揭露醜聞的紐約時報記者：黃仁勳隕落內幕和對AI 界的影響 — Transcript

Full transcript

1. 0:00Uh it's extraordinary to think of the
2. 0:02rise and fall of this man and how
3. 0:04rapidly it occurred. It's like something
4. 0:06out of Greek tragedy. You know, he he
5. 0:08was the most powerful man in AI, one of
6. 0:12the most powerful people in the world 30
7. 0:14days ago, right? And now he has nothing
8. 0:18really. I mean, have nothing. He has 3%
9. 0:20of Nvidia, which is still worth a lot,
10. 0:22but he he has no kind of executive power
11. 0:24anymore. So, it's just such a shocking
12. 0:28rise and fall. And it really shows you
13. 0:30that even in the age of AI, [snorts] AI
14. 0:32is taking over everything. It's still
15. 0:35these kind of human foibless that
16. 0:37determine what happens, right?
17. 0:50Ryan,
18. 1:05Mega Speed LS1.
19. 1:26Um, last night we published a story
20. 1:29about a second relationship that he had,
21. 1:33extrammarital relationship, uh, with,
22. 1:36uh, an executive at NVIDIA, this vice
23. 1:39president of healthcare, Kimberly Pal.
24. 1:42And you know, Jensen was already kind of
25. 1:44teetering um given some of the early
26. 1:47reporting from Anna and others um about
27. 1:50his first relationship. And that kind of
28. 1:52seemed to set the wheels in motion even
29. 1:55more. And so um the board met and you
30. 1:59know this morning before the market
31. 2:01opened they put out the announcement
32. 2:22Many people on the board are themselves
33. 2:24Nvidia billionaires. I think there's
34. 2:26four different members of the board who
35. 2:28own more than a billion dollars in
36. 2:29Nvidia stock and they've never seen this
37. 2:32company without Jensen. Jensen is the
38. 2:34only CEO that this company has ever had
39. 2:36for over 33 years. Some of these guys
40. 2:39have went been with him for the very
41. 2:41beginning from the very first day. So,
42. 2:43it's very hard I think for them to let
43. 2:45go of Jensen even when it it's almost
44. 2:47forced that they have to. I suspect that
45. 2:50they were looking for some way not to do
46. 2:52that. Maybe, you know, maybe you go to
47. 2:54counseling Jensen, maybe you step back
48. 2:56for a little bit and then maybe you come
49. 2:57back a little while later. But look, you
50. 2:59know, the stock is down quite a lot
51. 3:01already. It went down today again. that
52. 3:03costs them, that cost the board members
53. 3:05real money. Uh, and so they're
54. 3:07attempting to act in the not so much in
55. 3:08the interests of morality or in the
56. 3:11interests of even good taste, but in the
57. 3:13interests of Nvidia shareholders. And
58. 3:15and that's a tough decision to make.
59. 3:17It's not clear that this company is
60. 3:18worth more with Jensen gone than it is
61. 3:20with a very tarnished Jensen in charge.
62. 3:37You know, in my interviews with him,
63. 3:39Jensen went to such great lengths to
64. 3:42present himself as a family man. It was
65. 3:44something he really really emphasized.
66. 3:46He talked about his wife all the time.
67. 3:49Every time I talked to him, he started
68. 3:51talking about his wife, the sacrifices
69. 3:52she had made uh to make Nvidia possible.
70. 3:55she stepped away from her own career to
71. 3:57raise her kids and then bringing his own
72. 3:59kids into the company even as he's
73. 4:02openly consorting with another
74. 4:04executive. It's almost hard to imagine
75. 4:06that that would happen. In some ways,
76. 4:07it's just a classic tale of masculine
77. 4:10hubris and overconfidence, right? But uh
78. 4:13I think probably the board when they
79. 4:16started to put all this together was
80. 4:17like there's a such a severe error in
81. 4:21judgment here that it calls everything
82. 4:24else into question. I you know I think
83. 4:26also Jensen created a reality bubble
84. 4:28around himself. He had inside the
85. 4:31company he was like it was almost like a
86. 4:33cult leader when he when he got up and
87. 4:35talked. He he was he was like a prophet.
88. 4:38But the thing was everything he said
89. 4:39actually came true. And then when
90. 4:42something he said came true, everybody
91. 4:43in the room made a billion dollars. And
92. 4:46so this created this uh this uh I think
93. 4:49this thing that you couldn't even
94. 4:50disagree with Jensen inside the company
95. 4:53and he wouldn't let you anyway. He just
96. 4:54started screaming at you if you didn't
97. 4:55agree with his point of view. You know,
98. 4:57I disagree with Jensen once and it was
99. 4:59like lighting a fuse on a stick of
100. 5:01dynamite and when it went off he just
101. 5:03started screaming at me. I'm a
102. 5:04journalist. I don't work for the
103. 5:06company. Like there's nothing to be
104. 5:07gained from with that. But that
105. 5:09indicated to me like, "Oh, this guy has
106. 5:12no breaks." Like this guy, there's no
107. 5:14one telling this guy, "You're acting out
108. 5:17of turn here." Uh, you know, even his PR
109. 5:19people, they they were there and they
110. 5:21just let him yell at me for 20 minutes.
111. 5:23You know, there's no one around him
112. 5:24saying, "Jensson, reel it back in
113. 5:26there." He had no one like that in his
114. 5:28life.
115. 5:51Uh, I think the board had maybe
116. 5:54considered a couple possible successors,
117. 5:56but Jensen almost deliberately never
118. 5:58groomed a successor. It's almost like
119. 6:00asking, "Oh, now that Michael Jordan's
120. 6:02gone from the Bulls, like who do we
121. 6:03replace him with?" Like, there's not an
122. 6:05obvious candidate that of anyone
123. 6:07anywhere who can do what Jensen did. He
124. 6:10really was a dynamo in this way, a force
125. 6:12of nature who could both see very deeply
126. 6:14into the future of computer science and
127. 6:16also, you know, be the very tough CEO
128. 6:19screaming at suppliers on Taiwan to get
129. 6:21things delivered on time and who's also
130. 6:23smoozing with customers. Unique to
131. 6:25Jensen, right? It's incredible some of
132. 6:27the stuff he was able to pull off. So,
133. 6:29you can't really replace him. Um, but
134. 6:31she can only ever be a temporary or
135. 6:33interim CEO. And the reason for that is
136. 6:36although she's a very good executive,
137. 6:38she is not a world-class computer
138. 6:41scientist. She she's a finance person.
139. 6:43And so, if you're not a world-class
140. 6:46computer scientist over the long term,
141. 6:48you cannot be the CEO of Nvidia or
142. 6:50really any microchip company. The CEO of
143. 6:53the company has to be an electrical
144. 6:54engineer. They have to understand
145. 6:57computing from the level of the
146. 6:58transistor scaling all the way up to the
147. 7:01level of the data center. And this was
148. 7:03something the board really told me. This
149. 7:05is the reason they they believed in
150. 7:07Jensen so much right from the beginning.
151. 7:09He actually didn't have that many
152. 7:10business skills at first. But he was a
153. 7:12worldclass computer scientist and that
154. 7:14was the bottleneck. That was the thing
155. 7:16that was hard to find. So finding
156. 7:18somebody who who has all of these rare
157. 7:20qualities and putting them together in
158. 7:22one person, it it's really tough. It's
159. 7:24really hard to find that person. The
160. 7:26board is going to have a hard time, an
161. 7:28impossible time honestly replacing Jens
162. 7:41IP.
163. 7:46Uh yeah, I mean there's going to be
164. 7:48downstream impacts for weeks if not
165. 7:50months. I mean you mentioned the
166. 7:52anthropic IPO you know Nvidia is
167. 7:54supposed to invest in anth it already is
168. 7:56an investor in anthropic and is about to
169. 7:58invest more uh you know there's Nvidia
170. 8:01is involved in a lot of these kind of
171. 8:03circular funding deals where not only
172. 8:04supplies chips but you know financing
173. 8:08uh to companies large and small across
174. 8:11the AI and tech ecosystem.
175. 8:13um you know does that continue can it
176. 8:16continue without this kind of
177. 8:18charismatic figure in Jensen Huang
178. 8:21calling the shots um you know if it's a
179. 8:23CF CFO type um now leading the company
180. 8:26will they become a little more cautious
181. 8:28with their spending a little more
182. 8:30cautious with the decision-m um you know
183. 8:33that's that's all going to play out the
184. 8:35next couple again weeks and months and
185. 8:37we'll see what happens
186. 8:38>> ultimately the determinant of whether
187. 8:40this succeeds or fails is is not the
188. 8:42chip companies They themselves are
189. 8:44downstream of the the frontier labs and
190. 8:47to the extent that the frontier labs
191. 8:48make products that people want to use or
192. 8:50even have to use to stay productive or
193. 8:53don't the chips will the chips will be
194. 8:55there to supply them one way or another
195. 8:57is my is is my feeling I think however
196. 8:59if AI disappoints you know as a as a
197. 9:02product if AI ends up being kind of an
198. 9:04expensive and broken toy which it can
199. 9:06sometimes seem like then that then it
200. 9:09won't work.