CNBC & Fox Today On NVIDIA Stock, Anthropic IPO, Micron Stock - NVDA Update — Transcript
Full transcript
- 0:00pave the way for for what we can talk
- 0:01about because you have ties with the
- 0:02administration. You you with the Trump
- 0:04accounts and everything else. You know
- 0:06everything that's going on. You know the
- 0:08the secretaries of both Treasury and and
- 0:10Commerce and you know the president
- 0:12>> and you put this that intro with the
- 0:15backdrop of what happened with
- 0:17>> uh with Jensen and then Elon's comments
- 0:20about Michael Jordan and and and some
- 0:23efforts to bring
- 0:24>> Jensen back. How's it all played? and
- 0:27and you also probably heard uh Palmer
- 0:31Lucky say that that the Chinese
- 0:33communist government doesn't want the
- 0:35chips going companies.
- 0:38>> Correct. Correct.
- 0:38>> Well, that's a whole creates a whole new
- 0:40dynamic for for whether this was a bad
- 0:43thing Jensen reportedly was doing or
- 0:45not.
- 0:46>> Well, first first let's start here. I
- 0:49agree with Elon and Palmer that it's
- 0:53crazy to take this person off the floor
- 0:55unless he violated national security
- 0:58laws. Right? If he did that, let me be
- 1:00absolutely clear. Jensen's not above the
- 1:02law. Elon's not above the law. Palmer,
- 1:05no founder is above the law. Okay?
- 1:07Nobody in Silicon Valley that I've
- 1:08talked to thinks that that's the case.
- 1:11What I worried about last Friday, Joe,
- 1:14is this decision to get rid or the
- 1:16pressure on him to resign occurred so
- 1:19quickly. It felt to me like a bit of a
- 1:21rush to judgment.
- 1:23>> Has the board taken their time? And then
- 1:24I know Scott Bessant, the president,
- 1:27other people have opinions about what
- 1:29this company should do. And I was
- 1:31worried about undue pressure on the
- 1:32board. So I wrote the letter last week
- 1:35that said, "Take your time. Your
- 1:36obligation is to shareholders of the
- 1:38company. Take your time. get all the
- 1:40facts and get it right. It's
- 1:42increasingly clear to me that there may
- 1:45not be a violation of law. And if
- 1:48there's not a violation of law, at some
- 1:50point this board is going to have to
- 1:52explain why they did what they did. And
- 1:54I will tell you this based upon the
- 1:57inbounds to me since we wrote the
- 1:59letter. There are a lot of be people who
- 2:02will have an appetite to go after this
- 2:04board. Okay? if it's not clear that
- 2:07there was a violation of law here. And
- 2:10so I think the board move very quickly.
- 2:13It seems to me that they and Besset may
- 2:15know things that I don't know and that
- 2:17we don't know. And if that's the case,
- 2:19then I think it'll be an easy
- 2:20explanation. But if it's if if it's not
- 2:22the case,
- 2:23>> then I think they have a lot of
- 2:25explaining to do to shareholders and
- 2:26other stakeholders as to why they move
- 2:29so hastily to get rid of. have a
- 2:31national policy on exactly what we do
- 2:34want to happen and and how much wiggle
- 2:36rumor is there in this law and did
- 2:37President Trump kind of wink and a nod
- 2:40to Lutnik and and Jensen about doing
- 2:43this?
- 2:43>> Well, I know I I know this the law has
- 2:46been moving around since the since Biden
- 2:49left office. Okay,
- 2:51>> there was a general policy when Biden
- 2:54left office that had export controls on
- 2:56the diffusion of American chip and AI
- 2:59technology. Mhm.
- 3:00>> You heard this president, the AIS are at
- 3:02the time, David Saxs, Howard Lutnik, and
- 3:05others say, "We're going to change that.
- 3:07>> We want the American AI stack to win
- 3:09around the world. We want more
- 3:11diffusion, not less diffusion." And
- 3:13Jensen was very open in Washington that
- 3:16he thought it was a mistake to ban, you
- 3:19know, us from from sending chips to
- 3:21China. it would lead to the ascendancy
- 3:24of the Huawei stack around the world
- 3:26just like happened with the mobile
- 3:28network around the world when we uh slow
- 3:30rolled 5G around the world. So that was
- 3:32a legitimate argument. Those are
- 3:34legitimate policy disagreements between
- 3:36what I think are the more China hawks in
- 3:38Washington and let's call it the more
- 3:40China constructivist or engagement. When
- 3:42you think about the people who want to
- 3:44engage with China, Tim Cook, right,
- 3:46Jensen Hang, the Michael Dells of the
- 3:49world, uh Elon Musk, those people doing
- 3:51business with China, and I think the
- 3:53president is in this camp, they believe
- 3:55the world is a safer place when you're
- 3:57doing business together, when you're
- 3:58meeting with your adversary, rather than
- 4:00poking them in the eye. It's a
- 4:02legitimate policy disagreement. I think
- 4:04the burden of proof is on those who've
- 4:06gone after Jensen to demonstrate not
- 4:09just that they violated what you wanted
- 4:11to be the law, but you violated the
- 4:13actual law.
- 4:14>> Here's the thing. It sounds like u and
- 4:16you know Bessant was on our air. Joe did
- 4:18a great interview with him where he
- 4:21effectively said that he has this
- 4:23position. I think he suggested that
- 4:25Rubio has this position. From the
- 4:27reporting it suggested that Greer has
- 4:29this position, maybe Jay Clayton has
- 4:31this position. That they're all on this
- 4:32side. and then Lutnik and maybe the
- 4:35president is on this other side.
- 4:37>> Yeah.
- 4:37>> Well, first I would say when the
- 4:39president boarded Air Force One to go to
- 4:41China this week, um he was asked this
- 4:44question and he said, "Listen, I have a
- 4:45camp in my administration,
- 4:48probably referring to Bessant that uh
- 4:50wants to wants to be tougher and they're
- 4:52willing to blow up the trade
- 4:54relationship over it. And I have a group
- 4:56on this side that wants to be easier on
- 4:57them and I'm the president of the United
- 4:59States and I'll let them I'll let the
- 5:00cage match ensue." Oh, I thought he was
- 5:02just saying I'm going with the easier
- 5:03side.
- 5:04>> And no, I think he said I'll let the
- 5:05cage match ensue and then I will make
- 5:07the decision B. Now, you might imagine
- 5:10on your way to negotiate with she on
- 5:12this policy having a good good cop and a
- 5:14bad cop on this issue. That's an
- 5:16effective trade strategy. And if you
- 5:18read the art of the deal,
- 5:20>> right, this would be out classic case
- 5:22study out of the art of the deal. So, I
- 5:24don't know where the president is going
- 5:26to come down, but I do believe this is
- 5:28true. Today it is legal to sell H200s to
- 5:32China. The company came out on its
- 5:34earnings call and said, notwithstanding
- 5:36the fact that it's legal, we're going to
- 5:37stop selling H200s to China. At the
- 5:41beginning of this week, it was also
- 5:42legal to sell your most advanced chips
- 5:45to any data center in Indonesia, in
- 5:47Malaysia, in Singapore, etc. They
- 5:50changed the rules earlier this week to
- 5:52say we're going to make it tougher. You
- 5:54have to now have additional licenses to
- 5:56sell to those folks. So the real issue
- 5:58here is this.
- 6:00>> Did Mega Speed did we violate
- 6:03>> Go ahead. Go ahead, Becky.
- 6:04>> Right. That that Brad, that's just it.
- 6:07It's like what happened with Mega Speed?
- 6:08Was Mega Speed a front? Did the company
- 6:11know that? Were those chips being
- 6:13siphoned off and sent back to China? I
- 6:15think those are questions we still don't
- 6:17know the answers to. Um I I hear what
- 6:19you're saying about the changing
- 6:20logistics about what's going to be legal
- 6:22now. And in the question, Becky, but you
- 6:24know, it's it's strange. It's strange
- 6:26that the Singapore line disappeared
- 6:28from, you know, the the earnings
- 6:30release. Um, was that last year when
- 6:32that disappeared? There there are
- 6:34questions around that.
- 6:35>> To me, what another thing we learned
- 6:37this week, there's a release out of the
- 6:38Commerce Department. Is that for 14
- 6:40months, the Department of Justice has
- 6:42actually been investigating this and I
- 6:44think they said they named 17
- 6:45defendants, but I didn't see any
- 6:47defendants from Nvidia or Jensen or
- 6:49anybody else. So, it's not like they
- 6:50just learned about this. In fact, I
- 6:51think Nvidia said in that release, they
- 6:53put it on their do not sell list mega
- 6:55speed over a year ago. So, the fact is
- 6:58this. I don't know. I'm not in the
- 7:00intelligence community. I'm not privy to
- 7:02these things. The president is
- 7:04>> and the president has said what he said
- 7:06knowing all of those things. I imagine
- 7:09that the board is in possession of more
- 7:10information. So, I'm going to be, you
- 7:12know, reserve judgment. And all I've
- 7:14said to the board is take your time, do
- 7:16the right thing based upon full
- 7:18information and facts, but I also don't
- 7:20want them to rush to judgment before
- 7:22they've seen all of that.
- 7:23>> The president said, do you mean hire
- 7:25Jensen back?
- 7:26>> Correct. I the the president I I think
- 7:28has implied that he sees no reason. I
- 7:30mean, he wouldn't say that you should
- 7:32bring Jensen back if if he was seeing
- 7:34information that disqualified him from
- 7:36you coming back. And by the way, the
- 7:38other piece, and Becky made this point
- 7:39earlier, if in fact there is a bigger
- 7:42problem at Nvidia, Colette and the board
- 7:46and everybody would be involved in it.
- 7:47>> I mean, you you would think. And so,
- 7:50again, I've been around a lot of boards.
- 7:52I've been doing this for 26 years. I'm a
- 7:54corporate securities lawyer by training.
- 7:57Uh I was deputy secretary of state in
- 7:59Indiana. I I dealt with a lot of
- 8:00corporate issues with respect to boards.
- 8:02So, I'm coming at that this with that
- 8:04perspective. I want to show a lot of
- 8:06difference and respect to this board.
- 8:07It's a hard hard position that they're
- 8:10put in, but they need to know that if
- 8:14they're rushing to judgment here and
- 8:15insinuating that there was a violation
- 8:18of some bigger law and there wasn't, at
- 8:20some point they're going to have to
- 8:21explain that to shareholders.
- 8:23>> Nvidia told employees that it plans to
- 8:25move on from Jensen Hong as CEO. Uh it
- 8:28looks to me like Hong is out.
- 8:30>> I think so. Is Nvidia the same capable
- 8:33of doing the same greatness without
- 8:35Jensen Hawk?
- 8:36>> I hate to say it as a [clears throat]
- 8:37shareholder uh Steuart, but no. And I'll
- 8:39tell you why. Um Colette Crest, the CFO,
- 8:41and by the way, she's been CFO for 13
- 8:43years. But my point is she is a guardian
- 8:46of the company. Jensen Huang was the
- 8:49founder, the visionary, and admittedly
- 8:51the salesperson, the frontman for AI. So
- 8:54Nvidia without him is not the same
- 8:56company. That said, Nvidia just
- 8:58increased its guidance. It still is the
- 9:01company that enables AI. They're going
- 9:03to grow 75% this year on the top line.
- 9:06Uh and the shares are trading at about
- 9:0810 times earnings. That's half the
- 9:10valuation of the S&P 500 with three
- 9:13times the growth. You have to own Nvidia
- 9:15here. I own it. I've been buying more.
- 9:17It's been painful, but I still like
- 9:19Nvidia.
- 9:19>> But I do think it's important they have
- 9:21a $3.1 billion a month bill that goes to
- 9:24SpaceX for compute. I think it's
- 9:26interesting by the way that Elon Musk
- 9:27really one of the people who was just
- 9:29saying travesty what happened Jensen
- 9:32well you know what I would if I wanted
- 9:35cheaper GPUs I'd be trashing the heck
- 9:38out of Nvidia and it seems like
- 9:39everyone's doing it who needs GPUs off
- 9:42record or on record.
- 9:44>> It's a pretty tepid bounce for Nvidia
- 9:46this morning.
- 9:46>> No bounce at all. I think it just goes
- 9:48right back to where it was. Maybe it
- 9:50even goes down, right? Because it's not
- 9:52>> the stock is the company's gonna have a
- 9:54hard time getting a CEO if it has
- 9:57someone as powerful as Jensen really
- 9:59saying things that would he he does have
- 10:02to moderate himself if he thinks there
- 10:04can be a CEO unless he thinks he's a
- 10:06CEO. President you made this point that
- 10:08if he keeps saying look I'd be happy to
- 10:10come back. Why would a candidate want to
- 10:13want to consider it? There are people
- 10:14who are being bet on radically on in
- 10:16this site that we're affiliated with Khi
- 10:19who I think they may not real the
- 10:21betterers may not realize are sensitive
- 10:23the idea that if you take the job but
- 10:25chances out there saying he wants the
- 10:27job your life would be miserable. Most
- 10:29these people I look there are people who
- 10:31misery loves money.
- 10:33>> Yeah.
- 10:33>> But I do think that we have to keep and
- 10:35I don't want to put Nvidia front and
- 10:36center but it is interesting. You see
- 10:38all the stocks that are climbing. Nvidia
- 10:40was at 154 this morning at 430. It's
- 10:42done nothing. we go straight down and
- 10:44that again is in part because Jensen
- 10:46lives
- 10:47>> um you know the my understanding
- 10:49[clears throat] is that the uh the
- 10:51search is moving as fast as they can.
- 10:53>> Oh yeah.
- 10:54>> Um
- 10:55>> does what does that mean? Does that mean
- 10:57you even get to you know do you get to a
- 10:59a new CEO early next year? That would be
- 11:02incredibly fast given everything they
- 11:04need to go through at Nvidia. Uh it is
- 11:07fascinating Jim to continue to hear this
- 11:09uproar amongst so many certainly in the
- 11:12tech community talking about Jensen of
- 11:14course as they believe irreplaceable and
- 11:17and don't understand why the board moved
- 11:20as quickly as they did.
- 11:22No CEO who's ever been uh I don't even
- 11:24know how to say what he did with par
- 11:26with perhaps more than one executive uh
- 11:28has been done that and has kept their
- 11:31job
- 11:32>> even look I think money is important but
- 11:35there there's ethics
- 11:37>> look I have a huge position my chat
- 11:39trust I I'm unlike most these people I
- 11:41see on air I'm going to talk against my
- 11:42trust ethics and national security the
- 11:44New York Times piece was death was just
- 11:47incredibly horrible
- 11:49>> incredibly horrible about what the the
- 11:51company knew about sending it right to
- 11:52the PLA even though Jensen said it
- 11:54wouldn't go to the PLA and I'm thinking
- 11:56about national security and I'm thinking
- 11:57about ethics or I just say you know what
- 12:00I want to make a lot of money those
- 12:01things don't matter at a certain point
- 12:03you know how hated rich people are going
- 12:04to be I mean rich people don't have
- 12:06ethic everyone anyone who's watching the
- 12:08330 million people out there they do
- 12:11this stuff they get fired immediately
- 12:12they're forgotten they're just they
- 12:13disappear but if you're rich enough in
- 12:15this country are you allowed to do
- 12:17whatever you want
- 12:18>> that's it is a fascinating I'm finding
- 12:20this period fascinating in terms of the
- 12:22idea that somehow rules don't apply. And
- 12:24I get it.
- 12:26>> It's amazing. I've gotten some very nice
- 12:27emails from people who are otherwise
- 12:29very cordial, but saying you don't I
- 12:30don't understand why you keep talking
- 12:32about this. These are and we're talking
- 12:33again about Jensen Wong and Nvidia
- 12:35because the stock is not moving today on
- 12:37what otherwise may be a very positive
- 12:39day overall given the anthropic IPO and
- 12:40how strong it's going to be. But they I
- 12:43feel like I'm living in a very different
- 12:44world because the world that I grew up
- 12:46in where I followed governance for all
- 12:47these years, typically when you do
- 12:49violate the rules, when you lie to your
- 12:51board, that is a fireable offense. And
- 12:53everything I continue to hear from the
- 12:54board of Nvidia is in terms of the
- 12:56personal liability for the board
- 12:57members, not to mention somebody used a
- 12:59very colorful phrase with me today,
- 13:01basically just saying no effing way.
- 13:05>> God, I are they bringing him back. could
- 13:07be I have another one that I I'm afraid
- 13:09that if we do two of those back to back
- 13:11>> but you know what I mean you heard Brad
- 13:13Gersonner earlier on
- 13:14>> oh yeah he's been deport a bunch of
- 13:16losers
- 13:16>> obviously spending a lot of time on
- 13:18anthropic but also talking about this
- 13:20continued
- 13:21>> chorus um that he is certainly a part of
- 13:23of it was judged too quickly and
- 13:26>> but to that point guys this is a live
- 13:28picture that we're seeing right now
- 13:29outside of the NASDAQ to your point Jim
- 13:31I mean Jensen's behavior uh certainly
- 13:34does it adds to the mounting concerns
- 13:37about inequality, the impact of AI on
- 13:40employment, the obvious push back from
- 13:42data centers, the what resulted in the
- 13:44midterm blue wave. I mean, it's all
- 13:48the same,
- 13:48>> you know, and I look, I look, there's
- 13:50obviously no one who's been a bigger
- 13:51back, including all these rich hedge
- 13:53fund managers than I was for Jensen.
- 13:55>> I just am unfortunately or fortunately
- 13:57stuck with the ethics and the national
- 13:59security issues. Why? Because like
- 14:02David, I think that rich people should
- 14:05be bound by our rules. They should be
- 14:07bound. We do not have oligarchs in this
- 14:09country. We are a nation of men.
- 14:11>> Yeah. Yeah,
- 14:12>> if you believe I rules fascinating to
- 14:16see given the world that they live in
- 14:18and what they and by the way it also is
- 14:21colored by Elon Musk as well who I as
- 14:23I've made the point many times controls
- 14:25the SpaceX board has significant amount
- 14:27of voting uh stake in Tesla as well and
- 14:30does operate on his own and they all
- 14:32come and they all look at Elon and say
- 14:34well Elon's able to do it and Jensen is
- 14:37almost his equivalent in terms of his
- 14:38importance certainly to Nvidia why
- 14:40shouldn't Jensen be allowed to. But it
- 14:42it is different.
- 14:43>> What happens if the vice I heard someone
- 14:45say, "What happens if the vice president
- 14:46>> typical board does what they're doing?"
- 14:48>> I heard what happens if the vice
- 14:49president of healthcare really liked uh
- 14:51Jensen and Jensen had an open marriage.
- 14:54Who are you to judge? Who are you to
- 14:57judge the the Jensen? And I'm like
- 14:59saying, well, wait a second, you no, I
- 15:01just didn't know the history of war. And
- 15:03the person says to me, you when since
- 15:05when are you not a capitalist? I I I
- 15:07know I'm finding I'm getting the same
- 15:09I'm like same criticism just simply
- 15:11pointing out the things we would have
- 15:13we've sat here how many times through
- 15:14the years Carl and talked about his CEO
- 15:16being dismissed as a result of an affair
- 15:18with a director
- 15:20[laughter]
- 15:21how about Easterbrook at McDonald's boom
- 15:24>> yeah um
- 15:25>> out so fast his head was spinning faster
- 15:27than a big man
- 15:28>> I mean I know
- 15:29>> we still don't know what happened to the
- 15:30poor guy at Kroger but he you know
- 15:32>> well I do know but you know what
- 15:34>> you've never told
- 15:35>> well it's not my business to know. But I
- 15:37will tell you that it that one could
- 15:40have been worse, frankly. But no, I
- 15:41don't know anything. I don't know. But
- 15:42all I can tell you is this that what I'm
- 15:45watching you are two class of people in
- 15:46this country and I think that
- 15:49>> well
- 15:49>> those I think the pro
- 15:51>> it's only going to get stranger as this
- 15:53company Anthropic continues to move
- 15:55forward and uh
- 15:56>> right and we got to go back to anthropic
- 15:58but
- 15:58>> not just creating wealth but more
- 16:00importantly creating the models that are
- 16:02going to take us to super intelligence
- 16:05when we get to recursive
- 16:06self-improvement so many of their
- 16:07engineers are talking about is fairly
- 16:10close is a 2027 event and what that is
- 16:13going to mean in terms of the changes uh
- 16:17potentially potentially uh in society
- 16:20and again there are different sides of
- 16:22this but many at anthropics certainly
- 16:24view it that way about touching Adela
- 16:26are they smarter than we are what can
- 16:27they be done and the obvious answer is
- 16:29they are smarter than we are but you
- 16:31have to be judicious and remember that
- 16:32the humans should be
- 16:33>> smarter than we are the the agents will
- 16:36be smarter but humans have and then I
- 16:38listened to Nesh Aurora last night with
- 16:39the stock that was the best performer in
- 16:41the market and he's saying listen I have
- 16:43a solution for the open weight models by
- 16:46the way so does Kurts at Crowd Strike
- 16:49>> so does anthropic
- 16:50>> right right so [laughter]
- 16:51people just look when I talked to the
- 16:53anthropic people they wanted to talk to
- 16:55me about two things and by the way
- 16:56they're delightful I like them uh one
- 16:59about Jim are you worried are you
- 17:01worried as we are about guard rails and
- 17:03the second is do you think that we can
- 17:05be a model for philanthropy and I I was
- 17:08quite taken by both very
- 17:10>> public benefit
- 17:11owns. I'm sorry. The foundation owns 5%
- 17:13of the shares I believe for [music]
- 17:16public benefit as well, right?
- 17:17>> I mean, I know I feel I was taken in by
- 17:20Jensen. Now I'm like, please, you know,
- 17:22I'm so off. I I I I think about it every
- 17:25day. I go my come my wife said, "What
- 17:27are you down about?" I said, "I love
- 17:29Jensen. I don't know how this happened.
- 17:31I don't know how it happened."
- 17:32>> I Jim, you and I are in similar. I I
- 17:34It's fascinating to watch this happen
- 17:37with Jensen. I know we've got anthropic
- 17:38in the IPO. We'll be talking all about
- 17:40it. But this movement, Travis Kalanick,
- 17:42Elon Musk, Brad Gersonner and on and on
- 17:44is fascinating in terms of the way they
- 17:47see the world.
- 17:48>> Uh the rate of growth of this company
- 17:49has been extraordinary. But I I want to
- 17:51understand the unlock and also
- 17:53understand the competition in the space
- 17:55because increasingly you have openweight
- 17:57models. You have competitors in open AI
- 17:59which are trying to offer tokens at
- 18:01cheaper prices. uh you have folks like
- 18:03Microsoft which offer your services and
- 18:05yet are also trying to sell uh their own
- 18:07models and they say oftentimes that some
- 18:10of those models or at least some of the
- 18:11features of their models can do some of
- 18:13the same things maybe not at the
- 18:14frontier level. Um how you think about
- 18:17that in terms of what that means longer
- 18:18term to the margins for the company. So
- 18:20I think I think a couple concepts um one
- 18:23concept is the frontier and we have
- 18:25found that there are these incredible
- 18:27returns to the frontier. You know in
- 18:30most products or goods you have
- 18:32competition on price. Here the
- 18:34competition is really on quality because
- 18:36the returns to greater cognitive ability
- 18:39are so high. Right? As we climb the
- 18:42exponential of cognitive ability, you go
- 18:44from doing chat bots to being able to do
- 18:47coding to being able to do biology and
- 18:50robotics and the economic value of
- 18:52things further along the frontier is so
- 18:54much larger than things earlier in the
- 18:56frontier that that's where most of the
- 18:59value accumulates. The second idea is
- 19:02that if you're good at the frontier, you
- 19:04are also good at producing smaller and
- 19:07more efficient models away from the
- 19:08frontier. So if you look at Haiku 5.5,
- 19:12we were able to drop the price by 4x
- 19:15from Haiku 4.5 and by many times from
- 19:18Opus and and Fable 5 family uh which are
- 19:22you know which which it's which it's
- 19:24much cheaper then and so we are able to
- 19:26also compete using our the same
- 19:29technology that we build our models
- 19:31models on the frontier the same compute
- 19:33multipliers to make these cheap models
- 19:35that compete away from the frontier. Um,
- 19:38I got to ask because it's been in the
- 19:39middle of all this. Uh, Nvidia is an
- 19:41investor in your company and in this
- 19:43offering. Um, at the same time, you've
- 19:45been on the other side of the issue with
- 19:48Jensen Wong as it relates to chips in
- 19:50China. And so, what have you thought as
- 19:52you've watched this whole drama play
- 19:54out?
- 19:57>> Yeah. So, you know, I would say we've
- 19:58been a, you know, partner partner to
- 20:00Nvidia for for a long time. Um, you
- 20:02know, we work with them on uh data
- 20:04centers. we've had a long and productive
- 20:06partnership with them and you know of
- 20:08course we're not always going to agree
- 20:10on policy issues with our partners and
- 20:12so you know we have one policy position
- 20:15that you know comes from our conviction
- 20:16of where the where the technology is
- 20:18going and we continue to hold that and
- 20:20some of our partners have a different
- 20:21position.
- 20:22>> I would also just add I think something
- 20:23that is a little bit unique about
- 20:24Anthropic is just how many different
- 20:26partners we work with for compute. So we
- 20:28work with Nvidia, but we also work with
- 20:30with Amazon, with Google, with Broadcom.
- 20:33And I do think this has been, you know,
- 20:34helpful strategic advantage for us that
- 20:36we have so many different incredible
- 20:38partners who we get to work with
- 20:39together to design some of the best
- 20:40chips in the world.
- 20:41>> You guys think Jensen should come back?
- 20:43I should ask.
- 20:44>> We're obviously not going to comment on
- 20:45anything about
- 20:46>> We're not going to come. That's up to
- 20:47Nvidia.
- 20:48>> As we await the first trade for
- 20:50Anthropic, Millius Research initiating
- 20:52coverage this morning with a buy rating
- 20:54and a price target of $155 per share.
- 20:57Joining us now, the analyst behind the
- 20:58call, Ben Writz. Welcome. So, how do you
- 21:00get to to 155?
- 21:02>> Well, what we did was we compared it to
- 21:04the hyperscalers. I mean, it's the
- 21:06closest we can get. There's a capex
- 21:08intensity that's kind of similar. They
- 21:10obviously are all AI companies in their
- 21:12own ways and a couple are tinkering with
- 21:14models and they're partnering and
- 21:16competing. And so, what we thought was
- 21:19that at maturity, the margins for
- 21:21anthropic will be probably better than a
- 21:24hyperscaler. They have higher gross
- 21:25margins. So we thought on EV to sales is
- 21:28a good approximation of the potential uh
- 21:31valuing it on the earnings power. So
- 21:32what we're valuing it is at um six times
- 21:36our 29 EV to revenues which puts it at
- 21:39the current average of the forward
- 21:41multiple for hyperscalers and that could
- 21:43be conservative and the reason it's
- 21:45conservative is much better growth and
- 21:47the leverage here could be insane like
- 21:49because you have an 85% gross margin. So
- 21:52>> people worry about that margin. They
- 21:54they worry about the price of tokens and
- 21:56what that will do for margins.
- 21:58>> Well, it's going down.
- 21:59>> I mean, yeah, but their cost to serve is
- 22:02going down. 2 years ago, their cost to
- 22:03serve was like $25 per dollar revenue.
- 22:07Now it's 14 and it might go into the
- 22:09singledigit cents pretty soon. The other
- 22:12thing is is that they're a leader in
- 22:13reinforcement learning technologies and
- 22:15other model. They're they're ahead of
- 22:17the game in models, making them more
- 22:19efficient, and their cost to train is
- 22:22going down and the speed to train is
- 22:25going down. So, they're going to be
- 22:26putting out models at such a rapid clip
- 22:28that are on the frontier, then they're
- 22:31going to still get a premium. And so,
- 22:33their cost to serve and cost to train go
- 22:35down that makes them more valuable and
- 22:38worth paying for. So, I think they can
- 22:39stay ahead of it,
- 22:40>> which is why it's so interesting to
- 22:42listen to them talk about safety and
- 22:44guardrails, right? I mean what what
- 22:45cadence of new models is too slow?
- 22:49>> Well, look, you know, we we are entering
- 22:52um a a realm of recursive
- 22:55self-improvement and their models are
- 22:57recursive already and the big limiting
- 23:00factor to RSI. Uh those are models that
- 23:03basically train themselves and improve
- 23:04themselves. The the the big argument is
- 23:07should there be a human in the loop or
- 23:09not? And these things move so fast kind
- 23:11of it's like you know we're like uh you
- 23:14know like you know like like this is
- 23:16like tortoise in the hair like we're
- 23:17like the tortoise and going yeah okay
- 23:20like we're in the loop. Okay. So uh so
- 23:23it is a it actually is slowing them
- 23:26down. They could go faster but they want
- 23:28to make it safer. So, what their big
- 23:31challenge is going to be is how do they
- 23:34introduce recursive self-improvement
- 23:36while keeping a either a human in the
- 23:37loop or convincing us that they've
- 23:40trained it in a matter that's safe? And
- 23:42I'm not smart enough. Maybe those two
- 23:44people you just interviewed are smart
- 23:45enough, but we're going to have to have
- 23:47some trust that they have a safety
- 23:49mission and can do this. And um
- 23:51>> otherwise, we're all in big trouble.
- 23:53>> Well, you know, that's why I have a buy
- 23:55and a big target because I'm dead
- 23:56anyway. It doesn't work, right? I mean,
- 23:59wait,
- 24:00>> world only ends once.
- 24:01>> Have you been to Costco lately, too, or
- 24:03no?
- 24:03>> Yeah. What's the worst thing that can
- 24:04happen? I die. Oh, wait. That's in the
- 24:06risk factors. Okay. So, you know, um,
- 24:09so, so what what the heck?
- 24:10>> Um, what percent to the degree that
- 24:13Claude helps create nextgen Claude now,
- 24:16>> like what percent? I've seen 90% in
- 24:18terms of the the the coding part of
- 24:21that. Are we already there? Well, I I
- 24:24think like like I said, I mean, I don't
- 24:25even know how to answer that question
- 24:27100%. Like I don't know how much they're
- 24:29throttling it to to get to it in order
- 24:32to be safe, right?
- 24:33>> But um the big question is how like if I
- 24:36were able to speak to them, I would talk
- 24:37to them about how much is human in the
- 24:39loop. How much can you do without that
- 24:41and keep to your mission and make the
- 24:44government happy and then potentially be
- 24:46the leader in regulation so that you
- 24:48make sure however I've been impressed uh
- 24:51I've heard a lot Greg Brown uh Mr. Uh
- 24:55Greg Brown, the tall guy we saw standing
- 24:57there.
- 24:57>> Yeah.
- 24:58>> Yeah. Yeah. Well, he's in DC. Tom Brown.
- 25:00Greg Brown. I've got a lot of names in
- 25:02my head. Uh Tom Brown is uh in the DC
- 25:05area and he's done a lot of work with
- 25:07them on the policy side. So, you know,
- 25:09perhaps they're getting on the right
- 25:10side of the the administration.
- 25:12>> We have a Neil Michael on tomorrow.
- 25:14We'll ask about that.
- 25:14>> Yeah. Yeah. Yeah. Ask him.
- 25:16>> Um compute is obviously the most
- 25:18important part of this. I mean, these
- 25:19new models need to be trained on
- 25:21something.
- 25:21>> Yep. any concerns overall about the
- 25:24ability to actually get that compute uh
- 25:26andor the pace at which it will be made
- 25:29available given certainly the election
- 25:31we had a few weeks ago.
- 25:32>> Oh, sure. I worry about that every day
- 25:34for every [clears throat] company I
- 25:35cover right now. And
- 25:36>> how do you see it at this I mean, again,
- 25:38it's hard to know, but
- 25:39>> so the way I see it is if it's tight,
- 25:42these guys have the resources and the
- 25:44partnerships to lead and, you know, on a
- 25:47pound-for-pound basis, their access to
- 25:49compute is going to be better than
- 25:50everybody else's, maybe even better than
- 25:52OpenAI's. And so they have uh a very
- 25:57their plan is very Broadcom ccentric as
- 26:00well as SpaceX. And we, you know, at
- 26:02Melius have a pretty good line of sight
- 26:05on those gigawatts that are coming for
- 26:07them because we cover a lot of the
- 26:09companies involved and it does seem like
- 26:10there's land power shell at least at
- 26:12least a decent, you know, obviously for
- 26:1427 and then heading into 28. But David
- 26:18relative to everybody else, they're
- 26:19going to have an advantage. So if others
- 26:21can't if they can't get compute, none of
- 26:23their competitors will.
- 26:24>> That was that was how they answered
- 26:25actually on Nvidia question that they
- 26:27have a lot of suppliers. closed models
- 26:29like theirs are going to run more
- 26:31efficiently on the architectures that we
- 26:33have. So they're going to benefit that
- 26:35way and then the you could argue token
- 26:39prices if we don't have enough compute
- 26:41may even go up due to scarcity. So
- 26:43they're going to be in the pole
- 26:44position,
- 26:44>> right? What about the uh cost of
- 26:46switching for their customers which some
- 26:48see is very low and therefore wonder
- 26:50whether there's a real mode here or
- 26:52whether hey the latest model is I like
- 26:54this one, I'm going to that one and I'm
- 26:55getting rid of you. Anthropic.
- 26:57>> Yeah. I mean there's there's I guess you
- 27:00know people use open AI chat and people
- 27:03use are using Grock more and more. Um
- 27:06look it it's always a threat. I think
- 27:08that's why they're expanding into other
- 27:11uh markets. The the real thing is like
- 27:13the app like they they really are
- 27:16attacking the application stack too. So
- 27:18it's like multiple products get attached
- 27:20then that moat becomes tighter. The more
- 27:23products you use it becomes tighter. Is
- 27:25this going to trade like a hyperscaler,
- 27:26a semiconductor? What? Like how?
- 27:28>> Oh my god. I hope I hope well you know I
- 27:30had a a few really good years there on
- 27:32the semis but you know I I hope it
- 27:34trades uh like anthropic. I think like
- 27:36look this is the pure play. This is what
- 27:38everybody was working towards sort of
- 27:40like an IPO of an AI first disruptor
- 27:44that is going to you know augment labor
- 27:47take some labor create labor that never
- 27:49could have been invented before and it
- 27:52it's going to trade like anthropic. I
- 27:53don't want to sound but the maybe the
- 27:55closest thing is like Google in 2005
- 27:58kind of thing where you you where they
- 28:00were ahead on the algorithms and then
- 28:02like they just outpaced everybody. It
- 28:05might be the most akin to that that kind
- 28:07of Google like run back in the day.
- 28:10>> That's what I see.
- 28:10>> All right, we're just seeing Anthropic
- 28:12open up almost 34% there. You saw the
- 28:15collapse at the NASDAQ market site. ENT
- 28:18Anthropic is now a public company and
- 28:22the price one, you know, it's above $2
- 28:24trillion, right? This is historic
- 28:26record-breaker in so many different
- 28:28ways. Carl,
- 28:28>> uh, opened right at 11150.
- 28:32So, we're going to watch this now. We've
- 28:34gotten kind of good at this in terms of
- 28:36the way of large issues and see how we
- 28:38manage through the rest of this morning
- 28:40and into this afternoon. Uh, but what a
- 28:42remarkable moment. Um, and
- 28:44congratulations so far to the NASDAQ and
- 28:47all involved for making this pretty much
- 28:49picture perfect.
- 28:50>> I got to ask you about uh, you know,
- 28:52Nvidia. Uh, so folks, the dark line is
- 28:55your four PE ratio. The stock is hasn't
- 28:58been this cheap. You got to go back to
- 28:59when they were selling only gaming con
- 29:01gaming chips. Uh, so it's the green line
- 29:04is how much sales they have. It's become
- 29:07extraordinarily valuable. I mean,
- 29:09successful and yet it's trading like
- 29:11it's going out of business. What are you
- 29:12telling clients with Nvidia?
- 29:14>> Patience. There's going to be a
- 29:16>> Are you buying the dip at all? Do you
- 29:18dip? I don't buy dips. You don't buy
- 29:19dips?
- 29:20>> No, I buy dips. I buy reality.
- 29:21>> So, where would you buy? Because a lot
- 29:23of people are saying, "Okay, if it holds
- 29:24this, I think 150 and we start to turn
- 29:25around, that might be an area not
- 29:27catching a falling knife, but you I
- 29:29mean, you want to take advantage of the
- 29:31opportunity once the coast is clear. I
- 29:33think the opportunity starts to
- 29:35formulate at the 145 level, but I've got
- 29:38institutional clients that are thinking
- 29:40it's going lower than that
- 29:41>> really.
- 29:42>> And that's just because Jensen, you
- 29:45know, not only was he the father of uh
- 29:48super intelligence, but he was the
- 29:49capital allocator and he had such an
- 29:52influence um globally
- 29:55uh on that trade that with him out of
- 29:57the picture that that's somewhat
- 29:59problematic. And I think every single
- 30:01deal that they've done is going to be
- 30:03turned over uh by not only the the SEC
- 30:07but everybody that's going to go and
- 30:08look what they did uh in China and what
- 30:11deals they structured. So you got a huge
- 30:13overhang for a lot a lot of time here.
- 30:16So
- 30:16>> go through that. The dims have to be
- 30:18sworn in. They'll exactly.
- 30:19>> So in other words, just be patient
- 30:21because this isn't about value right
- 30:22now. It's about getting the a okay of
- 30:24the clearance I guess that the coast is
- 30:26clear legally.
- 30:27>> You don't catch a falling knife until it
- 30:29hits the ground. it stopped spinning and
- 30:30we're we've hit the ground. We're
- 30:32spinning and it's getting slower, but
- 30:34it's not time yet.
- 30:35>> Great stuff, Dad. Thank you very much.
- 30:36We're crunched for time. We don't have
- 30:38that much time. I know you like Google.
- 30:40I know you like SpaceX and I know you're
- 30:41buying Micron. I just I just want to ask
- 30:44you quickly if you can answer it in 15
- 30:46seconds. When does Micron turn around?
- 30:50>> Uh buybacks December 9th. Um Sanjay
- 30:53Meeroth, I heard him yesterday. He said
- 30:55December 9, the buybacks will kick in.
- 30:57Their LTAs which are secured till 2030.
- 31:00They are very hard to break. So that is
- 31:02free free cash flow and they are going
- 31:05to do most of the FCF towards buybacks.
- 31:08That's a huge amount of money to support
- 31:10the market cap.
- 31:11>> Annie, thank you very much. Got I owe
- 31:12you some time next time.
- 31:14>> All right. I hope you're all doing well
- 31:15today and staying calm in this market.
- 31:17Thursday was a rough day throughout the
- 31:19market as Anthropic began trading on the
- 31:20NASDAQ and there was an AMD short report
- 31:22related to China which weighed on AMD
- 31:24and AI infrastructure stocks more
- 31:26broadly during a time of severely
- 31:27negative sentiment. I'm not going to
- 31:28spend much time on the AMD short report.
- 31:31It appears that the writers of the
- 31:32report found that a Chinese server
- 31:33manufacturer was added to the US entity
- 31:35list. That company continued publishing
- 31:37server BIOS updates incorporating newer
- 31:39AMD firmware after the blacklisting. And
- 31:41the writers argue that AMD attributed
- 31:43more than 20% of its 2025 revenue to
- 31:45customers build in China and Hong Kong.
- 31:47And they also argue that some companies
- 31:49associated with the blacklisted company
- 31:51could become subject to additional
- 31:52restrictions. Importantly, the short
- 31:54report does not demonstrate that AMD
- 31:56supplied anything illegally. The writers
- 31:58of the report do not know how the
- 32:00blacklisted company obtained the newer
- 32:01firmware or whether any transfer was
- 32:03licensed. But given the fragile
- 32:04sentiment in recent weeks, market
- 32:05participants chose to sell first and ask
- 32:08questions later. This situation with AMD
- 32:10reminds me a lot of the Deep Seek
- 32:12situation in early 2025 where the market
- 32:14got it completely wrong. That's all I'm
- 32:15going to say about that for now. The
- 32:16Wall Street Journal published a story
- 32:18saying that the $500 billion initiative
- 32:19that Jensen Hong announced back in
- 32:21August alongside six of Wall Street's
- 32:23top financiers has yet to fund a single
- 32:25deal and its first transaction may not
- 32:27close until 2027. This is also
- 32:29interesting given that Anthropic is now
- 32:31public as the $500 billion was primarily
- 32:33meant for the Frontier model companies.
- 32:35This story weighed on tech stocks
- 32:36Thursday as there have recently been
- 32:38concerns about the cost and availability
- 32:39of capital. Market participants are also
- 32:41wondering what the future of Nvidia's
- 32:43financial support for ecosystem partners
- 32:45will look like given that Jensen is no
- 32:46longer at the company. Jensen was
- 32:48probably the main advocate for the $500
- 32:49billion arrangement and now his absence
- 32:51raises questions about what the
- 32:53initiative will look like moving
- 32:54forward. In other news, Nvidia announced
- 32:55it will present at the 30th annual UBS
- 32:58Global Technology and AI conference on
- 33:00December 1st starting at 6:35 a.m.
- 33:02Pacific, 9:35 a.m. Eastern. Originally,
- 33:04Jensen was scheduled to deliver a GTC
- 33:06keynote on December 1st at GTC
- 33:08Washington DC, but that keynote has been
- 33:10cancelled in light of recent events. In
- 33:11other news, the House Foreign Affairs
- 33:13Committee released Commerce Department
- 33:14licensing information covering Nvidia
- 33:16and AMD chips. Approximately 3.4 million
- 33:19chips were requested, nearly 900,000
- 33:21were approved, and this covers December
- 33:232025 through May 2026. Approvals
- 33:25included 150,000 for 10 cent, 75,000 for
- 33:28Alibaba, 43,000 for Bayu, and 4,000 for
- 33:32BYD. As a reminder, Nvidia told us on
- 33:33the recent earnings call that they have
- 33:35stopped shipping H200 products to China.
- 33:37There have also been recent reports that
- 33:39hundreds of thousands of Nvidia GPUs are
- 33:41stuck in Hong Kong because the Chinese
- 33:42government is not issuing the necessary
- 33:44permits. So, put simply, it appears that
- 33:46the Chinese government does not want the
- 33:48chips coming into the country, but
- 33:49Chinese tech companies do want the
- 33:50chips. There's a slew of additional
- 33:52stories and headlines related to Nvidia
- 33:54China, the Commerce Department, and the
- 33:55president. I don't know which stories
- 33:56are true and which are false. It's worth
- 33:58mentioning that this is all happening as
- 34:00the president visits China, so that may
- 34:02have something to do with it. The Nvidia
- 34:03China headlines also weighed on
- 34:05sentiment Thursday. Also on Wednesday,
- 34:06Moody's revised Nvidia's credit outlook
- 34:08from positive to stable, citing Jensen
- 34:10Huang's departure, succession,
- 34:11uncertainty, and investigations. That
- 34:13said, Moody's reaffirmed Nvidia one
- 34:16credit rating. I noticed that Wed Bush's
- 34:17Matt Bryson reiterated his by rating on
- 34:20Nvidia and actually raised his price
- 34:21target from $275 to $293 per share.
- 34:24According to Tip Ranks, Matt is ranked
- 34:26number 20 out of more than 12,500
- 34:28analysts. So, that's nice to see given
- 34:30that many analysts have recently
- 34:32abandoned their ratings and lower their
- 34:33price targets. There's a lot of
- 34:34uncertainty in the short to medium-term,
- 34:36which is weighing on the stock. That
- 34:38said, as I've said repeatedly in recent
- 34:40videos, Nvidia's product roadmap for the
- 34:42next couple years is largely already set
- 34:43in motion, and I think the team at
- 34:45Nvidia will likely execute successfully
- 34:46on that road map with or without Jensen.
- 34:48That hasn't changed. The main question
- 34:50has to do with the outy years and the
- 34:52long-term vision for the company. That
- 34:53remains uncertain, but the roadmap for
- 34:55the next couple years remains intact.
- 34:57And last I checked, Nvidia was trading
- 34:58at roughly 10 times next 12 months
- 35:00earnings, while the S&P 500 was trading
- 35:02at roughly 17 times next 12 months
- 35:04earnings. Meanwhile, Nvidia is expected
- 35:06to grow revenue more than 70% next year,
- 35:08while S&P 500 revenue growth is expected
- 35:10to be somewhere around 10% over that
- 35:13same time frame. So, I understand the
- 35:14nervousness in the short term given
- 35:16Jensen's sudden departure and questions
- 35:17regarding the future of company
- 35:19leadership. I even said repeatedly going
- 35:20into earnings that to put it bluntly,
- 35:22market participants don't really care
- 35:23about quarterly results given
- 35:25fundamental uncertainties about the
- 35:26future of company leadership. That said,
- 35:28if we can just get through this
- 35:29challenging period in the short to
- 35:30medium term, and considering that the
- 35:31stock is very cheap given the growth
- 35:33that's ahead, and given that I expect
- 35:34Nvidia's team will be able to execute on
- 35:36the product roadmap over the next couple
- 35:38years with or without Jensen, I can't
- 35:40help but think that this is a very
- 35:41attractive setup. This is not financial
- 35:43advice, and I'm not telling you what to
- 35:44do. I'm just sharing some news, my
- 35:46opinions, and some thoughts that I think
- 35:47may be helpful to consider. If you need
- 35:49help with your unique situation,
- 35:50consider reaching out to a certified
- 35:52financial professional who can help you
- 35:53with your unique circumstance. I think
- 35:55the stock's very cheap. I think the
- 35:56product roadmap is largely intact for at
- 35:58least the next couple of years. I expect
- 36:00Nvidia to grow revenue and earnings at a
- 36:02significantly greater rate than the S&P
- 36:04500 over that time. And yet, Nvidia is
- 36:06trading at a roughly 40% discount to the
- 36:08index on a forward P basis last I
- 36:10checked. I know the past month and a
- 36:11half has been rough for AI
- 36:13infrastructure stocks and especially
- 36:14rough over the last couple weeks with
- 36:15the news about Jensen. I understand the
- 36:17short-term nervousness and the
- 36:18uncertainties that are weighing on
- 36:20Nvidia stock over the short to
- 36:22medium-term. But based on what I can see
- 36:23currently, the fundamentals remain very
- 36:25strong. The stock is very cheap and I
- 36:27think the situation will largely be
- 36:28resolved over the coming months. That
- 36:30said, I don't know where the bottom is,
- 36:31and I don't know how long this will
- 36:32last. With that in mind, it's important
- 36:34to remember that as an investor, you
- 36:36never have to buy a position all at one
- 36:37time, and you never have to sell a
- 36:38position all at one time. There's
- 36:40nothing wrong with gradually dollar cost
- 36:41averaging into a position consistently
- 36:43in small increments over a long period
- 36:45of time. And that's usually a good way
- 36:47for most investors to gain exposure in a
- 36:49relatively stress-free manner. That way,
- 36:51if you buy and the stock goes lower,
- 36:52great. You have money left over to buy
- 36:54more at the lower prices because you
- 36:55didn't buy it all at one time. On the
- 36:56other hand, if you buy and the stock
- 36:58goes higher, great. You're up on your
- 36:59position. Again, I'm not a financial
- 37:01adviser and I'm just sharing some
- 37:02thoughts that I think may be helpful to
- 37:03consider. I'll briefly mention some
- 37:05memory news. Trendforce is projecting a
- 37:07decline in NAND flash prices in 2027 due
- 37:10to substantial cuts to enterprise SSD
- 37:12orders from Chinese CSPs coupled with a
- 37:14potential contraction in consumer
- 37:15electronics demand. At the same time,
- 37:17Trendforce reportedly projected that
- 37:18HBM4 prices could rise substantially in
- 37:212027 as DRAM supply tightens. Trendforce
- 37:24reportedly projected at its road show
- 37:26and sold that HBM4 prices would reach 3
- 37:28to $5.50 50 per gigabit versus $25 per
- 37:32gigabit this year, representing a
- 37:34year-over-year increase of approximately
- 37:3646 to 168%. I know that's a wide range,
- 37:39but the underlying point is that they
- 37:41expect HBM4 prices to be meaningfully
- 37:43higher in 2027 than in 2026. I also
- 37:46noticed that Micron CEO and CTO spoke at
- 37:48the Tomorrow Xummit on Wednesday. Key
- 37:50points include 75% of Micron's planned
- 37:532027 output is already sold. Management
- 37:55expects memory supply to be tighter in
- 37:572027 and 2028 than it is in 2026. Demand
- 38:00remains robust with no recent
- 38:02deterioration, and management discussed
- 38:03stronger shareholder returns after
- 38:05restrictions related to the chips act
- 38:06expire on December 9th. I continue to be
- 38:08bullish on Micron despite the recent
- 38:10sell-off. While there are similarities
- 38:12between this current moment and the
- 38:13typical memory cycles of the past, there
- 38:15are also fundamental differences. I
- 38:16expect Micron's pricing power and
- 38:18therefore the durability of both
- 38:19earnings and share repurchases to last
- 38:21longer than the typical memory cycles of
- 38:23the past. Looking ahead, Nvidia is
- 38:24scheduled to present at the 30th annual
- 38:26UBS Global Technology and AI conference
- 38:29on December 1st, starting at 6:35 a.m.
- 38:31Pacific, 9:35 a.m. Eastern. Now, in case
- 38:34you're new to the channel, I want to
- 38:35make sure that you have at least a basic
- 38:36understanding of the underlying
- 38:38long-term thesis. So, let's cover that.
- 38:39Now, I don't know what's going to happen
- 38:40in the short term, but from a long-term
- 38:42perspective, I am very confident that
- 38:43Nvidia will be worth much more in future
- 38:45years than it is today based purely on
- 38:47the fundamental growth of the business.
- 38:48Based on everything I'm seeing, the
- 38:50world is still compute constrained and I
- 38:52expect that to continue at least through
- 38:54calendar 2028. In a computed
- 38:55environment, developers will use
- 38:57whatever viable compute they can get
- 38:58their hands on. Today, there are no GPUs
- 39:00that are sitting dark due to a lack of
- 39:02demand, like there was fiber sitting
- 39:03dark due to a lack of demand at the
- 39:05height of the dot bubble. Back then,
- 39:07companies were laying fiber in the hopes
- 39:08that use cases and demand would
- 39:09eventually show up. Today, we are seeing
- 39:11the complete opposite. As I've said many
- 39:13times, when market participants compare
- 39:15this AI revolution to the dotcom bubble,
- 39:17they ignore the fact that the internet
- 39:18is already here this time. This means
- 39:20that mass adoption of the technology and
- 39:22new use case development at scale are
- 39:24immediately possible. We don't have to
- 39:25wait years for it to show up. It's
- 39:26already here. The world is compute
- 39:28constrained, which means there is not
- 39:29enough supply to satisfy demand. New
- 39:31capacity is utilized as soon as it comes
- 39:33online. The hyperscalers are monetizing
- 39:35capacity as soon as it comes online.
- 39:37Each of the hyperscalers spoke about
- 39:39being supply constrained on their most
- 39:41recent earnings calls. Additionally,
- 39:43many of the clouds are building out into
- 39:44contracted demand. They're not blindly
- 39:46building in the hopes that demand will
- 39:47eventually show up. No, they're building
- 39:49out because they have signed contracts
- 39:51and in some cases significant
- 39:52prepayments from their paying customers.
- 39:54This AI revolution is fundamentally
- 39:56different from the dotcom bubble. And
- 39:582026 will be a pivotal year for the AI
- 40:00industry thanks to the rapid adoption of
- 40:02Agentic AI and the proliferation of
- 40:03Agentic systems in the world's leading
- 40:05enterprises. The leading AI labs
- 40:07revenues are surging right now. Agentic
- 40:09coding and the implementation of agentic
- 40:11systems in large enterprises are new use
- 40:12cases that are increasing inference
- 40:14demand significantly that subsequently
- 40:16is increasing compute demand. The rapid
- 40:18adoption of agentic AI is why we're
- 40:20seeing an inflection in inference
- 40:22demand. It's why we're seeing the
- 40:23leading AI labs revenues surge. I think
- 40:25the leading labs surging revenues may be
- 40:27the initial proof point that grabs
- 40:29market participants attention and causes
- 40:31them to realize that there will be a
- 40:32clear ROI on AI infrastructure. I think
- 40:34the leading labs surging revenues will
- 40:36also help assure investors of the
- 40:38longevity of Nvidia's growth since these
- 40:40labs revenues are directly tied to
- 40:41compute. If they had more compute, they
- 40:43would have greater revenues. It really
- 40:44is that simple. Demand is not the
- 40:46problem. The problem is a lack of supply
- 40:48to meet the demand. As I've said
- 40:49previously, I expect the world to be
- 40:51computed at least through calendar 2028,
- 40:53possibly longer. And so regardless of
- 40:55what happens in the short term, it's
- 40:56important for long-term investors to
- 40:58remain focused on the fundamentals,
- 40:59maintain a long-term perspective, and
- 41:01remember that we are only in the early
- 41:03stages of aic systems being adopted at
- 41:05scale. This will increase compute demand
- 41:07significantly, and after that, the next
- 41:08surge in compute demand will likely be
- 41:10fueled by physical AI. We're no longer
- 41:12talking about digital agents performing
- 41:14digital tasks. With physical AI, we're
- 41:16talking about physical AI agents
- 41:17performing physical tasks in the real
- 41:19world. NVIDIA CFO has called physical
- 41:21AI, quote, a multi- trillion dollar
- 41:23opportunity. and the next leg of growth
- 41:24for Nvidia. This industry will
- 41:26fundamentally transform society and
- 41:28Nvidia has positioned themselves to
- 41:30benefit massively. NVIDIA sells the
- 41:31hardware for the data centers where the
- 41:33models are trained. They offer omniverse
- 41:34where the models are taught and tested.
- 41:36And Nvidia also sells the hardware that
- 41:38allows ondevice real-time inference
- 41:39through NVIDIA AGX allowing robots to
- 41:42have intelligent interactions with the
- 41:43real world even when they are not
- 41:44connected to a data center. Notice that
- 41:46Nvidia is taking a holistic platform
- 41:48approach to physical AI and they're
- 41:50embedding themselves as the underlying
- 41:51foundation supporting all of it. Over 3
- 41:53million developers are already building
- 41:55on the Nvidia robotic stack and this is
- 41:57not getting enough attention. As for
- 41:58production ramps, Blackwell Ultra has
- 42:00ramped and remains in high demand. Vera
- 42:02Rubin is rolling out to customers.
- 42:04Nvidia Gro 3 LPX is in full production.
- 42:06Later on, we're expecting the launch of
- 42:08Reuben Ultra in 2027 and finement after
- 42:10that in 2028. We have a clear data
- 42:12center product roadmap stretching into
- 42:142028. I don't think we are anywhere near
- 42:16any type of bubble bursting type of
- 42:17event. With all of this in mind, I
- 42:19seriously think that Nvidia still has
- 42:21plenty of runway ahead of it. And I
- 42:22think this company will be worth
- 42:23substantially more in future years than
- 42:25it is today. At least that's my view of
- 42:26the situation. Quick note before I wrap
- 42:28up. All of the compilations on this
- 42:30channel are edited by Finn Vid with
- 42:31original structure and commentary.
- 42:33Occasionally, the same edits appear
- 42:34elsewhere on YouTube. If you're looking
- 42:36for the original version, it's always
- 42:37here on this channel. Thanks for
- 42:38watching Finn Vid. I appreciate your
- 42:40support. Remember to stay calm in this
- 42:41market. Remember to maintain a long-term
- 42:43perspective and do not make any hasty or
- 42:45irrational decisions. With all of that
- 42:47being said, I hope you all have a great
- 42:48rest of the day. And I'm curious to hear
- 42:50your thoughts about Nvidia in the
- 42:51comments below. Please leave a like on
- 42:52this video so more people will see it.
- 42:54And while you're down there, please
- 42:55consider subscribing. It's free and you
- 42:57can always change your mind. Thanks for
- 42:59watching and hopefully I'll see you in
- 43:00the next
About this transcript
This page contains the full transcript of CNBC & Fox Today On NVIDIA Stock, Anthropic IPO, Micron Stock - NVDA Update by FinVid, generated from the public captions YouTube serves with the video. The transcript has 9,010 words across 1,328 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.