00:03:20,169
他们讨论到

110
00:03:20,169 --> 00:03:21,229
他们进行计算

111
00:03:21,229 --> 00:03:23,520
作为真正的个人贷款

112
00:03:23,520 --> 00:03:24,659
在六点处核销

113
00:03:24,659 --> 00:03:25,919
1%对比2.8%

114
00:03:25,919 --> 00:03:26,460
9%

115
00:03:26,460 --> 00:03:29,879
已报告贷款在核销门槛前数日处置

116
00:03:29,879 --> 00:03:32,599
低估损失影响公允价值模型

117
00:03:32,599 --> 00:03:36,099
产生2.59亿美元不当收益

118
00:03:36,099 --> 00:03:43,838
学生贷款业务似乎主要为管理层奖金创造公允价值收益

119
00:03:43,838 --> 00:03:48,118
如果我使用十年期国债利率下27个基点的折现率

120
00:03:48,118 --> 00:03:49,299
暗示负风险

121
00:03:49,299 --> 00:03:50,039
溢价

122
00:03:50,039 --> 00:03:52,960
沙发安全贷款业务实为卖家

123
00:03:52,960 --> 00:03:55,219
批发融资贷款

124
00:03:55,219 --> 00:03:55,860
哦

125
00:03:55,860 --> 00:03:57,020
这很有趣

126
00:04:00,460 --> 00:04:02,849
看似伪装借贷

127
00:04:02,849 --> 00:04:04,330
计入手续费收入

128
00:04:04,330 --> 00:04:08,780
结构类似安然事件

129
00:04:08,780 --> 00:04:10,580
就像他们拿这和安然比较

130
00:04:10,580 --> 00:04:10,840
对

131
00:04:10,840 --> 00:04:12,949
这太精彩了

132
00:04:12,990 --> 00:04:14,550
我们看到3.12亿美元

133
00:04:14,550 --> 00:04:17,430
显然存在未记录债务

134
00:04:17,430 --> 00:04:19,889
A似乎借入3.12亿美元

135
00:04:19,889 --> 00:04:22,769
我们认为未在财务报表中体现

136
00:04:22,769 --> 00:04:27,910
替代抵押贷款的平台业务疑似伪装借贷

137
00:04:27,910 --> 00:04:29,490
计入手续费收入

138
00:04:29,490 --> 00:04:30,410
再次

139
00:04:30,410 --> 00:04:31,769
与安然对比

140
00:04:31,769 --> 00:04:34,420
调整我们视为虚高的估值

141
00:04:37,399 --> 00:04:39,560
将索菲2025年的

142
00:04:39,560 --> 00:04:41,720
调整EBITDA至90%

143
00:04:41,720 --> 00:04:43,420
从你的角度

144
00:04:43,420 --> 00:04:45,459
从十亿美元降至一亿美元

145
00:04:45,459 --> 00:04:48,370
管理层靠稀释你获利

146
00:04:48,370 --> 00:04:49,329
好的

147
00:04:49,329 --> 00:04:50,050
所以

148
00:04:50,050 --> 00:04:50,709
你知道的

149
00:04:50,709 --> 00:04:51,970
显然

150
00:04:52,939 --> 00:04:54,420
浑水即将

151
00:04:54,420 --> 00:04:56,420
你知道现在就要攻击了

152
00:04:56,420 --> 00:05:01,949
我快速浏览后的第一反应是这里

153
00:05:01,949 --> 00:05:03,029
这就是我想到的

154
00:05:03,029 --> 00:05:04,350
我直接告诉你吧

155
00:05:04,350 --> 00:05:06,660
正是我所想的 我就像

156
00:05:06,660 --> 00:05:06,860
哦

157
00:05:06,860 --> 00:05:07,399
就是这样

158
00:05:07,399 --> 00:05:08,779
我期望更多

159
00:05:08,779 --> 00:05:09,980
我只想和你们坦诚

160
00:05:09,980 --> 00:05:11,259
我期望更多

161
00:05:11,259 --> 00:05:12,060
我当时心想

162
00:05:12,060 --> 00:05:13,319
如果你要针对一家公司

163
00:05:13,319 --> 00:05:13,899
到目前为止

164
00:05:13,899 --> 00:05:14,899
你必须得来

165
00:05:14,899 --> 00:05:16,740
比这更强势

166
00:05:16,740 --> 00:05:17,470
没错

167
00:05:17,470 --> 00:05:18,790
所以谷歌Gemini

168
00:05:18,790 --> 00:05:20,389
谷歌Gemini对此有何看法

169
00:05:20,389 --> 00:05:22,750
我将分享更深入的观点

170
00:05:22,750 --> 00:05:23,470
这里一个

171
00:05:23,470 --> 00:05:26,480
财务工程奖金目标声明

172
00:05:26,480 --> 00:05:31,879
浑水指控管理方通过表内外结构操纵贷款以达成奖金目标

173
00:05:31,879 --> 00:05:36,060
安东尼·诺托曾著名未出售过任何SOFI股份

174
00:05:36,060 --> 00:05:38,100
如果他在为奖金做财务工程

175
00:05:38,100 --> 00:05:39,360
他应该套现

176
00:05:39,360 --> 00:05:39,740
相反

177
00:05:39,740 --> 00:05:43,339
他今日以市价购入二万八千九百股

178
00:05:43,339 --> 00:05:44,240
真正的工程师

179
00:05:44,240 --> 00:05:47,819
不会在崩盘时用自有资金加仓

180
00:05:47,819 --> 00:05:48,240
对啊

181
00:05:48,240 --> 00:05:49,360
所以股价暴跌了

182
00:05:49,360 --> 00:05:49,560
我说

183
00:05:49,560 --> 00:05:50,000
今天一度十六

184
00:05:50,000 --> 00:05:51,019
多左右

185
00:05:51,019 --> 00:05:52,339
SOFA的峰值

186
00:05:52,339 --> 00:05:54,120
我认为在三十到三十二美元左右

187
00:05:54,120 --> 00:05:56,339
现在跌至约五成

188
00:05:56,339 --> 00:05:59,379
三百一十二万美元未记录债务

189
00:05:59,379 --> 00:06:00,319
浑水声称

190
00:06:00,319 --> 00:06:03,699
至少三百一十二万美元存在重大财务误导

191
00:06:03,699 --> 00:06:05,870
未记录债务缺口合规问题

192
00:06:05,870 --> 00:06:10,850
这可能指仓库设施或特殊目的实体结构

193
00:06:10,850 --> 00:06:12,908
标准金融科技借贷

194
00:06:12,908 --> 00:06:15,309
这些已在SEC文件中披露

195
00:06:15,309 --> 00:06:17,369
称其为未记录

196
00:06:17,369 --> 00:06:18,408
因为它们离表

197
00:06:18,408 --> 00:06:22,829
按会计规则离表是术语游戏

198
00:06:22,829 --> 00:06:25,129
并非发现欺诈

199
00:06:25,129 --> 00:06:26,250
不良贷款率

200
00:06:26,250 --> 00:06:27,610
这确实非常重要

201
00:06:27,610 --> 00:06:28,470
在我看来

202
00:06:28,470 --> 00:06:30,790
注意不良贷款率

203
00:06:30,790 --> 00:06:35,129
浑水说六点一百分比对比二点八九百分比对吧

204
00:06:35,129 --> 00:06:37,410
空头交易者常计算不良贷款

205
00:06:37,410 --> 00:06:41,939
使用毛额数据或不同分母

206
00:06:41,939 --> 00:06:45,059
比如滞后数据制造更吓人的数字

207
00:06:45,059 --> 00:06:47,459
索维斯二点八九百分比存在差距

208
00:06:47,459 --> 00:06:48,059
经审计

209
00:06:48,059 --> 00:06:49,980
遵循公认会计原则

210
00:06:49,980 --> 00:06:51,300
年度化数据

211
00:06:51,300 --> 00:06:53,470
如果信用真的那么差

212
00:06:53,470 --> 00:06:55,370
所以我不会认为

213
00:06:55,370 --> 00:06:56,069
呃

214
00:06:56,069 --> 00:06:57,649
持续性盈利缺口

215
00:06:57,649 --> 00:06:59,860
始于2023年末

216
00:06:59,860 --> 00:07:01,810
学生贷款公允价值收益

217
00:07:01,810 --> 00:07:02,569
该主张

218
00:07:02,569 --> 00:07:08,129
学生贷款业务仅存在是为了产生公允价值收益以获取奖金

219
00:07:08,129 --> 00:07:09,629
这其实挺可笑的

220
00:07:09,629 --> 00:07:12,829
学生贷款是SoFi的核心产品

221
00:07:12,829 --> 00:07:15,269
吸引高收入人群

222
00:07:15,269 --> 00:07:15,629
医生

223
00:07:15,629 --> 00:07:16,300
律师

224
00:07:16,300 --> 00:07:20,920
随后购买保险房贷并进行公允价值会计

225
00:07:20,920 --> 00:07:25,060
这是待售贷款行业的标准做法

226
00:07:25,060 --> 00:07:26,740
这不是奖金漏洞

227
00:07:26,740 --> 00:07:29,668
这是资产定价方式

228
00:07:35,189 --> 00:07:37,930
即提前获取现金流

229
00:07:37,930 --> 00:07:40,709
是金融科技模式的基础

230
00:07:40,709 --> 00:07:42,689
向机构伙伴出售贷款

231
00:07:42,689 --> 00:07:45,019
并非借贷而是流动性管理

232
00:07:45,019 --> 00:07:50,439
这就是SoFi循环资本增长无需投入

233
00:07:50,439 --> 00:07:50,899
你知道的

234
00:07:50,899 --> 00:07:52,860
打造无限资产负债表

235
00:07:52,860 --> 00:07:55,019
虚增九千五百万美元

236
00:07:55,019 --> 00:07:59,819
EBIT通过各种会计手段虚增九千五百万美元

237
00:07:59,819 --> 00:08:00,240
指控

238
00:08:00,240 --> 00:08:01,160
浑水公司

239
00:08:01,160 --> 00:08:02,100
反驳

240
00:08:02,100 --> 00:08:04,060
虽然EBITDA可调控

241
00:08:04,060 --> 00:08:05,720
净利润更难掩盖

242
00:08:05,720 --> 00:08:07,519
SoFi转向盈利缺口

243
00:08:07,519 --> 00:08:08,879
净利润盈利能力

244
00:08:08,879 --> 00:08:12,509
包含股票期权和营销费用

245
00:08:12,509 --> 00:08:15,670
EBITDA（息税折旧摊销前利润）

246
00:08:15,670 --> 00:08:16,269
折旧与摊销

247
00:08:16,269 --> 00:08:17,269
顺便说一下

248
00:08:17,269 --> 00:08:19,069
如果你在e比特德论中犹豫不决

249
00:08:19,069 --> 00:08:20,089
其次

250
00:08:20,089 --> 00:08:23,800
资金正在流入这个缺口

251
00:08:23,800 --> 00:08:24,839
底线是绿色

252
00:08:25,879 --> 00:08:29,149
所以你知道我现在的情况是

253
00:08:29,149 --> 00:08:31,750
这是否让我作为沙发股东

254
00:08:31,750 --> 00:08:33,710
不想购买sofi股票

255
00:08:33,710 --> 00:08:35,389
这是否让我想卖掉灵魂

256
00:08:35,389 --> 00:08:37,129
我股票当然不会卖出

257
00:08:37,129 --> 00:08:40,590
我有点感到不快

258
00:08:40,590 --> 00:08:41,269
我没有买入一些

259
00:08:41,269 --> 00:08:42,529
所以我今天持有i股票

260
00:08:42,529 --> 00:08:44,070
我想查看一下报告

261
00:08:44,070 --> 00:08:45,990
等我全部做完时

262
00:08:45,990 --> 00:08:47,519
市场已经收盘了

263
00:08:47,519 --> 00:08:51,139
如果有人因这份报告明天抛售沙发股

264
00:08:51,139 --> 00:08:52,809
我可能会介入买入一些股份

265
00:08:52,809 --> 00:08:55,450
我将继续零星收集sofi股票

266
00:08:55,450 --> 00:08:56,990
暂时在二十美元以下

267
00:08:56,990 --> 00:08:57,730
这是一只股票

268
00:08:57,730 --> 00:09:00,129
我已在十美元以下重仓买入

269
00:09:00,129 --> 00:09:02,299
所以我愿意支付

270
00:09:02,299 --> 00:09:04,179
你知道我们可以说最高到二十美元

271
00:09:04,179 --> 00:09:05,980
我认为这体现了公司实力

272
00:09:05,980 --> 00:09:07,919
以及我对未来发展的看法

273
00:09:07,919 --> 00:09:11,159
当公司发布这类简报时

274
00:09:11,159 --> 00:09:14,679
通常会提供一些具体数据如

275
00:09:14,679 --> 00:09:15,110
哦

276
00:09:15,110 --> 00:09:15,690
你知道的

277
00:09:15,690 --> 00:09:16,590
我们预计是六点

278
00:09:16,590 --> 00:09:17,309
一点百分比

279
00:09:17,309 --> 00:09:17,830
两点八

280
00:09:17,830 --> 00:09:20,639
九个百分点之类的数据对吧

281
00:09:20,639 --> 00:09:21,279
是的

282
00:09:21,279 --> 00:09:25,000
我觉得这种只见树木不见森林的情况

283
00:09:25,000 --> 00:09:25,399
你知道的

284
00:09:25,399 --> 00:09:26,360
如果我们想辩论

285
00:09:26,360 --> 00:09:26,940
这是否

286
00:09:26,940 --> 00:09:29,340
他们应该这样操作还是那样

287
00:09:29,340 --> 00:09:29,818
对吧

288
00:09:29,818 --> 00:09:33,418
空头永远不会百分之百满意

289
00:09:33,418 --> 00:09:33,979
就像

290
00:09:33,979 --> 00:09:35,818
无论哪家公司

291
00:09:36,470 --> 00:09:40,389
整个股市中没有不存在的公司

292
00:09:40,389 --> 00:09:42,330
我无法提出做空理由

293
00:09:42,330 --> 00:09:43,789
让我再说一遍

294
00:09:43,789 --> 00:09:47,169
股市中没有不存在的公司

295
00:09:47,169 --> 00:09:48,990
我无法提出做空理由

296
00:09:48,990 --> 00:09:49,590
我说

297
00:09:49,590 --> 00:09:51,809
即使从顶部我们即将迎来英伟达

298
00:09:51,809 --> 00:09:52,809
听着

299
00:09:52,809 --> 00:09:54,529
即使有英伟达

300
00:09:54,529 --> 00:09:57,450
他们声称英伟达在做

301
00:09:57,730 --> 00:09:58,850
我们还是叫它

302
00:09:58,850 --> 00:09:59,690
呃

303
00:09:59,690 --> 00:10:00,769
可疑的事情

304
00:10:00,769 --> 00:10:02,649
却 somehow 将芯片运入中国

305
00:10:02,649 --> 00:10:03,370
就像那样

306
00:10:03,370 --> 00:10:05,090
人们曾试图主张这一点

307
00:10:05,090 --> 00:10:06,090
他们声称

308
00:10:06,090 --> 00:10:06,509
哦

309
00:10:06,509 --> 00:10:08,690
你知道公司都在重复订购芯片

310
00:10:08,690 --> 00:10:11,759
因此人们提出了各种各样的说法

311
00:10:11,759 --> 00:10:13,539
关于像英伟达这样的公司

312
00:10:13,539 --> 00:10:15,000
像索菲这样的公司

313
00:10:15,000 --> 00:10:16,539
这简直是小菜一碟

314
00:10:16,539 --> 00:10:17,039
你知道的

315
00:10:17,039 --> 00:10:21,120
写一篇简短的报告并发布

316
00:10:21,120 --> 00:10:24,009
所以我并不介意

317
00:10:24,009 --> 00:10:24,330
不过

318
00:10:24,330 --> 00:10:26,129
让我非常明确地说

319
00:10:26,129 --> 00:10:28,580
我喜欢简短的报告啊

320
00:10:28,580 --> 00:10:29,200
你知道的

321
00:10:29,200 --> 00:10:31,120
对于新手投资者

322
00:10:31,120 --> 00:10:33,070
他们对此感到非常沮丧

323
00:10:33,070 --> 00:10:35,149
我其实很感激

324
00:10:35,149 --> 00:10:37,049
因为这些人

325
00:10:37,049 --> 00:10:40,450
他们的职责就是发现可疑之处

326
00:10:40,450 --> 00:10:41,250
而我则想

327
00:10:41,250 --> 00:10:42,269
这就是你发现的全部

328
00:10:42,269 --> 00:10:43,789
他们可能花了数周

329
00:10:43,789 --> 00:10:46,139
甚至数月时间核查

330
00:10:46,139 --> 00:10:51,330
试图找出任何可疑的线索

331
00:10:51,330 --> 00:10:53,210
我就是说这就是你找到的全部

332
00:10:53,210 --> 00:10:53,929
就这些了

333
00:10:53,929 --> 00:10:55,129
哦好的

334
00:10:55,129 --> 00:10:56,629
谢谢啊

335
00:10:56,629 --> 00:10:57,950
我简直

336
00:10:57,950 --> 00:10:59,919
深入得令人惊叹

337
00:10:59,919 --> 00:11:04,659
通常这些人总在找小问题

338
00:11:04,659 --> 00:11:05,639
我就是说就是这样

339
00:11:05,639 --> 00:11:07,639
你花了数周或数月调查

340
00:11:07,639 --> 00:11:08,480
这就是你找到的全部

341
00:11:08,480 --> 00:11:09,639
我其实感到

342
00:11:09,639 --> 00:11:10,320
如果有的话

343
00:11:10,320 --> 00:11:14,220
我今天对索菲的信心反而更强了

344
00:11:14,220 --> 00:11:15,519
基于这份简报

345
00:11:15,519 --> 00:11:16,419
因为我觉得

346
00:11:16,419 --> 00:11:18,000
这就是我找到的全部了

347
00:11:18,000 --> 00:11:20,480
你知道这就是我的观点

348
00:11:20,480 --> 00:11:22,440
我不介意和我的公司一起经历这些

349
00:11:22,440 --> 00:11:23,919
其实我还挺喜欢的

350
00:11:23,919 --> 00:11:25,220
但这是来自

351
00:11:25,220 --> 00:11:26,200
有人已经掌控了市场

352
00:11:26,200 --> 00:11:28,629
现在已经有十七多年了

353
00:11:28,629 --> 00:11:30,990
我认为那些对此感到沮丧的人

354
00:11:30,990 --> 00:11:34,029
可能是持有SOI看涨期权的人

355
00:11:34,029 --> 00:11:37,990
你可能持有SOI十七美元的看涨期权

356
00:11:37,990 --> 00:11:39,830
你在查看看跌报告

357
00:11:39,830 --> 00:11:40,370
你可能会说

358
00:11:40,370 --> 00:11:42,370
你会直接把我的煎饼翻到地上

359
00:11:42,370 --> 00:11:43,289
兄弟别闹了

360
00:11:43,289 --> 00:11:45,549
这就是为什么别碰这些麻烦事

361
00:11:45,549 --> 00:11:46,090
你知道的

362
00:11:46,090 --> 00:11:47,570
只需长期持有优质公司

363
00:11:47,570 --> 00:11:49,159
你想要拥有

364
00:11:49,159 --> 00:11:50,279
你深信不疑

365
00:11:50,279 --> 00:11:52,019
你相信它未来一年会大幅上涨

366
00:11:52,019 --> 00:11:53,100
在未来几年

367
00:11:53,100 --> 00:11:54,659
直接买入股票

368
00:11:54,659 --> 00:11:56,350
不要玩看涨期权

369
00:11:56,350 --> 00:11:57,870
你无法预测市场变化

370
00:11:57,870 --> 00:11:59,110
市场可能随时崩盘

371
00:11:59,110 --> 00:11:59,590
我们清楚

372
00:11:59,590 --> 00:12:02,710
P500可能在未来一年下跌20-30-40%

373
00:12:02,710 --> 00:12:04,529
但这种情况大概率不会发生

374
00:12:04,529 --> 00:12:05,690
但有可能发生

375
00:12:05,690 --> 00:12:07,409
那你的看涨期权怎么办

376
00:12:07,409 --> 00:12:09,120
它们都会过期作废

377
00:12:09,120 --> 00:12:09,980
你知道的

378
00:12:09,980 --> 00:12:11,639
更多关于SOI的看跌报告

379
00:12:11,639 --> 00:12:13,879
人们因此感到恐慌

380
00:12:13,879 --> 00:12:15,820
抛售股票不再买入

381
00:12:15,820 --> 00:12:19,500
股价就在15-20美元区间徘徊

382
00:12:19,500 --> 00:12:20,279
而你拥有

383
00:12:20,279 --> 00:12:20,799
你知道的

384
00:12:20,799 --> 00:12:22,120
一年后到期的期权

385
00:12:22,120 --> 00:12:23,059
三十美元的行权价

386
00:12:23,059 --> 00:12:23,720
不是这样的

387
00:12:23,720 --> 00:12:24,740
我们不玩这种游戏

388
00:12:24,740 --> 00:12:28,109
兄弟专注买入优质公司

389
00:12:28,109 --> 00:12:32,129
长期持有你热爱的伟大公司股票，保持简单

390
00:12:32,129 --> 00:12:32,908
正如所说

391
00:12:32,908 --> 00:12:33,629
保持简单

392
00:12:33,629 --> 00:12:33,969
愚蠢

393
00:12:33,969 --> 00:12:36,940
无需复杂化游戏让自己焦虑

394
00:12:36,940 --> 00:12:38,320
因为现在有做空报告

395
00:12:38,320 --> 00:12:40,539
你的股票短期受损

396
00:12:40,539 --> 00:12:41,519
对吧好的

397
00:12:41,519 --> 00:12:41,639
现在

398
00:12:41,639 --> 00:12:43,450
让我们来谈谈英伟达和AMD

399
00:12:43,450 --> 00:12:45,929
然后是SaaS股票，这里到底发生了什么

400
00:12:45,929 --> 00:12:47,529
我们即将迎来一些非常重要的内容

401
00:12:47,529 --> 00:12:48,809
股票下跌了30美分

402
00:12:48,809 --> 00:12:52,309
你的股票在过去十年上涨了22000%

403
00:12:52,309 --> 00:12:54,889
我们应该关注过去半小时的情况吗

404
00:12:56,960 --> 00:12:58,539
我猜我们都应该关心

405
00:12:58,539 --> 00:12:59,320
嗯

406
00:12:59,320 --> 00:13:04,870
你知道我们的增长正在更大规模加速

407
00:13:04,870 --> 00:13:05,850
这对很多人来说很意外

408
00:13:07,169 --> 00:13:11,200
终端市场正在快速增长

409
00:13:11,200 --> 00:13:13,200
你知道我们已经到了这个阶段

410
00:13:13,200 --> 00:13:15,649
AI被称为

411
00:13:15,649 --> 00:13:17,789
推理拐点的到来

412
00:13:17,789 --> 00:13:20,159
所以我们已经达到了推理拐点

413
00:13:20,159 --> 00:13:22,120
因此我们的增长正在加速

414
00:13:22,120 --> 00:13:24,359
我们希望所有人都能享受这个过程

415
00:13:24,359 --> 00:13:27,759
人们被5000亿到1万亿美元的数字困住了

416
00:13:27,759 --> 00:13:29,379
我喜欢2025年的表现

417
00:13:29,379 --> 00:13:29,958
2027年

418
00:13:29,958 --> 00:13:32,389
让我想起你提到

419
00:13:32,389 --> 00:13:34,590
昨天我们谈到迈克尔·戴尔时

420
00:13:34,590 --> 00:13:35,600
吉姆

421
00:13:35,600 --> 00:13:36,299
你还记得吗

422
00:13:36,299 --> 00:13:37,159
当我告诉你

423
00:13:37,159 --> 00:13:39,559
让我非常清晰地说明英伟达的问题

424
00:13:39,559 --> 00:13:41,639
让我用最简单的方式解释

425
00:13:41,639 --> 00:13:44,019
我现在正在观看的观众

426
00:13:44,019 --> 00:13:44,399
好的

427
00:13:44,399 --> 00:13:47,159
英伟达的数据非常惊人

428
00:13:47,159 --> 00:13:48,460
将继续保持惊人表现

429
00:13:48,460 --> 00:13:51,009
我们都清楚问题所在

430
00:13:51,009 --> 00:13:55,690
就像街对面的孩子有个柠檬水摊

431
00:13:55,690 --> 00:13:56,409
好的

432
00:13:56,409 --> 00:13:58,450
你知道的

433
00:13:58,450 --> 00:14:03,068
假设只有少数公司或少数公司

434
00:14:03,068 --> 00:14:06,279
只有少数邻居能大量购买柠檬水

435
00:14:06,279 --> 00:14:08,700
嗯疯狂抢购他们的柠檬水

436
00:14:08,700 --> 00:14:09,419
对吧

437
00:14:09,419 --> 00:14:10,759
这些

438
00:14:10,759 --> 00:14:14,480
只有少数公司能实质性改变数据

439
00:14:14,480 --> 00:14:14,740
对吧

440
00:14:14,740 --> 00:14:16,080
只有少数邻居

441
00:14:16,080 --> 00:14:16,879
我一直在说这些公司

442
00:14:16,879 --> 00:14:18,179
只有少数邻居

443
00:14:18,179 --> 00:14:19,299
我们都在疯狂花钱

444
00:14:19,299 --> 00:14:20,879
在这个柠檬水摊上

445
00:14:20,879 --> 00:14:25,129
很快我们就要支付额外费用

446
00:14:25,129 --> 00:14:28,789
从我们的业务或

447
00:14:28,789 --> 00:14:30,509
你知道我们的工作

448
00:14:30,509 --> 00:14:32,710
每一笔钱现在都到了临界点

449
00:14:32,710 --> 00:14:35,690
所有资金都将流向柠檬树摊位

450
00:14:35,690 --> 00:14:35,950
现在

451
00:14:35,950 --> 00:14:39,190
此时此刻的孩子们将会面临更多困难

452
00:14:39,190 --> 00:14:41,589
比如收入增长受阻

453
00:14:41,589 --> 00:14:42,308
因为

454
00:14:42,308 --> 00:14:45,729
所有可能在柠檬摊位大量消费的人

455
00:14:45,729 --> 00:14:47,600
这些人的消费会显著改变数据

456
00:14:47,600 --> 00:14:49,460
他们已经花到了最大支出额度

457
00:14:49,460 --> 00:14:51,779
要在柠檬摊位多消费

458
00:14:51,779 --> 00:14:54,440
只能疯狂借贷

459
00:14:54,440 --> 00:14:56,409
你知道的

460
00:14:56,409 --> 00:14:57,490
没人愿意这么做

461
00:14:57,490 --> 00:14:58,009
本质上

462
00:14:58,009 --> 00:14:59,610
这就是当前的情况

463
00:14:59,610 --> 00:15:00,529
对于英伟达来说

464
00:15:00,529 --> 00:15:02,210
像Meta这样的公司

465
00:15:02,210 --> 00:15:02,789
像谷歌

466
00:15:02,789 --> 00:15:03,629
像亚马逊

467
00:15:03,629 --> 00:15:07,659
所有对英伟达数据影响重大的巨头

468
00:15:07,659 --> 00:15:09,820
它们基本已达到支出上限

469
00:15:09,820 --> 00:15:10,279
现在

470
00:15:10,279 --> 00:15:13,139
它们能在这些芯片上投入的金额

471
00:15:13,139 --> 00:15:15,559
它们没有多余的钱了

472
00:15:15,559 --> 00:15:20,259
基本上要用全部现金流投入资本支出

473
00:15:20,259 --> 00:15:21,320
此时此刻

474
00:15:21,320 --> 00:15:25,100
它们没有大量额外支出的空间

475
00:15:25,100 --> 00:15:26,159
而他们

476
00:15:26,159 --> 00:15:27,399
你可能会说

477
00:15:27,399 --> 00:15:29,679
如果小公司购买英伟达芯片

478
00:15:29,679 --> 00:15:29,919
是的

479
00:15:29,919 --> 00:15:33,139
但这不会实质性改变英伟达业务

480
00:15:33,139 --> 00:15:35,049
尤其是营收规模如此庞大时

481
00:15:35,049 --> 00:15:35,889
现在我们谈论的

482
00:15:35,889 --> 00:15:40,450
是一家年营收数百亿甚至上千亿的公司

483
00:15:40,450 --> 00:15:45,049
某个成功的小型AI公司可能花费

484
00:15:45,049 --> 00:15:45,549
你知道的

485
00:15:45,549 --> 00:15:48,840
四亿美元购买英伟达芯片

486
00:15:48,840 --> 00:15:50,279
只是微不足道

487
00:15:50,799 --> 00:15:53,000
詹森提到要投入万亿以上

488
00:15:53,000 --> 00:15:53,460
而

489
00:15:53,460 --> 00:15:53,940
你知道的

490
00:15:53,940 --> 00:15:55,340
未来几年采购几批芯片

491
00:15:55,340 --> 00:15:57,700
基本上在这两年期间

492
00:15:57,700 --> 00:15:59,539
四亿美元的芯片订单

493
00:15:59,539 --> 00:16:01,100
对英伟达来说毫无影响

494
00:16:01,100 --> 00:16:02,500
这就是问题所在

495
00:16:02,500 --> 00:16:06,129
希望我用最简单的方式解释清楚

496
00:16:06,129 --> 00:16:08,809
希望这个解释能让大家明白

497
00:16:08,809 --> 00:16:09,929
评论区告诉我你的看法

498
00:16:09,929 --> 00:16:10,330
好的

499
00:16:10,330 --> 00:16:12,440
查克可能在那里

500
00:16:12,440 --> 00:16:13,379
我说，嗯

501
00:16:13,379 --> 00:16:13,679
是的

502
00:16:13,679 --> 00:16:15,740
我十年前就做了

503
00:16:15,740 --> 00:16:17,190
他说尼莫克拉

504
00:16:17,190 --> 00:16:17,690
他说

505
00:16:17,690 --> 00:16:18,389
尼莫克拉

506
00:16:18,389 --> 00:16:19,850
这可能更大

507
00:16:19,850 --> 00:16:21,889
我每天早上以尼莫克拉开始

508
00:16:26,649 --> 00:16:29,450
而不是涉及呃

509
00:16:29,450 --> 00:16:30,909
过去五分钟的边际变化

510
00:16:31,909 --> 00:16:32,110
嗯

511
00:16:32,110 --> 00:16:35,620
首先我在演讲中提到

512
00:16:35,620 --> 00:16:39,889
并引用了之前使用过的数据

513
00:16:39,889 --> 00:16:42,289
因为我需要给所有人一个基准

514
00:16:42,289 --> 00:16:44,129
所以去年

515
00:16:44,129 --> 00:16:51,639
我说在这个时间节点我们有五百亿美元的高确定性

516
00:16:51,639 --> 00:16:56,359
关于黑威尔和鲁因的可见性

517
00:16:56,359 --> 00:16:57,198
就是这样

518
00:16:57,198 --> 00:16:59,979
我们公司当然还销售其他产品

519
00:16:59,979 --> 00:17:01,219
你知道的

520
00:17:01,219 --> 00:17:01,918
CPU

521
00:17:01,918 --> 00:17:03,359
我们有存储处理器

522
00:17:03,359 --> 00:17:04,199
我们有网络设备

523
00:17:04,199 --> 00:17:06,809
所以我们销售各种产品

524
00:17:06,809 --> 00:17:09,829
除此之外我们还有五百亿美元的可见性

525
00:17:09,829 --> 00:17:12,578
现在在这个时间节点

526
00:17:12,578 --> 00:17:17,489
还有两个月到2027年底

527
00:17:17,489 --> 00:17:20,519
我们已经具备高确定性

528
00:17:20,519 --> 00:17:21,319
高确定性

529
00:17:21,319 --> 00:17:26,569
可见性达到一千亿美元以上的黑威尔和鲁因

530
00:17:26,569 --> 00:17:27,509
仅此而已

531
00:17:27,509 --> 00:17:28,789
只有黑威尔和鲁因

532
00:17:28,789 --> 00:17:33,559
我们仍有两个月的新订单将至

533
00:17:33,559 --> 00:17:36,519
我们还有英伟达销售的所有其他产品

534
00:17:36,519 --> 00:17:39,630
因此我想给所有人一个基准

535
00:17:39,630 --> 00:17:44,309
生态系统需要了解英伟达的发展方向

536
00:17:44,309 --> 00:17:46,150
当然我们与供应链协同规划

537
00:17:46,150 --> 00:17:48,289
我们协助客户进行产能建设

538
00:17:48,289 --> 00:17:50,569
我们准备所有技术合作伙伴

539
00:17:50,569 --> 00:17:56,650
以支持一千亿美元的黑威尔和鲁因

540
00:17:56,650 --> 00:17:59,029
这真是了不起

541
00:17:59,029 --> 00:18:02,319
你还在关注英伟达股票

542
00:18:02,319 --> 00:18:02,980
你正在查看

543
00:18:02,980 --> 00:18:07,170
有人出来提到今年到明年之间一千亿美元的增长

544
00:18:07,170 --> 00:18:08,329
仅靠几款芯片

545
00:18:08,329 --> 00:18:12,230
这太疯狂了对吧 股价短期内为何不动

546
00:18:12,230 --> 00:18:13,650
听着

547
00:18:13,980 --> 00:18:17,579
他们完全扼杀了史诗级数据的悬念

548
00:18:17,579 --> 00:18:20,339
英伟达股价真的非常令人兴奋

549
00:18:20,339 --> 00:18:26,960
从2022年末到2024年末

550
00:18:26,960 --> 00:18:31,430
当时股价如此激动人心的原因在于你并不清楚

551
00:18:31,430 --> 00:18:33,970
英伟达的潜力究竟有多大

552
00:18:33,970 --> 00:18:35,710
他们带来的这些数据

553
00:18:35,710 --> 00:18:37,369
简直令人目瞪口呆

554
00:18:37,369 --> 00:18:39,539
他们刚刚公布了什么数据

555
00:18:39,539 --> 00:18:41,039
他们是如何实现如此增长的

556
00:18:41,039 --> 00:18:43,569
现在简直令人兴奋

557
00:18:43,569 --> 00:18:44,970
他直接解释道

558
00:18:44,970 --> 00:18:46,309
大致这就是我们的计划

559
00:18:46,309 --> 00:18:46,809
你知道的

560
00:18:46,809 --> 00:18:49,779
现在你可以从另一端业务测算数据

561
00:18:49,779 --> 00:18:51,319
或许他们会在此处出售一些芯片

562
00:18:51,319 --> 00:18:53,670
但这完全扼杀了所有期待

563
00:18:53,670 --> 00:18:55,190
从短期来看

564
00:18:55,190 --> 00:18:55,829
这对投资者不利

565
00:18:55,829 --> 00:18:57,029
如果你是长期投资者

566
00:18:57,029 --> 00:18:57,890
我认为这是好事

567
00:18:57,890 --> 00:18:58,369
不过

568
00:18:58,369 --> 00:19:00,670
我认为更重要的是

569
00:19:00,670 --> 00:19:01,319
你知道的

570
00:19:01,319 --> 00:19:05,679
英伟达未来一年或几年的预期表现

571
00:19:05,679 --> 00:19:06,878
在营收方面

572
00:19:06,878 --> 00:19:09,618
从长期角度看这非常有帮助

573
00:19:09,618 --> 00:19:11,078
但从短期来看

574
00:19:11,078 --> 00:19:15,359
这完全扼杀了兴奋感

575
00:19:15,359 --> 00:19:17,319
这就是我的看法

576
00:19:17,319 --> 00:19:18,099
哦，明白了

577
00:19:18,099 --> 00:19:19,900
现在我们已经清楚了

578
00:19:19,900 --> 00:19:21,450
十二月的情况

579
00:19:21,450 --> 00:19:22,970
我昨天听到的消息

580
00:19:22,970 --> 00:19:29,460
我最大的收获是它不会让你赚钱

581
00:19:29,460 --> 00:19:31,359
但会赚取巨额利润

582
00:19:31,359 --> 00:19:33,220
而这并不包含在万亿规模内

583
00:19:33,220 --> 00:19:33,960
不

584
00:19:33,960 --> 00:19:34,400
是的

585
00:19:34,400 --> 00:19:36,859
我们开始生产布洛克三型

586
00:19:36,859 --> 00:19:38,640
这是一家非常特别的公司

587
00:19:38,640 --> 00:19:42,250
并且我们将向各部分添加格洛克

588
00:19:42,250 --> 00:19:45,059
薇拉·鲁宾如此杰出以至于

589
00:19:45,059 --> 00:19:48,220
很难为薇拉·鲁宾增加价值

590
00:19:48,220 --> 00:19:50,700
这台引擎如此强大且

591
00:19:50,700 --> 00:19:52,529
但工作负载的某个部分

592
00:19:52,529 --> 00:19:56,950
我认为有25%的工作负载可以添加格洛克

593
00:19:56,950 --> 00:20:03,119
对于25%的工作负载我们可以进一步提升机会

594
00:20:03,119 --> 00:20:06,099
因此人们会为此付费

595
00:20:06,099 --> 00:20:06,960
是的

596
00:20:06,960 --> 00:20:07,759
你知道的

597
00:20:07,759 --> 00:20:12,598
英伟达显然是一家令人惊叹的公司

598
00:20:12,799 --> 00:20:13,400
你知道的

599
00:20:13,400 --> 00:20:14,160
传奇

600
00:20:14,160 --> 00:20:15,680
这就是我眼中的传奇

601
00:20:15,680 --> 00:20:15,940
好的

602
00:20:15,940 --> 00:20:17,480
这里x有什么动态

603
00:20:17,480 --> 00:20:18,799
我们来分析一下

604
00:20:18,799 --> 00:20:19,700
好的在这里

605
00:20:19,700 --> 00:20:20,359
我们开始行动

606
00:20:20,359 --> 00:20:21,929
女士们先生们

607
00:20:22,608 --> 00:20:26,429
市场开始质疑莱西·宋的价值主张

608
00:20:26,429 --> 00:20:27,328
这非常重要

609
00:20:27,328 --> 00:20:29,128
如果你在购买任何SaaS股票

610
00:20:29,128 --> 00:20:32,150
你是否考虑在这里买入SaaS股票

611
00:20:32,150 --> 00:20:33,470
过去几个月的公司

612
00:20:33,470 --> 00:20:35,650
随着工作流程的兴起

613
00:20:35,650 --> 00:20:38,869
许多人认为这些平台可以被轻易取代

614
00:20:38,869 --> 00:20:40,210
我想知道你如何看待

615
00:20:40,210 --> 00:20:44,329
AI如何与这些软件公司合作证明其价值

616
00:20:44,849 --> 00:20:48,150
请告诉散户投资者

617
00:20:48,150 --> 00:20:52,950
他们完全错了而且

618
00:20:52,950 --> 00:20:55,910
这不是我保护你的责任

619
00:20:56,529 --> 00:20:59,210
你知道但对错分明

620
00:20:59,210 --> 00:20:59,890
错误

621
00:20:59,890 --> 00:21:00,869
嗯

622
00:21:00,869 --> 00:21:02,029
让我想想

623
00:21:02,029 --> 00:21:04,210
我们用一个例子

624
00:21:04,739 --> 00:21:05,919
番茄

625
00:21:05,919 --> 00:21:08,099
或者Cadence

626
00:21:08,099 --> 00:21:09,240
或者任何弟子

627
00:21:10,680 --> 00:21:13,279
他们的业务局限

628
00:21:14,980 --> 00:21:16,460
屁股坐在座位上

629
00:21:16,460 --> 00:21:18,859
工程师数量有限

630
00:21:19,388 --> 00:21:21,128
如果有更多工程师

631
00:21:21,128 --> 00:21:22,628
他们将授权更多工具

632
00:21:23,949 --> 00:21:26,250
如果有更多水管工

633
00:21:26,250 --> 00:21:26,970
你就能生活

634
00:21:26,970 --> 00:21:27,990
获得更多工具

635
00:21:27,990 --> 00:21:29,470
如果有木匠

636
00:21:29,470 --> 00:21:30,890
获得更多工具

637
00:21:30,890 --> 00:21:31,569
是的

638
00:21:31,569 --> 00:21:32,069
相当不错

639
00:21:32,069 --> 00:21:36,209
所以如果有更多工程师

640
00:21:36,209 --> 00:21:37,729
那么就会有人

641
00:21:37,729 --> 00:21:40,148
但未来百分之百要考虑

642
00:21:40,148 --> 00:21:41,519
简单来说

643
00:21:41,519 --> 00:21:43,079
想象一下

644
00:21:43,079 --> 00:21:46,569
你有一个小镇的五金店

645
00:21:46,569 --> 00:21:49,699
对吧而且你知道

646
00:21:49,699 --> 00:21:55,000
你的小镇人口只有大约一百人对吧

647
00:21:55,000 --> 00:21:58,660
在这些人中，真正会来你的五金店买东西的人有多少呢

648
00:21:58,660 --> 00:21:59,789
所以是成年人

649
00:21:59,789 --> 00:22:03,109
假设城市里的成年人数量大约有五十人左右

650
00:22:03,109 --> 00:22:06,809
剩下的都是儿童或超级年长的人

651
00:22:06,809 --> 00:22:07,569
他们甚至

652
00:22:07,569 --> 00:22:07,970
你知道的

653
00:22:07,970 --> 00:22:08,849
去五金店购物

654
00:22:08,849 --> 00:22:10,849
所以你可能有五十位顾客

655
00:22:10,849 --> 00:22:15,710
或许你的五金店能卖出更多商品

656
00:22:15,710 --> 00:22:18,329
如果人口数量更大

657
00:22:18,329 --> 00:22:20,690
如果镇上住着五百人会怎样

658
00:22:20,690 --> 00:22:22,309
对或五千人左右

659
00:22:22,309 --> 00:22:25,859
这将创造更大的销售机会

660
00:22:25,859 --> 00:22:27,140
所以没错

661
00:22:27,140 --> 00:22:27,799
百分之百正确

662
00:22:27,799 --> 00:22:30,140
这就是他在这里暗示的重点

663
00:22:30,140 --> 00:22:31,920
作为论点是

664
00:22:32,480 --> 00:22:37,890
这些软件公司一直受限于用户数量

665
00:22:37,890 --> 00:22:39,349
因为需要真实人类使用

666
00:22:39,349 --> 00:22:41,369
但若代理开始这样做

667
00:22:41,369 --> 00:22:43,509
理论上可以无限扩展

668
00:22:43,509 --> 00:22:45,190
我们将拥有getic工程师

669
00:22:45,190 --> 00:22:48,970
工程师要做什么？制造晶体管

670
00:22:48,970 --> 00:22:50,269
不

671
00:22:50,750 --> 00:22:54,369
工程师会使用我们相同的工具

672
00:22:54,369 --> 00:22:58,250
因为使用工具后

673
00:22:58,250 --> 00:23:03,159
需要将其还原为我能理解的结构化数据

674
00:23:03,679 --> 00:23:07,038
否则如果我无法理解或控制

675
00:23:07,038 --> 00:23:08,699
这又回到可控性问题

676
00:23:08,699 --> 00:23:11,539
我们不断回到这个主题

677
00:23:11,539 --> 00:23:12,279
即将消亡

678
00:23:12,279 --> 00:23:14,180
因为代理存在

679
00:23:14,180 --> 00:23:15,299
没有续集

680
00:23:15,299 --> 00:23:19,849
这就是商业真实性的根基将被摧毁

681
00:23:19,849 --> 00:23:22,609
当代理完成工作后

682
00:23:22,609 --> 00:23:25,720
请将成果放回续集

683
00:23:26,119 --> 00:23:28,000
这样我需要审阅时

684
00:23:28,000 --> 00:23:31,960
可以在代理设计完成后审阅

685
00:23:31,960 --> 00:23:34,429
它会使用综合工具和数据

686
00:23:34,429 --> 00:23:37,749
并将结果存入文件系统和数据结构

687
00:23:37,749 --> 00:23:39,420
设计的真实基础

688
00:23:39,420 --> 00:23:40,059
明白了

689
00:23:40,059 --> 00:23:40,960
这就是综合工具

690
00:23:40,960 --> 00:23:43,339
如果我要完全复现

691
00:23:44,380 --> 00:23:46,289
而非概率性生成

692
00:23:46,289 --> 00:23:47,130
你知道的

693
00:23:47,130 --> 00:23:48,250
如果我拥有一个代理

694
00:23:48,250 --> 00:23:52,069
用AI深度学习生成芯片

695
00:23:52,069 --> 00:23:54,809
大概会正常工作

696
00:23:55,289 --> 00:23:57,089
但不能只是大概有效

697
00:23:57,089 --> 00:23:58,829
我需要准确说明它的运作方式

698
00:23:58,829 --> 00:23:59,369
是的

699
00:23:59,369 --> 00:24:03,920
因此我需要具体的事实依据

700
00:24:03,920 --> 00:24:05,359
还有这些工具

701
00:24:05,359 --> 00:24:09,009
概要工具现在能保持连贯性

702
00:24:09,009 --> 00:24:10,630
因为我有这些代理

703
00:24:10,630 --> 00:24:15,150
我们拥有的概要工具许可证数量可能会激增

704
00:24:15,150 --> 00:24:15,529
是啊

705
00:24:15,529 --> 00:24:17,509
不是反过来

706
00:24:19,480 --> 00:24:21,019
人们有这种奇怪的想法

707
00:24:21,019 --> 00:24:25,279
认为AI会从零开始生成晶体管

708
00:24:25,279 --> 00:24:28,089
或者像纳米香蕉那样

709
00:24:28,089 --> 00:24:29,279
嗯哼

710
00:24:29,279 --> 00:24:30,099
我要去

711
00:24:30,099 --> 00:24:31,240
你知道吗

712
00:24:31,240 --> 00:24:34,160
他们想象视频工程师会下线

713
00:24:34,160 --> 00:24:35,339
有点像

714
00:24:35,339 --> 00:24:38,400
你知道的

715
00:24:38,640 --> 00:24:39,039
哦

716
00:24:39,039 --> 00:24:41,920
我喜欢这个剧本

717
00:24:41,940 --> 00:24:43,880
但你无法设计芯片

718
00:24:43,880 --> 00:24:44,220
不过

719
00:24:44,220 --> 00:24:48,640
你们告诉我必须逐层精细打造

720
00:24:48,640 --> 00:24:50,519
因为需要可验证

721
00:24:50,519 --> 00:24:51,940
需要可理解

722
00:24:51,940 --> 00:24:53,440
需要可重复

723
00:24:53,440 --> 00:24:55,009
需要可控

724
00:24:55,410 --> 00:24:58,630
所有这些都不能是概率性的

725
00:24:58,630 --> 00:24:59,250
没错

726
00:24:59,250 --> 00:25:00,009
百分之百

727
00:25:00,009 --> 00:25:00,809
嗯

728
00:25:00,809 --> 00:25:01,210
嗯

729
00:25:01,210 --> 00:25:03,839
詹森还有另一种精辟的表达方式

730
00:25:04,359 --> 00:25:04,960
他说

731
00:25:04,960 --> 00:25:05,180
好的

732
00:25:05,180 --> 00:25:05,440
想象一下

733
00:25:05,440 --> 00:25:06,119
你拥有

734
00:25:06,119 --> 00:25:06,880
你知道的

735
00:25:06,880 --> 00:25:08,000
奥图机器人

736
00:25:08,000 --> 00:25:09,279
或者任何机器人

737
00:25:09,279 --> 00:25:09,859
嗯

738
00:25:09,859 --> 00:25:11,250
你知道的

739
00:25:11,250 --> 00:25:12,609
谁知道需要多少年

740
00:25:12,609 --> 00:25:13,809
五年后

741
00:25:13,809 --> 00:25:14,410
十年后

742
00:25:14,410 --> 00:25:15,150
二十年后

743
00:25:15,150 --> 00:25:15,569
三十年

744
00:25:15,569 --> 00:25:16,400
随便怎样

745
00:25:16,400 --> 00:25:16,900
好吧

746
00:25:16,900 --> 00:25:18,240
最终我们会有家庭机器人

747
00:25:18,240 --> 00:25:21,420
然后机器人就会出去做庭院景观工作

748
00:25:21,420 --> 00:25:22,990
你知道它们会

749
00:25:22,990 --> 00:25:24,190
它们在洗碗

750
00:25:24,190 --> 00:25:24,970
还有所有这些事情

751
00:25:24,970 --> 00:25:26,710
你知道机器人会说

752
00:25:26,710 --> 00:25:27,210
你知道吗

753
00:25:27,210 --> 00:25:32,319
让我先建一个全新的洗碗机

754
00:25:32,319 --> 00:25:32,900
你知道不用

755
00:25:32,900 --> 00:25:36,180
它会使用现有的设备

756
00:25:36,180 --> 00:25:36,940
它会说

757
00:25:36,940 --> 00:25:37,339
你知道吗

758
00:25:37,339 --> 00:25:40,788
让我先造全新的吹叶机

759
00:25:40,788 --> 00:25:41,189
不用

760
00:25:41,189 --> 00:25:42,608
它会使用吹叶机

761
00:25:42,608 --> 00:25:43,669
它会使用工具

762
00:25:43,669 --> 00:25:46,028
这就是人们认为的

763
00:25:46,028 --> 00:25:46,469
哦

764
00:25:46,469 --> 00:25:46,808
你知道的

765
00:25:46,808 --> 00:25:49,599
像销售团队会被取代

766
00:25:49,599 --> 00:25:50,279
因为他们有

767
00:25:50,279 --> 00:25:50,740
你知道的

768
00:25:50,740 --> 00:25:51,660
代理和我却说不

769
00:25:51,660 --> 00:25:53,589
代理会给他们带来巨大好处

770
00:25:53,589 --> 00:25:55,029
你开玩笑吗

771
00:25:55,029 --> 00:25:57,049
这将创造更大的

772
00:25:57,049 --> 00:25:58,630
长期市场机会

773
00:25:58,630 --> 00:26:00,309
更大的长期市场机会

774
00:26:00,309 --> 00:26:03,959
长期来看他们的工具会被更广泛使用

775
00:26:03,959 --> 00:26:10,369
为什么要去开发新的销售软件

776
00:26:10,369 --> 00:26:12,529
当现有的已经完美运行

777
00:26:12,529 --> 00:26:14,569
不需要另一个了对吧

778
00:26:14,569 --> 00:26:16,450
为什么要复制Slack

779
00:26:16,450 --> 00:26:17,549
我觉得这没道理

780
00:26:17,549 --> 00:26:18,170
最终会

781
00:26:18,170 --> 00:26:19,730
可能是个更差的产品

782
00:26:19,730 --> 00:26:24,349
并在组织内引发一大堆问题

783
00:26:24,349 --> 00:26:25,069
你开玩笑吗

784
00:26:25,069 --> 00:26:28,750
这太疯狂了，就像遍历公司

785
00:26:28,750 --> 00:26:31,029
整个关于

786
00:26:31,029 --> 00:26:31,349
哦

787
00:26:31,349 --> 00:26:33,868
所有人都被取代简直是荒谬

788
00:26:33,868 --> 00:26:35,648
绝对荒谬

789
00:26:35,648 --> 00:26:36,169
你知道的

790
00:26:36,169 --> 00:26:38,528
不过这就是当前市场状况

791
00:26:38,528 --> 00:26:38,929
好的

792
00:26:38,929 --> 00:26:39,528
接下来这个

793
00:26:39,528 --> 00:26:41,880
我觉得这对MD更重要

794
00:26:41,880 --> 00:26:43,819
如果你是AMD股东

795
00:26:43,819 --> 00:26:45,680
天啊亚历克斯·坎特里克

796
00:26:45,680 --> 00:26:46,240
亚历克斯

797
00:26:46,240 --> 00:26:46,859
很高兴见到你

798
00:26:46,859 --> 00:26:49,269
你是怎么在昨天的幕后交流中来到这里的

799
00:26:49,269 --> 00:26:51,549
不过我们还是谈谈英伟达吧

800
00:26:51,549 --> 00:26:52,470
这有点奇怪

801
00:26:52,470 --> 00:26:53,369
英伟达

802
00:26:53,369 --> 00:26:54,829
它必须得证明自己

803
00:26:54,829 --> 00:26:55,410
此刻而言

804
00:26:55,410 --> 00:26:56,638
它表现得如此出色

805
00:26:56,638 --> 00:26:58,719
以至于持续保持良好表现

806
00:26:58,719 --> 00:27:00,659
或许还能做得更好

807
00:27:00,659 --> 00:27:01,038
什么

808
00:27:01,038 --> 00:27:04,849
你在GTC上发现最值得注意的点是什么

809
00:27:05,410 --> 00:27:05,809
嗯

810
00:27:05,809 --> 00:27:09,049
首先这凸显了人们对这家公司的期待

811
00:27:09,049 --> 00:27:09,470
仅仅

812
00:27:09,470 --> 00:27:13,839
英伟达可以告诉你将在两年内靠两款产品赚一万亿

813
00:27:13,839 --> 00:27:16,450
或近似数字在未来两年

814
00:27:16,450 --> 00:27:18,509
但人们却视此为失望

815
00:27:18,509 --> 00:27:21,859
但我们需要考虑当前的环境

816
00:27:21,859 --> 00:27:23,019
我们现在身处一个世界

817
00:27:23,019 --> 00:27:28,619
今年科技巨头的资本支出预计达七百亿美元

818
00:27:28,619 --> 00:27:31,390
较去年四百亿有所增长

819
00:27:31,390 --> 00:27:34,049
所以2027年会如何发展还不清楚

820
00:27:34,049 --> 00:27:37,519
如果保持当前增速或达一万亿

821
00:27:37,519 --> 00:27:38,380
如果你关注

822
00:27:38,380 --> 00:27:43,259
未来两三年可能有1500至2000亿美元投入

823
00:27:43,259 --> 00:27:44,160
英伟达宣称

824
00:27:44,160 --> 00:27:47,569
我们将获得一万亿份额人们开始

825
00:27:47,569 --> 00:27:49,690
他们对这一目标表示失望

826
00:27:49,690 --> 00:27:52,679
但总体而言我认为公司状况良好

827
00:27:52,679 --> 00:27:55,460
他们将占据大部分资本支出

828
00:27:55,460 --> 00:27:57,779
随着AI持续加速

829
00:27:57,779 --> 00:28:01,380
我们昨天讨论了如今各种产品应用

830
00:28:01,380 --> 00:28:03,660
这些在数月前还不存在

831
00:28:03,660 --> 00:28:05,500
我认为英伟达将从中受益

832
00:28:05,500 --> 00:28:07,059
短期内将持续获利

833
00:28:07,059 --> 00:28:07,940
几个月前

834
00:28:07,940 --> 00:28:09,279
人们充满期待

835
00:28:09,279 --> 00:28:10,019
嗯

836
00:28:10,019 --> 00:28:11,589
推动英伟达股价上涨

837
00:28:11,589 --> 00:28:13,950
因为大量投资者买入

838
00:28:13,950 --> 00:28:16,650
人们对云服务商

839
00:28:16,650 --> 00:28:17,150
你知道的

840
00:28:17,150 --> 00:28:18,829
谷歌亚马逊

841
00:28:18,829 --> 00:28:19,329
嗯

842
00:28:19,329 --> 00:28:20,069
微软

843
00:28:20,069 --> 00:28:23,670
对他们在AI领域的增长潜力感到兴奋

844
00:28:23,670 --> 00:28:25,390
尽管当前支出巨大

845
00:28:25,390 --> 00:28:28,170
双方似乎有些冷淡

846
00:28:28,170 --> 00:28:34,289
如何平衡基础设施厂商如英伟达的机会

847
00:28:34,289 --> 00:28:37,410
尽管英伟达这个名字听起来有点像一个单一实体

848
00:28:37,410 --> 00:28:39,220
而在利润率方面

849
00:28:39,220 --> 00:28:39,779
呃

850
00:28:39,779 --> 00:28:44,230
这对那些投入大量资金建设基础设施的大公司意味着什么

851
00:28:44,430 --> 00:28:44,829
嗯

852
00:28:44,829 --> 00:28:46,269
首先确实有一场狂热

853
00:28:46,269 --> 00:28:46,750
呃

854
00:28:46,750 --> 00:28:48,769
过去几年的股票狂热

855
00:28:48,769 --> 00:28:51,559
人们看到英伟达展现出的强劲增长

856
00:28:51,559 --> 00:28:52,960
都想参与其中

857
00:28:52,960 --> 00:28:54,200
于是纷纷涌入

858
00:28:54,200 --> 00:28:55,160
但如今

859
00:28:55,160 --> 00:28:58,180
我们确实知道英伟达有大量收入预期

860
00:28:58,180 --> 00:29:00,900
但其增长似乎略有停滞

861
00:29:00,900 --> 00:29:02,400
至少在2026年

862
00:29:02,400 --> 00:29:03,909
所以我认为

863
00:29:03,909 --> 00:29:04,669
这里仍有增长空间

864
00:29:04,669 --> 00:29:07,779
但关于利润率将如何变化

865
00:29:07,779 --> 00:29:11,619
即使深入这一领域的人

866
00:29:11,619 --> 00:29:12,579
无论是芯片制造商

867
00:29:12,579 --> 00:29:13,960
应用开发者

868
00:29:13,960 --> 00:29:14,859
超大规模云服务商

869
00:29:14,859 --> 00:29:15,799
他们也不太确定

870
00:29:15,799 --> 00:29:18,279
每个类别都有其合理性

871
00:29:18,279 --> 00:29:19,420
每个类别

872
00:29:19,420 --> 00:29:21,970
都可能成为最大受益者

873
00:29:21,970 --> 00:29:22,950
或一无所获

874
00:29:22,950 --> 00:29:24,430
想想模型构建

875
00:29:24,430 --> 00:29:25,009
例如

876
00:29:25,009 --> 00:29:28,970
如果构建类似通用人工智能这样的强大系统

877
00:29:28,970 --> 00:29:30,970
这应该具有经济价值

878
00:29:30,970 --> 00:29:32,730
但如果别人构建了

879
00:29:32,730 --> 00:29:33,710
能否垄断

880
00:29:33,710 --> 00:29:35,250
能否收取溢价

881
00:29:35,250 --> 00:29:37,589
还是会被商品化 同样适用于芯片

882
00:29:37,589 --> 00:29:40,789
我们在英伟达的推理讨论中看到这一点

883
00:29:40,789 --> 00:29:42,660
现在人们认为

884
00:29:42,660 --> 00:29:44,680
随着AI更广泛采用

885
00:29:44,680 --> 00:29:45,920
英伟达应处于最佳位置

886
00:29:45,920 --> 00:29:47,380
我认为确实如此

887
00:29:47,380 --> 00:29:49,319
但当我们从训练转向推理

888
00:29:49,319 --> 00:29:51,059
竞争将大幅增加

889
00:29:51,059 --> 00:29:53,359
我的观点是这充满动态变化

890
00:29:53,359 --> 00:29:54,299
波动性极大

891
00:29:54,299 --> 00:29:55,730
快速演变

892
00:29:55,730 --> 00:29:59,349
投资者只能暂时保持警惕

893
00:29:59,349 --> 00:30:00,759
随着市场洗牌

894
00:30:00,759 --> 00:30:05,579
我们可能讨论不足的一个领域是数据管理软件

895
00:30:05,579 --> 00:30:10,900
它介于基础设施与应用之间 在此之前

896
00:30:10,900 --> 00:30:15,349
所以听好了 英伟达显然领先AMD数年

897
00:30:15,349 --> 00:30:17,309
现在终于进入了番茄酱阶段

898
00:30:17,309 --> 00:30:18,049
而且你知道的

899
00:30:18,049 --> 00:30:20,750
显然随着450系列即将投产

900
00:30:20,750 --> 00:30:21,210
呃

901
00:30:21,210 --> 00:30:25,130
并在2026年下半年以重大规模扩产

902
00:30:25,130 --> 00:30:25,569
这确实

903
00:30:25,569 --> 00:30:25,970
你知道的

904
00:30:25,970 --> 00:30:28,349
AMD终于来到这里

905
00:30:28,349 --> 00:30:31,390
你知道我们首次与英伟达同台竞技了

906
00:30:31,390 --> 00:30:33,089
这对AMD来说非常令人振奋

907
00:30:33,089 --> 00:30:37,220
但你必须理解训练与推理的区别

908
00:30:37,220 --> 00:30:38,019
想想训练过程

909
00:30:38,019 --> 00:30:41,229
这些模型需要训练，还要考虑推理应用

910
00:30:41,229 --> 00:30:42,449
人们实际使用它们的情况

911
00:30:42,449 --> 00:30:46,709
没错，Lisa Sue已将公司定位得非常强势

912
00:30:46,709 --> 00:30:48,279
她多次强调过这一点

913
00:30:48,279 --> 00:30:49,119
嗯

914
00:30:49,119 --> 00:30:50,180
在推理领域

915
00:30:50,180 --> 00:30:55,109
推理才是更广阔的长期机会

916
00:30:55,109 --> 00:30:59,720
当我们思考训练机会及过去几年的发展

917
00:30:59,720 --> 00:31:01,619
这确实是个绝佳机会

918
00:31:05,519 --> 00:31:07,240
这已是巨大机遇

919
00:31:07,240 --> 00:31:09,779
但要理解推理机会

920
00:31:09,779 --> 00:31:13,720
长期来看规模更大

921
00:31:13,720 --> 00:31:17,039
在AMD与英伟达之间

922
00:31:17,039 --> 00:31:18,319
令人兴奋的是

923
00:31:18,319 --> 00:31:25,130
AMD将在2026年及未来加速增长

924
00:31:25,130 --> 00:31:28,730
而英伟达营收增速将放缓

925
00:31:28,730 --> 00:31:29,309
好的

926
00:31:29,309 --> 00:31:30,930
营收增速放缓

927
00:31:30,930 --> 00:31:32,789
不是营收减少

928
00:31:32,789 --> 00:31:34,660
营收增速放缓

929
00:31:34,660 --> 00:31:37,680
这绝非争论话题

930
00:31:37,680 --> 00:31:38,920
这是事实

931
00:31:38,920 --> 00:31:42,130
若认为英伟达会奇迹般增长

932
00:31:42,130 --> 00:31:42,950
百分之百

933
00:31:42,950 --> 00:31:43,910
百分之百五

934
00:31:43,910 --> 00:31:44,730
百分之二百

935
00:31:44,730 --> 00:31:45,990
我不知道该说什么

936
00:31:45,990 --> 00:31:47,490
这在数学上不可能

937
00:31:47,490 --> 00:31:50,789
因为公司已接近现金流峰值

938
00:31:50,789 --> 00:31:53,519
并大量投入相关领域

939
00:31:53,519 --> 00:31:54,970
但说到AMD

940
00:31:54,970 --> 00:31:58,490
即将迎来重大营收增长加速

941
00:31:58,490 --> 00:31:59,349
纯属数学问题

942
00:31:59,349 --> 00:32:00,509
这不是主观意见

943
00:32:00,509 --> 00:32:02,109
这是客观数学

944
00:32:02,109 --> 00:32:05,029
这就是为什么AMD

945
00:32:05,029 --> 00:32:07,069
若从解放日以来观察

946
00:32:07,069 --> 00:32:09,349
AMD的表现显著优于

947
00:32:09,349 --> 00:32:10,420
英伟达

948
00:32:10,420 --> 00:32:12,740
但我觉得这仍处于早期阶段

949
00:32:12,740 --> 00:32:17,289
我认为在未来一两年内大家要做好准备

950
00:32:17,289 --> 00:32:19,210
AMD与英伟达的业绩差距

951
00:32:19,210 --> 00:32:20,650
将会非常惊人

952
00:32:20,650 --> 00:32:21,170
在我看来

953
00:32:21,170 --> 00:32:22,089
这并非贬低

954
00:32:22,089 --> 00:32:24,130
英伟达是一家了不起的公司

955
00:32:24,130 --> 00:32:26,388
我认为英伟达股票仍有机会

956
00:32:26,388 --> 00:32:30,788
但需理解AMD将在接下来的时间里崛起

957
00:32:30,788 --> 00:32:33,539
同时也涉及所有数据的存储方式

958
00:32:33,539 --> 00:32:36,059
适合AI应用的数据库

959
00:32:36,059 --> 00:32:36,859
如何进行管理

960
00:32:36,859 --> 00:32:37,680
数据湖

961
00:32:37,680 --> 00:32:39,839
像Snowflake这样的公司

962
00:32:40,440 --> 00:32:42,059
Datadog等

963
00:32:42,059 --> 00:32:45,599
如何评估它们在生态中的地位

964
00:32:45,599 --> 00:32:47,740
以及它们的估值

965
00:32:47,740 --> 00:32:52,039
或它们与其它公司的讨论热度

966
00:32:52,319 --> 00:32:52,640
嗯

967
00:32:52,640 --> 00:32:54,180
我认为它们将非常重要

968
00:32:54,180 --> 00:32:56,279
这引出了更广泛的讨论

969
00:32:56,279 --> 00:32:57,599
软件是否仍有价值

970
00:32:57,599 --> 00:32:59,119
还是会一文不值

971
00:32:59,119 --> 00:33:01,059
总之

972
00:33:01,059 --> 00:33:06,180
即使进入主要通过聊天机器人完成计算的时代

973
00:33:06,180 --> 00:33:09,109
仍需要优质数据作为基础

974
00:33:09,109 --> 00:33:11,390
并非所有数据都会存入

975
00:33:11,390 --> 00:33:12,950
ChatGPT的内存

976
00:33:12,950 --> 00:33:14,579
无论技术如何进步

977
00:33:14,579 --> 00:33:18,619
我认为这些中间层公司至关重要

978
00:33:18,619 --> 00:33:21,440
短期内不会被取代

979
00:33:21,440 --> 00:33:24,180
尽管可能会出现类似

980
00:33:24,180 --> 00:33:24,539
哦

981
00:33:24,539 --> 00:33:26,059
我可以开发一个应用

982
00:33:26,059 --> 00:33:27,119
这令人印象深刻

983
00:33:27,119 --> 00:33:29,640
投资者需要区分

984
00:33:29,640 --> 00:33:32,730
这些功能与核心基础设施

985
00:33:32,730 --> 00:33:36,670
如Snowflake或Datadog提供的服务

986
00:33:36,670 --> 00:33:37,250
是的

987
00:33:37,250 --> 00:33:43,929
关于AI的另一个重要点是人们认为

988
00:33:44,209 --> 00:33:44,769
你知道的

989
00:33:44,769 --> 00:33:47,368
人们见证了过去几年AI的进步

990
00:33:47,368 --> 00:33:48,429
这非常令人惊叹

991
00:33:48,429 --> 00:33:48,848
没错

992
00:33:48,848 --> 00:33:51,660
但接下来的难度会大幅增加

993
00:33:51,660 --> 00:33:52,900
人们可能会认为

994
00:33:52,900 --> 00:33:53,299
哦

995
00:33:53,299 --> 00:33:54,119
你知道的

996
00:33:54,119 --> 00:33:54,599
通用

997
00:33:54,599 --> 00:33:55,000
你知道的

998
00:33:55,000 --> 00:33:55,779
通用人工智能

999
00:33:55,779 --> 00:33:56,220
人工智能

1000
00:33:56,220 --> 00:33:56,779
AGI

1001
00:33:56,779 --> 00:33:57,259
就像

1002
00:33:57,259 --> 00:33:58,720
它就在不远处

1003
00:33:58,720 --> 00:33:59,259
没错

1004
00:33:59,259 --> 00:34:00,019
可能

1005
00:34:00,019 --> 00:34:01,799
但也不一定

1006
00:34:01,799 --> 00:34:03,859
可能要很久以后

1007
00:34:03,859 --> 00:34:04,720
关于

1008
00:34:04,720 --> 00:34:05,279
你知道的

1009
00:34:05,279 --> 00:34:09,130
真的在这些语境下讨论

1010
00:34:10,469 --> 00:34:12,449
我这样告诉你

1011
00:34:12,449 --> 00:34:14,090
你们中有些人有孩子

1012
00:34:14,090 --> 00:34:15,949
当你有了孩子

1013
00:34:15,949 --> 00:34:19,630
通常前几个月里他们基本上

1014
00:34:19,630 --> 00:34:20,050
什么也不做

1015
00:34:20,050 --> 00:34:22,630
只是躺着，换尿布、喂食

1016
00:34:22,630 --> 00:34:24,199
其实变化不大对吧

1017
00:34:24,199 --> 00:34:26,219
大约六个月时

1018
00:34:26,219 --> 00:34:29,088
开始做出面部表情

1019
00:34:29,088 --> 00:34:30,829
活动更多了对吧

1020
00:34:30,829 --> 00:34:31,389
呃

1021
00:34:31,389 --> 00:34:33,190
最终开始爬行

1022
00:34:33,190 --> 00:34:34,989
然后开始走路

1023
00:34:34,989 --> 00:34:36,369
接着你可能开始跑步

1024
00:34:36,369 --> 00:34:39,269
随着月龄增长和岁月流逝

1025
00:34:39,269 --> 00:34:41,349
你看到孩子不断进步

1026
00:34:41,349 --> 00:34:41,909
你会说

1027
00:34:41,909 --> 00:34:42,690
天啊

1028
00:34:42,690 --> 00:34:45,949
他们马上就能飞了

1029
00:34:45,949 --> 00:34:46,610
因为你看

1030
00:34:46,610 --> 00:34:48,090
他们从一无所有

1031
00:34:48,090 --> 00:34:49,469
什么都不会到

1032
00:34:49,469 --> 00:34:50,730
开始活动到

1033
00:34:50,730 --> 00:34:53,409
开始爬行到如今开始走路

1034
00:34:53,409 --> 00:34:54,349
现在开始跑步

1035
00:34:54,349 --> 00:34:57,550
再过一两年就能像鸟儿一样飞了

1036
00:34:57,550 --> 00:34:59,769
当然这显然不现实

1037
00:34:59,769 --> 00:35:05,739
同样的现象可能发生在新事物上

1038
00:35:05,739 --> 00:35:06,099
你知道的

1039
00:35:06,099 --> 00:35:07,219
如果回到以前

1040
00:35:07,219 --> 00:35:10,539
现在大家都知道常识是

1041
00:35:10,539 --> 00:35:11,090
哦

1042
00:35:11,090 --> 00:35:11,570
你知道的

1043
00:35:11,570 --> 00:35:14,389
人类在最初几年快速学习

1044
00:35:14,389 --> 00:35:14,809
并不意味着

1045
00:35:14,809 --> 00:35:16,670
现在也会变得超级聪明

1046
00:35:16,670 --> 00:35:17,699
而且它会飞

1047
00:35:17,699 --> 00:35:19,360
你明白要成为超级英雄

1048
00:35:19,360 --> 00:35:21,010
这可不是正常运作的方式

1049
00:35:21,010 --> 00:35:23,570
这里的技术问题也是如此

1050
00:35:23,570 --> 00:35:25,130
这令人印象深刻极了

1051
00:35:25,130 --> 00:35:27,150
前几年构建过程的进展

1052
00:35:27,150 --> 00:35:29,360
但如果你这样想

1053
00:35:29,360 --> 00:35:29,539
哦

1054
00:35:29,539 --> 00:35:30,860
它只会以这个速度继续发展

1055
00:35:30,860 --> 00:35:31,420
或者类似的情况

1056
00:35:31,420 --> 00:35:32,860
不，这是巨大减速

1057
00:35:32,860 --> 00:35:34,139
然后人们会说

1058
00:35:34,139 --> 00:35:35,150
哦天啊

1059
00:35:35,150 --> 00:35:36,789
这太难了

1060
00:35:36,789 --> 00:35:40,230
为什么得不到我们想要的准确结果

1061
00:35:40,230 --> 00:35:41,429
嗯是的

1062
00:35:41,429 --> 00:35:43,650
我自己越来越频繁使用AI

1063
00:35:43,650 --> 00:35:45,929
我反而越来越失望

1064
00:35:45,929 --> 00:35:48,730
其实刚开始使用时

1065
00:35:48,730 --> 00:35:49,710
哎呀

1066
00:35:49,710 --> 00:35:50,960
这太疯狂了

1067
00:35:50,960 --> 00:35:52,760
越用越发现

1068
00:35:52,760 --> 00:35:54,239
我已经开始使用它

1069
00:35:54,239 --> 00:35:56,449
现在我每天都在用

1070
00:35:56,449 --> 00:35:59,409
越来越发现不同模型的缺陷

1071
00:35:59,409 --> 00:36:00,010
我心想

1072
00:36:00,010 --> 00:36:00,210
哦

1073
00:36:00,210 --> 00:36:01,210
我们甚至做不到这个

1074
00:36:01,210 --> 00:36:02,570
就连简单的事情也如此

1075
00:36:02,570 --> 00:36:03,829
有时候我甚至

1076
00:36:03,829 --> 00:36:06,449
哦你看不清这里的内容

1077
00:36:06,449 --> 00:36:07,269
我完全不明白

1078
00:36:07,269 --> 00:36:09,179
这就是这个人说的

1079
00:36:09,179 --> 00:36:09,639
这个人

1080
00:36:09,639 --> 00:36:11,019
这是这家公司说的

1081
00:36:11,019 --> 00:36:11,820
什么啊，我心想

1082
00:36:11,820 --> 00:36:12,719
哦糟糕了

1083
00:36:12,719 --> 00:36:15,639
这方面还有很多工作要做

1084
00:36:15,639 --> 00:36:16,820
所以啊

1085
00:36:16,820 --> 00:36:17,380
总之

1086
00:36:17,380 --> 00:36:18,079
我觉得这就是

1087
00:36:18,079 --> 00:36:18,539
你知道的

1088
00:36:18,539 --> 00:36:21,028
给这个局面泼点冷水

1089
00:36:21,028 --> 00:36:25,849
此时公司可能会意识到大客户们

1090
00:36:25,849 --> 00:36:26,068
哦

1091
00:36:26,068 --> 00:36:28,400
我们可能得减少开支

1092
00:36:28,400 --> 00:36:30,340
这时开始真正影响英伟达

1093
00:36:30,340 --> 00:36:32,019
这时英伟达的利润率下降

1094
00:36:32,019 --> 00:36:33,139
这才是可怕的部分

1095
00:36:33,139 --> 00:36:34,480
这就是人们会说

1096
00:36:34,480 --> 00:36:36,019
这就是他们缺乏热情的原因

1097
00:36:36,019 --> 00:36:36,820
大家都知道

1098
00:36:36,820 --> 00:36:39,250
今年甚至明年视频号都会成为爆款

1099
00:36:39,250 --> 00:36:41,949
他们担心两位数的股票代码

1100
00:36:41,949 --> 00:36:42,590
没错

1101
00:36:42,590 --> 00:36:45,920
因此英伟达陷入了困境

1102
00:36:45,920 --> 00:36:47,179
当股价处于高位时

1103
00:36:47,179 --> 00:36:49,449
就像我去年提到的

1104
00:36:49,449 --> 00:36:52,849
长期卡在约125到200区间

1105
00:36:52,849 --> 00:36:56,750
因为无法获得重大买盘支撑

1106
00:36:56,750 --> 00:36:58,489
我们长期看好英伟达

1107
00:36:58,489 --> 00:37:01,380
你更像是一个交易者

1108
00:37:01,380 --> 00:37:02,400
短期持有股票

1109
00:37:02,400 --> 00:37:05,460
人们来回交易股份，市场情绪转向乐观

1110
00:37:05,460 --> 00:37:06,599
可能小幅推高股价

1111
00:37:06,599 --> 00:37:07,559
市场情绪转向谨慎

1112
00:37:07,559 --> 00:37:08,440
股价会略有回落

1113
00:37:08,440 --> 00:37:08,829
比如

1114
00:37:08,829 --> 00:37:09,730
你知道的

1115
00:37:09,730 --> 00:37:10,670
这就是常态

1116
00:37:10,670 --> 00:37:11,010
确实如此

1117
00:37:11,010 --> 00:37:11,309
好的

1118
00:37:11,309 --> 00:37:11,469
各位

1119
00:37:11,469 --> 00:37:12,510
感谢大家一如既往的参与

1120
00:37:12,510 --> 00:37:13,469
非常感谢你们的到来

1121
00:37:13,469 --> 00:37:14,728
希望你们喜欢今天的视频

1122
00:37:14,728 --> 00:37:16,369
今日重点评论在下方

1123
00:37:16,369 --> 00:37:18,248
如果你想加入自己的私有股票群组

1124
00:37:18,248 --> 00:37:20,429
可通过千倍股票网获取

1125
00:37:20,429 --> 00:37:22,050
所有优质资源

1126
00:37:22,050 --> 00:37:24,570
重点评论就在下方

1127
00:37:24,570 --> 00:37:25,510
所有课程大纲

1128
00:37:25,510 --> 00:37:26,619
备考Discord社群

1129
00:37:26,619 --> 00:37:27,699
SoFi聊天室

1130
00:37:27,699 --> 00:37:28,579
AMD聊天室

1131
00:37:28,579 --> 00:37:29,619
这些同样适用

1132
00:37:29,619 --> 00:37:31,679
我们甚至还有英伟达专属社群

1133
00:37:31,679 --> 00:37:34,059
如果你想在X平台关注我

1134
00:37:34,659 --> 00:37:36,760
下方第二个重点评论

1135
00:37:36,760 --> 00:37:37,960
完全免费

1136
00:37:37,960 --> 00:37:38,880
如果你使用平台

1137
00:37:38,880 --> 00:37:40,460
祝大家一切顺利，度过美好一天

共 2 条，按点赞数排序。

采集完成，该视频无弹幕