저라면 앞으로 "이것" 하겠습니다. — Transcript

Full transcript

1. 0:14This person is Rick Rieder. He is the
2. 0:17Chief Investment Officer of Global Fixed
3. 0:18Income at BlackRock, the world's largest
4. 0:20asset manager, and he said that tech
5. 0:21stocks are really cheap. He made this
6. 0:23comment the very morning after the
7. 0:25Nasdaq 100 dropped 6.4%. Nvidia alone
8. 0:28had fallen 9.3%just the day before. The
9. 0:30Philadelphia Semiconductor Index fell
10. 0:32even further. It recorded its worst
11. 0:35decline since March 2020. Yet, the next
12. 0:37day, he went on Bloomberg TV and said
13. 0:39they were cheap. The market collapsed
14. 0:41like that in a single day, and he calls
15. 0:43it cheap? Isn't he telling us to catch a
16. 0:44falling knife? You might be thinking
17. 0:46that. So today, we will check one by one
18. 0:48with numbers whether Rieder is right and
19. 0:50if tech stocks are truly cheap.
20. 1:17To summarize what he said, most people
21. 1:19who bought US stocks were ultimately
22. 1:21holding the same few companies. And most
23. 1:22of them were likely semiconductors. But
24. 1:24just because everyone was crowded in
25. 1:25didn't mean they were all expensive. If
26. 1:27you look at the P/E ratio of tech stocks
27. 1:28, it's at the 17x level, which is far
28. 1:30below the five-year average of 25x.
29. 1:33And excluding the 2020 COVID lows, the
30. 1:35current valuation is even cheaper than
31. 1:38during the major 2022 market downturn or
32. 1:40the 2018 trade war. It is an absurdly
33. 1:43cheap price. You might think that their
34. 1:45growth rate must be low, but the dark
35. 1:47blue graph you see above represents AI
36. 1:49infrastructure-related stocks. As you
37. 1:51can see, AI infrastructure stocks are
38. 1:53projected to show an earnings-per-share
39. 1:55growth of over 80%by 2027.
40. 1:57Conversely, non-AI related stocks can
41. 2:00only grow at about 10%.
42. 2:02So, Nvidia, which is growing nearly
43. 2:04two-fold in a year, has a P/E of 10x,
44. 2:06while companies like Costco or Walmart,
45. 2:08which grow by just over 10%a year, are
46. 2:10trading at nearly 40x.
47. 2:12Isn't that somewhat ironic? I believe
48. 2:14this reflects a market that has
49. 2:16essentially taken for granted that AI
50. 2:17will fail and priced it in. Also, people
51. 2:19keep comparing this to the dot-com
52. 2:21bubble, but the graph makes it very
53. 2:22obvious. The graph during the dot-com
54. 2:24era showed an overwhelming rise, whereas
55. 2:26now we can see that the bubble has
56. 2:27already burst while rising, and it is
57. 2:29actually reverting to the mean. So, you
58. 2:31might be wondering why on earth it is
59. 2:33falling then. Rick Rieder finds the
60. 2:35answer in the recent mass liquidations
61. 2:37by hedge funds and the excessive
62. 2:38concentration of positions.
63. 3:22In other words, both institutions and
64. 3:23average retail investors had too high a
65. 3:25proportion of stocks in their assets.
66. 3:27And among those stocks, since everyone
67. 3:28was buying semiconductor stocks, there
68. 3:30was no one left to buy them. That seems
69. 3:32to be how he interprets it. So the "
70. 3:34rebalancing" Rieder speaks of is to
71. 3:36reduce the stock portion a little and
72. 3:38fill it back up with bonds, and here the
73. 3:40host asks what we are most curious about
74. 3:41. Is this just a simple adjustment of
75. 3:43weight, or is the bubble bursting? Let's
76. 3:45hear Rick Rieder's answer right away.
77. 4:18>> Ultimately, the Chief Investment Officer
78. 4:20of the world's largest asset manager
79. 4:22clearly points out that this is not an
80. 4:24AI bubble. The multiple valuation is
81. 4:26just too cheap. He is saying that while
82. 4:28there may be a brief correction, in the
83. 4:30end, it will follow the earnings. There
84. 4:32are networking and optical communication
85. 4:34components, which are the most important
86. 4:35parts that go into data centers. The
87. 4:37growth rate of this market is over a
88. 4:39staggering 95%, and as far as I know,
89. 4:40the optical communication sector has the
90. 4:42highest growth rate as a single market.
91. 4:44However, the valuations of the optical
92. 4:45communication companies you see now are
93. 4:47truly dismal.
94. 4:48Stocks like Marvell, Broadcom, Lumentum
95. 4:50, Credo, and Coherent are trading at
96. 4:52under 20 times earnings.
97. 4:54For reference, these are companies whose
98. 4:56annual revenue is growing by two to
99. 4:57three times.
100. 4:58In fact, the Lumentum CEO appeared on a
101. 5:00broadcast and said this about the stock
102. 5:02price plunge. Our stock price fell 40%
103. 5:04this week, and there is a huge gap
104. 5:06between that and the market reality.
105. 5:08We have an enormous backlog of orders,
106. 5:10and no matter how much we scale up
107. 5:12production, we might only be able to
108. 5:13meet about 30%of the required supply.
109. 5:16This is a report I created as an interim
110. 5:18check for the third-quarter earnings
111. 5:20season. The core of this ten-page report
112. 5:22is that AI demand will be enormous next
113. 5:24year. As you can see, I have brought in
114. 5:27remarks from 15 CEOs during their
115. 5:28earnings calls. First, there is the
116. 5:30OpenAI CFO. Sarah Friar says that
117. 5:32regardless of the midterm election
118. 5:33results, we must continue to invest in
119. 5:35data centers to meet customer demand.
120. 5:36She is saying that the AI demand is real
121. 5:39.
122. 5:40CoVE CEO Michael Intrator also
123. 5:41double-checks the statement that the
124. 5:43completion of already permitted data
125. 5:44centers will be allowed.
126. 5:46He said that we like that. He also said
127. 5:47that we should cooperate well moving
128. 5:49forward. Amazon CEO Andy Jassy also said
129. 5:51that customers use up the capacity as
130. 5:53fast as we install it. We don't see that
131. 5:55capacity easing up. The Microsoft CEO
132. 5:57also said we are doing more data centers
133. 5:58. He strongly argues that he has never
134. 6:00been as confident in the long-term
135. 6:02returns of our investments as he is now
136. 6:04. Mark Zuckerberg says the exact same
137. 6:06thing. The TSMC Chairman says the same
138. 6:08thing as well. What I can say now is
139. 6:09that demand is extremely strong. We
140. 6:11haven't seen any signs of a slowdown at
141. 6:12all. TSMC is the foundational company
142. 6:15that makes the chips. So, it is said
143. 6:17that they actually conducted an
144. 6:18investigation. Go to the data centers
145. 6:19and check if they are actually being
146. 6:21built. TSMC dispatched teams to the
147. 6:22sites, and the result was that this is
148. 6:24for real. You don't need to worry about
149. 6:26TSMC packaging demand. It is said that
150. 6:27the Chairman confirmed it by visiting
151. 6:28the sites himself. And the Intel CEO
152. 6:30also says it is the most severe supply
153. 6:32shortage in history.
154. 6:33The CEOs of AMD and Broadcom say they
155. 6:35need a massive amount, and SK Hynix says
156. 6:37their DRAM and HBM supply is sold out
157. 6:39for the next three years through 2029.
158. 6:41Micron is saying the exact same thing.
159. 6:43And now, let’s look at the leaders in
160. 6:44optical communications. The CEOs of
161. 6:45Coherent and Lumentum also say it is
162. 6:47completely sold out with no end in sight
163. 6:49. Customers are currently booking orders
164. 6:51for 2028. Not only that, I have compiled
165. 6:53statements from every sector as well. I
166. 6:55have organized only the truly major
167. 6:57companies. Whether it's SpaceX,
168. 6:58semiconductor equipment stocks, memory
169. 7:00storage, or optical communication
170. 7:01networks, if you look here, the story is
171. 7:03all the same. I won't list them all one
172. 7:05by one. To put it in one word, AI demand
173. 7:07next year will be incredibly strong. You
174. 7:08only need to look at this one sentence.
175. 7:09Everyone is saying this, so surely this
176. 7:11can't be a lie, right everyone? That is
177. 7:14what I think. All the evidence,
178. 7:15indicators, and data are telling the
179. 7:17same story.
180. 7:18They are all saying that demand for AI
181. 7:20is truly immense, so I honestly don't
182. 7:21understand why the AI bubble theory is
183. 7:23so prevalent.
184. 7:25For your reference, I have posted a link
185. 7:26to my Telegram in the pinned comment for
186. 7:28the full report, and I provide many
187. 7:30other reports there as well.
188. 7:31Please come in and receive this
189. 7:33information for free. Citibank also
190. 7:35released a very interesting report
191. 7:36regarding memory. Recently, memory
192. 7:38prices have surged more than tenfold.
193. 7:40This happened due to a mere 3.8%
194. 7:42mismatch in supply and demand.
195. 7:45The reason memory is a cyclical industry
196. 7:47has been that if many fabs are built and
197. 7:49supply increases, this negative turns
198. 7:50into a positive, causing prices to
199. 7:52plummet instantly. However, if you look
200. 7:55, it stays at about -3.6%even until 2028
201. 7:58. Even in 2029, when bottlenecks are
202. 8:00somewhat resolved, it is actually -0.3%
203. 8:02, meaning supply is still insufficient
204. 8:04compared to demand. But the problem
205. 8:06doesn't end there; they are projecting
206. 8:08that this shortage will worsen in 2030
207. 8:11and 2031. So, what Citi keeps trying to
208. 8:13say is that memory is no longer a
209. 8:14cyclical industry. This situation where
210. 8:16demand and supply continue to mismatch,
211. 8:18leading to shortages, could last for
212. 8:20more than five years. That is what they
213. 8:22are saying. And recently, there are many
214. 8:24interesting stories coming out regarding
215. 8:26LTA, or Long-Term Agreements. Actually,
216. 8:28after reviewing parts of the contracts,
217. 8:30the analysis shows that for hyperscalers
218. 8:32to break an LTA by paying a penalty fee
219. 8:35, the spot price would need to drop by
220. 8:37more than 60%from current levels for
221. 8:38them to see a profit.
222. 8:41But a 60%drop in spot prices almost
223. 8:43never happens unless there is such an
224. 8:45extreme supply glut that it triggers a
225. 8:47complete down-cycle.
226. 8:50I believe the LTA contracts will remain
227. 8:52valid until then. So, according to
228. 8:53current analyst estimates, we should
229. 8:55consider it almost certain that these
230. 8:56LTA contracts will not be broken. It’s
231. 8:58not just Citi; most Wall Street analysts
232. 9:01are providing this same analysis. But
233. 9:03why on earth aren't the stock prices
234. 9:05rising? With SK Hynix and Samsung
235. 9:06Electronics currently at a PER of three
236. 9:08and a PBR barely over 1, it’s truly
237. 9:10puzzling.
238. 9:11In terms of valuation, this is an
239. 9:13extremely cheap price, and it feels like
240. 9:15there is a floor somewhere that makes
241. 9:16you wonder if it can even drop any
242. 9:18further.
243. 9:19However, this is a structural problem
244. 9:21with the Korean market. There is no
245. 9:22supply and demand; it’s just not there.
246. 9:24When foreign investors pull out, there’s
247. 9:26no buying power to sustain it, leading
248. 9:28to constant dumping and a structural
249. 9:29decline in value.
250. 9:31Since SK Hynix and Samsung Electronics
251. 9:33can’t climb, it seems companies like
252. 9:35Micron and SanDisk are also trapped.
253. 9:38Of course, I partially agree with
254. 9:41pessimists who say all these assumptions
255. 9:43could crumble if AI demand suddenly
256. 9:46drops.
257. 9:48If people end up using AI less than
258. 9:49expected, all semiconductor chips would
259. 9:51face oversupply, leading to a collapse
260. 9:53in the prices of the GPUs used as
261. 9:55collateral, which could trigger a
262. 9:56massive credit crisis in both the bond
263. 9:58and stock markets.
264. 10:00However, I still believe that AI has a
265. 10:02very long way to go.
266. 10:05In reality, we’re only using about 1%,
267. 10:07and with larger markets like Agentic AI
268. 10:09and Physical AI remaining, I personally
269. 10:11don't worry that AI demand will drop
270. 10:13that significantly.
271. 10:15I’m curious to know what you all think.
272. 10:17If you have any thoughts, please leave a
273. 10:19comment, and subscribing or liking the
274. 10:21video is a huge help to the creator.