¿Ha tocado fondo Bitcoin?

[0:00] July 29th when we have the FOMC, are
[0:02] they going to raise rates? There's a
[0:04] possibility, a good possibility. It's
[0:06] like 35 to 40% that they will.
[0:09] If they don't, then I think Bitcoin will
[0:12] be above 70,000 when that time comes
[0:14] because I think people are going to
[0:15] start factoring in the fact
[0:17] >> What's going on, guys? Today we have a
[0:18] great conversation with Jordy Visser. We
[0:20] talk about what's going on with the AI
[0:22] mid-cycle slowdown, why he thinks that's
[0:24] over, why is he getting excited about
[0:25] Bitcoin. He thinks that the bottom is
[0:27] forming. On top of that, we talk about
[0:28] robotics, commodities, and what public
[0:30] stocks Jordy's getting interested in.
[0:32] This conversation is full of lots of
[0:33] nuance, and we cover some of the big
[0:35] stories, things like Apple and what's
[0:37] going on with Meta or XAI and Grok. So,
[0:40] you're going to enjoy it. Here's my
[0:41] latest conversation with Jordy Visser.
[0:43] All right, Jordy, this week Grok and
[0:45] Facebook both made very big
[0:46] announcements about their ambitious
[0:49] plans when it comes to AI. It looks like
[0:50] we have a price war on our hands as both
[0:52] of them start talking about
[0:53] affordability for what they can actually
[0:55] produce. What's your take as to these
[0:57] two kind of trying to make a comeback
[0:59] and and take over Open AI and Anthropic?
[1:02] >> Uh so,
[1:04] >> [snorts]
[1:05] >> Ooh, speechless.
[1:07] >> [laughter]
[1:07] >> No, I I was trying to think of the best
[1:09] way to kind of present this. I I really
[1:11] am
[1:13] believing that
[1:15] uh
[1:16] we have to start separating
[1:18] everything going on in AI and the models
[1:21] into two categories. I don't think
[1:23] anyone anymore can catch up to Anthropic
[1:27] and Open AI and what their plans are.
[1:29] Now, let me just kind of put that in
[1:31] into context. I've spent a lot more time
[1:34] on consumer agents. I've spent a lot
[1:35] more time on what these models are
[1:37] trying to do and what everyone's kind of
[1:40] going for. They're all trying to get
[1:42] some kind of revenue, all of them. So,
[1:44] when Meta talks about the cloud
[1:46] ambitions, they need some revenue coming
[1:49] in the door now, and if your business is
[1:52] going to be consumer agents,
[1:54] advertising, stuff like that, and
[1:56] personal AI assistants, where which is
[1:58] where I am now putting officially
[2:00] Meta, Apple, Google, Amazon. Now again,
[2:05] not completely, but they all have that
[2:08] part of the puzzle. Amazon, Google, and
[2:11] Microsoft obviously not only have the
[2:13] personal assistant, the co-pilot side,
[2:16] they also have the cloud business.
[2:18] Obviously, we have cloud business now
[2:20] where we're renting out compute for
[2:22] SpaceX. So, I think rather than getting
[2:25] to the model side, which again, these
[2:28] models are just going to be
[2:29] commoditized. Uh it's what what you're
[2:32] aiming to get your revenues from and
[2:34] clearly OpenAI and Anthropic are in a
[2:36] war uh with the enterprise side and then
[2:40] everyone else to me is kind of making
[2:42] sure that they're also spending in other
[2:44] areas where they need this and obviously
[2:46] SpaceX,
[2:48] you know, Tesla eventually, they've got
[2:50] different ambitions for how they're
[2:52] going to use AI. So, I think I I'm not
[2:54] really looking at the price wars as
[2:56] anything more that number one, the media
[2:58] is just kind of getting into this oh,
[3:00] the token index is going down, it means
[3:02] we're going to have problems. I think
[3:03] everyone's just re-examining where their
[3:05] ROIC is going to come from and they're
[3:07] strategically moving their
[3:08] decision-making towards that process.
[3:10] >> I love when we uh maybe don't disagree
[3:13] but have slightly different takes on
[3:14] this.
[3:15] >> Mhm.
[3:15] >> Um
[3:16] in building Silvia, I think you and I
[3:18] have talked in the last couple of weeks
[3:19] about, you know, the mandate from heaven
[3:21] 18 months ago everyone use AI. Now all
[3:23] of a sudden people start looking at the
[3:25] uh token prices or the the compute, you
[3:27] know, cost on a monthly basis was
[3:29] sky-high. They all start cutting that,
[3:31] but a lot of what people have been doing
[3:34] both, you know, what we've been doing at
[3:35] Silvia, but also in other companies I've
[3:37] talked with, they essentially were
[3:38] changing the architecture of how they
[3:40] use the models. So, it wasn't they were
[3:42] changing the model, it was just hey, we
[3:43] don't want to ping the model for if you
[3:45] ask what the date is or you know, that
[3:46] type of stuff, right? Now,
[3:48] I think that we've kind of gradually
[3:50] passed that. What we're actively looking
[3:51] at and I think a lot of other people are
[3:53] looking at, is how do you actually start
[3:54] to route queries to different models
[3:57] based on the complexity or the compute
[3:59] need of that query. And so, this feels
[4:02] like an area where you're going to see a
[4:04] significant drop-off in the compute cost
[4:07] of these companies in terms of what
[4:08] they're spending.
[4:09] And the reason why that becomes
[4:11] interesting is then it opens up this
[4:12] whole open-source thing. And so, to me,
[4:15] the Facebook and xAI announcements
[4:18] around cost is basically them trying to
[4:20] position themselves to say, "Hey, maybe
[4:22] you don't need to go to Chinese open
[4:23] source or look for American open source.
[4:25] What if you could use one of these
[4:26] closed-source models, but do it at a
[4:28] much cheaper rate?" It's almost like
[4:30] they're acknowledging the fact that all
[4:31] of these companies are now becoming very
[4:32] cost-conscious on this. I don't know if
[4:35] it'll work. Like, I don't know if
[4:36] they'll actually drive more adoption,
[4:38] but it seems odd that two different
[4:40] companies who probably are not sharing
[4:43] their plans with each other in the same
[4:45] week come out, and they're leading with,
[4:47] "Hey, this is a cost-effective way to
[4:49] get super intelligence, you know, via
[4:50] the model, right?"
[4:52] >> Yeah, so I I I don't agree with you, so
[4:54] we'll keep going. Um
[4:56] I The part I agree with is the fact that
[4:59] and and I I put together a visual
[5:01] because a bunch of people reached out
[5:02] and asked me the question on from a
[5:05] variety of angles on this point. And
[5:07] then I had some people that I've known a
[5:09] long time in the tech side that are all
[5:10] bearish on this going, "Oh, everyone's
[5:12] going to move to cheaper model." And I'm
[5:13] like, "That's
[5:14] it We got to start to understand where
[5:16] we are." So, I'm going to give it to you
[5:17] in a different way. You've probably not
[5:19] heard it this way. Um and you're looking
[5:21] at it from the angle of Sylvia, which I
[5:22] get.
[5:23] In my opinion, having managed hundreds
[5:25] of people over the years, um you bring
[5:28] in a resume, and you don't need to hire
[5:32] the most expensive people throughout
[5:34] your firm for every job. Jobs are now
[5:37] finally getting delegated to the models
[5:39] that make sense, meaning a 100 a 100 uh
[5:42] IQ might be enough for something if it
[5:44] does if you don't need it quickly. So,
[5:46] it's getting back to the way I look at
[5:48] aggregate payrolls, which is hours
[5:50] worked,
[5:51] then you've got hourly earnings. So,
[5:55] and how many people you're employing,
[5:56] that is the aggregate payroll, and I
[5:58] think every workforce is figuring out
[6:02] these models that the very high end,
[6:04] they're super high IQ, they're very
[6:06] cheap. And Dylan Patel gave an interview
[6:09] that I listened to today on WisdomTree.
[6:12] It's a great podcast, I highly
[6:13] recommend. He talked about his spend.
[6:15] So, this is Dylan Patel from Semi
[6:16] Analysis, very sophisticated. His spend
[6:20] went from $100,000
[6:22] all in for his firm as of
[6:26] December last year.
[6:29] It's now 11 million
[6:31] ARS as he said, annualized revenue
[6:34] spend. Now, what he talked about as the
[6:37] importance of that is he's not changing
[6:39] it, he's not looking to get it done. He
[6:41] said in some cases, because the
[6:43] productivity is expanding, he's able to
[6:45] generate more revenues. I think every
[6:48] business is going to be different. I
[6:49] think for the enterprises that had no
[6:52] usage for them whatsoever, they're
[6:55] figuring out where they should be making
[6:57] the spend. And in some case in every
[6:59] case, they will be having workflows that
[7:01] are using cheaper models, open source
[7:03] possibly, then you're going to have more
[7:06] expensive high IQ making decisions where
[7:08] they're very effective and very
[7:10] cost-effective because they do the speed
[7:13] in 1/10 or 1/100 the time that those
[7:16] workflows do. So, it just depends on
[7:18] where it is. The end result is going to
[7:19] be a mixture of all these different
[7:21] models, completely agree. We were always
[7:22] going to end up there, but I think one
[7:24] of the things that's been challenging is
[7:26] that since Opus 4.5, the model
[7:29] intelligence for coding agents just
[7:31] keeps getting better and better, but the
[7:33] deployment from enterprises just
[7:36] started. And so, they spent the first 6
[7:38] months actually learning how much the
[7:40] spend would be. Now, they're going to
[7:42] start going through a slight
[7:43] readjustment. Palantir is obviously out
[7:46] there promoting the fact that you don't,
[7:48] you know, they're going another angle.
[7:50] Yo, you don't want to have this. You
[7:51] don't want to be exposed. They're giving
[7:53] all the negatives to having your own
[7:54] models. I think we're in the the price
[7:56] discovery phase for everything. The
[7:58] models are going to keep getting more
[7:59] efficient. They're going to keep getting
[8:00] cheaper. They're going to keep getting
[8:01] smarter. I think there's going to be a
[8:03] mixture of both, but I don't think this
[8:05] is going to affect the overall spend. I
[8:06] think adoption and usage is going up
[8:08] significantly, dramatically each day.
[8:12] >> Yeah, and and maybe as a
[8:14] a nuance to it,
[8:16] I actually think that the number of
[8:18] tokens consumed will continue to grow.
[8:20] So, I agree with that. But I think that
[8:22] what companies are trying to figure out,
[8:24] at least what we're trying to figure
[8:24] out, is how do you consume more tokens,
[8:27] but spend less money. And I think what
[8:29] we are starting to see is that, you
[8:31] know, if you go hit 4.7 or 4.8 or, you
[8:34] know, whatever,
[8:35] uh and someone's asking, "What is the
[8:36] date?" Not only is there a cost
[8:38] associated with it, which obviously is a
[8:40] uh like a pain point for the company,
[8:42] but there's a a big latency. And so,
[8:44] it's actually like a bad experience for
[8:46] the user. Versus if you can, you know,
[8:49] kind of hit a lightning-fast open-source
[8:51] model, make sure it's accurate, et
[8:52] cetera, then it becomes uh somewhat
[8:54] interesting. Um but but some companies
[8:57] don't care, right? They don't care about
[8:58] the cost. May- maybe Dylan Patel is just
[9:00] like, "Look, I'll spend $50 million
[9:02] because every dollar that I spend I'm
[9:03] getting back $10 in revenue." Then, you
[9:06] know, it kind of doesn't matter, right?
[9:07] >> Yeah, and that and that's the point is I
[9:10] the the way this weeds out the bad
[9:12] companies, and I I think his exact quote
[9:14] was,
[9:15] "If you're not spending more money every
[9:17] year, you're going out of business."
[9:19] Like if you're trying to to optimize
[9:21] based on I don't want to spend X amount
[9:23] of dollars, and you're cutting back on
[9:25] your AI spend,
[9:26] it means you're going out of business.
[9:28] And I happen to agree with that.
[9:29] Meaning,
[9:30] your budget for AI should be expanding
[9:32] every single year, no questions asked.
[9:35] It should absolutely be going up. And
[9:37] that is the only way you're going to
[9:38] survive as a business. So that makes
[9:40] companies that are startups have a huge
[9:42] advantage and that's why semi analysis
[9:43] is very different than say the sem you
[9:46] know, they cover data centers, they
[9:48] cover semi they cover semiconductors,
[9:50] they cover the entire AI build out. If
[9:51] you're Goldman Sachs and Morgan Stanley,
[9:54] you've got a huge payroll, you have a
[9:55] huge thing, you're going to have a
[9:57] different approach than a company that's
[9:58] just starting was only spending $100,000
[10:00] a year on AI in December.
[10:03] >> Yeah, that makes sense. What about
[10:04] Apple? Apple obviously made the Siri
[10:06] announcement kind of an AI Siri
[10:09] integration and then they also announced
[10:10] that they're going to have to increase
[10:11] prices because of memory. What do you
[10:14] think's happening with that business?
[10:16] >> I actually so
[10:19] last week uh
[10:22] because I've been spending so much time
[10:23] on Meta and because I've so I listened
[10:26] to more podcast with Meta involved. I
[10:29] listened to one with Alex Kantrowitz
[10:32] this week. He had on the CTO from Meta.
[10:35] And I'll be covering that in the
[10:36] YouTube. Whenever there's a story that's
[10:39] going around that I'm trying to figure
[10:40] out what's behind it. They're getting
[10:42] into cloud. They're saying the agentic
[10:44] world is is not where they thought it
[10:46] would be. Well, then I got to go look
[10:48] up. So I start doing a ton of research
[10:50] on consumer agents and this is the way I
[10:52] want to break it down for people as I
[10:54] get into the Apple discussion.
[10:56] So far all we've done is coding agents
[10:57] and I want people to understand that
[10:59] that is basically the language of AI. So
[11:02] we're using more coding agents and that
[11:05] agentic side where the workflows are
[11:07] getting longer, they're getting more
[11:08] complex, they need some kind of memory.
[11:12] We've we've seen token usage go up
[11:14] exponentially.
[11:16] Consumer agents are going to make it
[11:18] have another move higher. And in
[11:20] listening to Meta multiple times and
[11:23] people talk about it, the complexity
[11:26] with consumer agents and what has to go
[11:28] on, the commerce side, the hey, book me
[11:31] a trip to Italy. Here's what I care
[11:34] about. That's far more complicated than
[11:36] build me this spreadsheet, build me this
[11:38] model. You have to do a lot more things.
[11:40] You have to visit a lot more sites. It
[11:42] takes a lot more compute. So, the step
[11:45] function of compute is going to go
[11:47] exponentially when these guys can solve
[11:49] this. The reason the Apple thing is
[11:51] interesting is the same thing with Meta
[11:52] because
[11:54] Google, Amazon, Amazon has Alexa, Google
[11:57] has personal
[11:58] intelligence, Apple which is trying to
[12:00] get Siri to work, and then Meta which is
[12:03] trying to monetize Facebook, Instagram,
[12:05] everything they have in some way with
[12:07] advertisements, but also with AI
[12:09] combined in, as well as all these places
[12:11] having some kind of hardware. Again,
[12:13] Alexa,
[12:14] the phone. You go through it. Everyone
[12:17] has hardware out of this group. So, the
[12:18] question is
[12:20] think about the how difficult it is that
[12:22] wow, Opus 4.8, 4. We've got Fable 5, but
[12:25] yet Siri still doesn't work. Alexa still
[12:28] doesn't work. How's that possible? Well,
[12:30] the complexity of what they're trying to
[12:32] solve is impossible. I was in my Tesla
[12:37] using Grok again this week and Grok in a
[12:39] Tesla is phenomenal. The latency is is
[12:42] is very good. Why can't they do this
[12:45] with Siri? Well, the car is on all the
[12:47] time. I'm just driving in it. If you had
[12:49] your phone open all the time, you could
[12:51] converse with ChatGPT, it would work.
[12:54] So, I think the most interesting thing
[12:56] with Apple is the take that I normally
[12:58] have. It sold off made all-time highs
[13:00] when they announced Siri personal
[13:02] intelligence and they showed what it was
[13:04] able to do. You had some people saying
[13:06] it was horrible. You had a lot of people
[13:08] saying it was shockingly good.
[13:10] Then they had the memory price side.
[13:13] Okay, memory prices are going higher.
[13:14] We're going to have to raise prices. The
[13:16] stock got hit both of those days. On the
[13:18] Siri day, it made all-time highs and
[13:20] sold off, and then it sold off on the
[13:22] memory side, leaving it down 10 per-
[13:24] more than 10% off the highs.
[13:26] And then with no news, it went right
[13:29] back up to the highs.
[13:30] This is not a cheap stock. This tells me
[13:33] this is a bullish reaction to bad news.
[13:36] And I like looking into things. I think
[13:39] the market, and particularly the very
[13:41] smart people, are starting to look ahead
[13:43] to the next 6 months. So, this is the
[13:45] first warning I'm going to give to
[13:46] people. Uh the infrastructure trade was
[13:49] a great 6-month trade. I still think
[13:51] you're going to make a lot of money
[13:52] being long the receivers or the the
[13:54] people getting the money on the
[13:56] infrastructure build-out. But, it's not
[13:58] going to be as easy. So, I'm calling the
[14:00] the AI mid-cycle slowdown is over when
[14:02] it comes to terms of I think now
[14:04] everyone's a little bit too negative on
[14:06] the AI build-out side and they're
[14:07] worried about this compute thing. Where
[14:09] I think this is all shifting to is two
[14:11] themes. One is when consumer agents
[14:14] come, you have to reprice Apple, you
[14:16] have to reprice Meta, you have to
[14:18] reprice these stocks cuz the ROIC is
[14:20] going to come through, and I think
[14:21] people are getting more bullish on these
[14:23] companies with the reality that that's
[14:25] coming. The second angle is this is
[14:27] where you start getting into the crypto
[14:29] side because of the agentic commerce.
[14:30] They're linked together, and so I'm
[14:32] looking towards, you know, a year from
[14:34] now, the major story being how did we
[14:36] miss Apple? I'm not convinced in Meta
[14:39] yet cuz it doesn't have the same
[14:40] technical picture, but I am going to
[14:42] watch the hyperscalers much more closely
[14:44] cuz I think we're getting into a side
[14:45] where the surprise could be that we're
[14:47] moving into consumer agents and away
[14:49] from the coding agents.
[14:51] >> What do you do with Grok in your Tesla?
[14:54] >> This was really really more
[14:55] conversational. So, I wanted In fact,
[14:58] the irony is I wanted to have a
[14:59] conversation about exactly what I just
[15:01] talked about. So, the way that I I use
[15:03] it when I'm driving, especially in Maine
[15:05] for say an hour, and I've just listened
[15:07] to a podcast, I will pause the podcast,
[15:10] I will go to Grok, and I will now have a
[15:11] conversation to expand that
[15:13] conversation.
[15:14] It's much better in the car than it is
[15:17] walking around with ChatGPT. And the
[15:19] reason is because it's on all the time,
[15:22] and the signal is much better inside the
[15:24] car. So, Grok is more human-like for me
[15:27] having the conversation. And so again,
[15:29] this thing of having a conversation with
[15:31] a really smart person, it's literally,
[15:33] "Hey, I want to talk about Apple and
[15:35] Meta and consumer agents." And I'll just
[15:37] give you one, you know, one
[15:38] [clears throat] factoid that came out of
[15:39] that. Goldman Sachs had a report out
[15:42] this year saying consumer agents will
[15:43] consume
[15:45] about 30 times
[15:47] 30 times what coding agents will consume
[15:50] from a compute basis. This is why there
[15:53] is no solution to compute. If you think
[15:56] there is excess capacity of compute, you
[15:58] are absolutely wrong.
[16:00] >> Ladies and gentlemen, I'm a software
[16:02] guy, but every once in a while I come
[16:03] across a piece of hardware that I just
[16:04] simply can't live without, and that's
[16:06] Plaud for me now. Now, this whole thing
[16:08] is basically a hedge against me losing
[16:10] my memory, because when I have all these
[16:12] conversations, I talk to founders,
[16:14] investors, all kinds of people every
[16:15] single day. There's three things that I
[16:17] always want to take away from the
[16:18] conversation: what I learned, what are
[16:20] the next steps action items, and then
[16:21] what notes do I need to remember for
[16:23] some period in the future. You can even
[16:25] mark important moments during the
[16:26] discussions so that you can quickly
[16:27] reference them later, whether it's a
[16:28] founder meeting, an investor call, a
[16:30] podcast interview, or maybe even just
[16:32] want to sit down and talk to your
[16:33] spouse. I could instantly go back and
[16:35] forth and find key insights, decisions,
[16:37] action items that I talked about in that
[16:39] conversation. Over time, every
[16:41] conversation all of a sudden becomes an
[16:42] asset instead of a forgotten moment. And
[16:44] in a world where information is
[16:46] abundant, the ability to retain and
[16:47] leverage knowledge is a very real
[16:49] competitive advantage. That's why Plaud
[16:51] Note Pro caught my attention recently.
[16:52] It's a credit card-size note-taker,
[16:54] slim, portable, perfectly integrated
[16:56] with my phone. And the part that I love
[16:57] about it is that it automatically
[16:58] captures conversations, turns it into
[17:00] board-ready summaries. You can have
[17:02] structured action items, and it does
[17:03] flawless transcription. It is really
[17:05] magical and you should try it out. So,
[17:07] go check them out. This hardware has
[17:09] completely changed my life, and I think
[17:10] that you'll end up finding it very
[17:12] valuable. So, if you want to use Plaud
[17:13] and get the same value out of it that
[17:15] I've gotten out of it, go check them out
[17:16] at plaid.ai/pomp.
[17:19] You can go to plaid.ai/pomp and get 15%
[17:22] off if you use the code pomp or just go
[17:24] click on the link in the description.
[17:25] So, go to plaid.ai/pomp
[17:28] today.
[17:28] >> Now, when we look at the compute issue,
[17:31] obviously that's one of the big things
[17:32] people saw meta's announcement, they
[17:34] started getting worried about that.
[17:37] We continue to see volatility in a lot
[17:40] of these names. Bloom Energy, now
[17:42] there's a whole bunch of these short
[17:43] reports that are starting to come out
[17:45] and it almost feels like the people who
[17:46] are pessimistic, they are just waiting
[17:49] to pick their shot and they think that
[17:50] all this stuff is overvalued. Can short
[17:53] sellers make money in a market like
[17:55] this? Like how do you think about the
[17:57] other side, right? You're bullish on
[17:59] this, I'm bullish on this, many people
[18:00] are buying these names. What about the
[18:02] other side? Like do they have any
[18:05] credible critiques that you think people
[18:07] should pay attention to?
[18:09] >> Well, of course and let's go back to me
[18:12] selling out of Micron but still
[18:14] believing the memory trade is going to
[18:16] continue.
[18:17] In fact, one of the things just to give
[18:19] people a preview
[18:21] um
[18:23] part of the reason I I sold out of
[18:24] Micron was because the speculation
[18:26] everywhere was starting to become
[18:28] evident.
[18:29] And again, it
[18:31] it was leverage.
[18:33] So, when I see leverage forming and
[18:36] obvious in Korea and I see leverage in
[18:39] macro strategies and I see leverage in
[18:42] fundamental long short strategies where
[18:44] six months ago they were arguing with me
[18:46] on memory already being expensive and
[18:48] now they're bragging about the fact that
[18:50] this has five years to go. I just start
[18:53] to reduce the risk and I think there's
[18:54] better places to put my money.
[18:56] Um
[18:58] and now I have to get my brain around
[18:59] the fact that to buy Micron back, I'm
[19:01] going to have to buy it above where I
[19:03] sold it. There's absolutely no question.
[19:05] Now, I don't have a problem with that
[19:06] cuz I've been a trader my life. But I do
[19:08] think that there are ways Micron went up
[19:10] to 1300. So, if you sold it at, you
[19:13] know, 1250
[19:15] and you bought it back at 900, you can
[19:17] trade, you know,
[19:19] bubbles, parabolas, and all these things
[19:21] that I've talked about. You will have 30
[19:23] to 40% corrections. And in a lot of the
[19:25] these names, we've had some of the
[19:26] largest corrections in the history of
[19:28] their stocks. And that's that's, you
[19:30] know, serious. Samsung,
[19:33] uh, over a 10-day period was down 20%.
[19:36] Um, and that's despite
[19:39] making, uh, enough money to basically to
[19:42] be more than they've made in their
[19:43] 40-year history. That's what's expected
[19:45] for this year. So, you can have price
[19:47] falls. You can short Bloom Energy. I
[19:50] just think people are making a mistake
[19:52] if they're thinking about this longer
[19:53] term. My general rule is as a macro
[19:56] person, I try to find which way the
[19:58] river is flowing.
[20:00] And I'm either in that side of the river
[20:02] going up, or I'm just not in the trade.
[20:04] And that's the way I thought about
[20:05] Micron. I still think you'll see 4,000,
[20:07] 5,000 in Micron before this is done. I
[20:10] think it'll be harder to ride it from
[20:11] 1,000 to 5,000 than it was to go from
[20:14] 100 to 500. And that's my general take
[20:16] is I I'm going to trade it more often.
[20:18] I'll get out. I'll get in. And I'll look
[20:20] for points. Uh, but I do think for
[20:22] people who want to short them, they just
[20:23] have to have a shorter time horizon.
[20:25] >> You you mentioned Samsung. Talk a little
[20:27] bit about them. They reported these like
[20:29] blowout earnings, but they sold off.
[20:31] What what is going on there?
[20:33] >> Yeah, like I said, I mean, the numbers
[20:35] are astonishing. And for people who who
[20:37] just haven't read the story,
[20:40] uh, they're they're expected to make 217
[20:45] billion dollars this year.
[20:47] I mean, I
[20:47] >> With a B? With a B?
[20:49] >> That's how much they're expected to
[20:50] make. Not the revenues, how much their
[20:52] earnings are supposed to be this year.
[20:55] Over the last 40 years, they haven't
[20:57] made a combined 217 billion dollars.
[21:01] So, you have to put that in the context
[21:03] and realize that, number one, when
[21:04] people say this is a bubble,
[21:07] it's not a bubble. Like the earnings are
[21:08] massive. Number two, the amount of chips
[21:10] they're selling and how far into the
[21:13] future they're going to be selling this
[21:14] stuff,
[21:16] it's just not easy to build memory
[21:17] capacity. So, Samsung along with SK
[21:20] Hynix and Micron have benefited
[21:22] significantly and they're going to
[21:24] continue to benefit significantly. But
[21:26] like I said, if you go back over the
[21:28] last 25 years, the the the only time
[21:31] they had a correction greater than 20%
[21:34] over 10-day period, I believe was when
[21:36] the world shut down for a virus that was
[21:38] going to kill everyone and it was very
[21:40] similar to one that happened just after
[21:42] Lehman Brothers. Those are the only two
[21:44] times in 25 years
[21:46] and yet it's still near the all-time
[21:48] highs even though it's off 20% and it's
[21:50] still a parabola. Here's the thing I
[21:52] would say. I saw a lot of people be very
[21:54] negative because one of the things that
[21:56] we like to do is, "Hey, they sold off on
[21:59] good news."
[22:01] The problem is they were already down
[22:02] 15%
[22:04] off the highs when that happened. So, I
[22:07] I don't care about it nearly as much.
[22:09] What I look for in those types of sell
[22:10] the news things is when the stock is at
[22:13] all-time highs, it opens higher and then
[22:14] it sells off and it catches everyone and
[22:17] then that usually leads to a two-month
[22:19] or so correction. These things have been
[22:21] correcting now for about four to six
[22:23] weeks and I say four to six weeks even
[22:25] though they made new all-time highs. You
[22:27] had started to see the weakness in late
[22:29] May and that's when I really started to
[22:31] pound the table on, "Okay, I really
[22:33] think they had a mid-cycle slowdown
[22:35] there and all we've done is wipe out the
[22:37] leverage at this point across three
[22:40] buckets. Systematic
[22:41] quant strategies have been forced to
[22:43] take down their leverage. We've also
[22:45] seen Korean investors forced take down.
[22:48] We've seen the ETFs, the two-time ETF in
[22:51] Hong Kong, go down 62%. Like we've seen
[22:54] enough in my opinion that we have a more
[22:57] balanced market and if you looked at
[22:59] what SK Hynix did with their ADR
[23:02] issuance and how over subscribed it was
[23:04] on the IPO coming to the market. But
[23:06] then when you read that most of the
[23:08] buyers or a lot of the buyers were long
[23:10] only people, this is the way long only
[23:12] managers are showing their hand, which
[23:14] is they missed this party on the way up.
[23:16] It moved too fast for how fast they
[23:18] deploy money. They're going to be
[23:19] underneath the buy things. So I'm very
[23:20] comfortable that we've done enough
[23:22] damage. Maybe we make a little bit more
[23:24] new lows during earning season,
[23:26] particularly on some stocks, but the
[23:28] volatility is way too high in the
[23:29] momentum side. It's going to come down
[23:31] and it's going to come down through
[23:32] lower correlation amongst the names.
[23:35] >> I know as the AI trade is maybe gearing
[23:38] up to to get out of this mid-cycle
[23:40] slowdown. You've become more bullish on
[23:42] Bitcoin. What's going on there?
[23:45] >> Uh well, it's really I mean, we talked
[23:47] about it, but I'll just re-emphasize a
[23:50] couple things and then I'll get into a
[23:51] bigger topic that I'm starting to to to
[23:54] say to managers and explain why I even
[23:57] I'm starting a YouTube specifically on
[23:59] crypto targeted for September.
[24:02] As people get to know me, I still I'll
[24:05] always be a trader. I'll always look for
[24:06] the best timing on things, but when
[24:08] you're running a business,
[24:10] you want to start things at a time when
[24:12] people are going to start to be looking
[24:13] for them. And I think this is a really
[24:16] important story. So the first thing is I
[24:18] always look for technical signs and I
[24:21] finally got my first RSI divergence
[24:23] since the peak at the end of last year.
[24:28] And that divergence was using basically
[24:31] a 4-hour RSI and I look for points where
[24:35] the price makes new low, which happened
[24:37] when we broke through 60,000 recently,
[24:40] but the RSI is higher than it was at the
[24:42] prior low. Plain and simple. Now, that
[24:44] was the first time that we got one. So
[24:45] as a trader, I go, well, now I can buy
[24:48] something when we get back above 60 and
[24:50] I'll just stop myself back out below the
[24:52] lows. That's the way a trader trades and
[24:54] that's the way that I've always thought.
[24:55] It's the way I was brought up and I'm an
[24:57] Elliott Wave guy, and I still believe
[24:58] we're entering a very, very big thing.
[25:00] My belief is that over the course of the
[25:02] next year, we are near the bottom end of
[25:04] the range of Bitcoin. And see how I said
[25:06] that, near? Could we go to 50? Yeah.
[25:08] Could we go to 45? Yeah. Do I think
[25:10] we'll be over 100 a year from now? Yeah.
[25:12] So, what do I care whether I buy
[25:14] something at 60 or whatever? Yes, my my
[25:17] percentage gain will be better, but if
[25:19] we're above 100, I don't care. So,
[25:21] that's the first thing. The second thing
[25:23] is I've highlighted that
[25:25] we probably all underestimated, and I
[25:28] know I did, the impact that the AI
[25:30] capital suck would have from the rest of
[25:33] everything. It's not just Bitcoin
[25:35] people. It's the hyperscalers as well.
[25:37] Basically, my benchmark arbitrage of
[25:39] money rotating into the space that's
[25:41] going the fastest,
[25:43] Micron went up 20 times. A 20 bagger. 20
[25:47] bagger. You don't get that in big
[25:49] companies, and this is a big company.
[25:51] So, for that to happen,
[25:53] you didn't need to buy Bitcoin. You
[25:55] didn't need to buy crypto. You didn't
[25:56] need to buy any of this stuff. Every
[25:58] single person in Silicon Valley talked
[26:00] about how startups, if you had AI
[26:02] attached to it, nobody wanted crypto. At
[26:04] the end of the day, people want to
[26:05] invest in things that are working. And
[26:07] so, once it started to move lower, which
[26:09] coincided in October with the
[26:12] release of Opus 4.5, but it also
[26:15] coincided with another strong secular
[26:17] trend that has still been in place until
[26:19] last week, which was rate cuts. We had
[26:22] 150 basis points of rate cuts still in
[26:25] the market as of October late September
[26:28] of last year. That peak there, some of
[26:30] it came out because of the actual cut,
[26:32] but then a lot of it came out because we
[26:35] started to build in hikes. And we have
[26:37] about we have a full cut plus
[26:41] percent of Sorry, hike plus a hot
[26:44] probability of another one before the
[26:46] end of the year. So, here's what I'm
[26:47] telling you is going to happen in my
[26:49] mind.
[26:50] Um that's all going to change, or at
[26:51] least the rate of change is going to be
[26:53] there. And I'm a rate of change guys. I
[26:54] do not think they want to hike. I did a
[26:57] piece on the Fed this week and made sure
[26:59] that everyone
[27:00] was highlighted that hey, the Fed war
[27:03] has spoken. He believes AI is going to
[27:05] be a productivity boom and that there
[27:08] might be a point at the beginning where
[27:10] it's a little inflationary before we get
[27:12] the deflationary, but the last thing he
[27:13] wants to do is go on some type hiking
[27:16] cycle because of inflation. We're going
[27:18] to get a negative CPI print. The
[27:19] question is
[27:21] July 29th when we have the FOMC, are
[27:23] they going to raise rates?
[27:25] There's a possibility, a good
[27:27] possibility. It's like 35 to 40% that
[27:29] they will.
[27:30] If they don't, then I think Bitcoin will
[27:33] be above 70,000 when that time comes
[27:35] because I think people are going to
[27:36] start factoring in the fact
[27:39] that maybe they're not going at all this
[27:40] year before the midterms. Does he really
[27:42] want to hike before the midterms when he
[27:43] was brought in by Trump? So, I think
[27:46] that may be on hold for a while and it's
[27:48] not that it's a big deal because it
[27:50] means they're not cutting rates, but
[27:51] since hikes are expected, it is a
[27:53] positive relative to expectations.
[27:56] Those are the ways that I'm kind of
[27:57] looking at it, but I want to leave one
[27:59] more thing for people and this gets
[28:01] into the bigger picture. Scott Bessent
[28:03] gave a speech at the New York Economic
[28:05] Club and I posted something on X
[28:09] on Friday
[28:11] that Matt Hogan had put out. I do think
[28:13] people should read the speech. I think
[28:14] they should also read the op-ed from
[28:16] Mohamed El-Erian on it. This is the fact
[28:19] guys and we talk about it here.
[28:22] The administration is very focused on
[28:24] changing the way the US does business
[28:27] around the globe.
[28:29] And part of that change is making sure
[28:32] that they don't lose the control of the
[28:34] financial system or be the leader of the
[28:36] financial guardrails.
[28:38] He specifically said that digital
[28:40] assets, tokenization, stablecoins are
[28:44] all part of the administration's focus
[28:47] on this new economic order.
[28:49] You have to understand what that means.
[28:52] They are focused on crypto being part of
[28:54] the guardrails, and this coincides with
[28:57] AI. So, the reason I named it AI macro
[28:59] nexus, the reason I do my YouTube and
[29:01] it's about 90% a combination of the
[29:04] impact AI is having on the macro world
[29:06] we exist in today, and 10% on crypto and
[29:09] what that means. We're now at the point
[29:12] where the reason I'm doing YouTube where
[29:13] it's going to be about 70% starting with
[29:16] crypto and then showing the difference
[29:18] of the macro world and AI connected to
[29:20] it is because we are about to enter the
[29:22] AI agentic commerce side, which means
[29:25] the velocity of money is going to
[29:26] increase. The dormant assets that remain
[29:29] in the form of housing, real estate
[29:31] around the globe are going to be
[29:33] liquefied and turned into active cash
[29:36] over the next 5 years through
[29:37] tokenization. These are all major
[29:39] positives, and that's the reason why I
[29:41] think you're at the beginning of a
[29:42] multi-year bull market in Bitcoin. And
[29:45] whether it starts in October, whether it
[29:47] starts in December, or whether it starts
[29:49] right now, I think we're at the bottom
[29:52] end of the range for the next year.
[29:54] >> I recently saw a statistic that showed
[29:56] uh stablecoins are spiking on the
[29:58] weekends in terms of volume, which makes
[30:00] sense the banks are closed, right? So,
[30:02] people want to uh access it. We also saw
[30:05] the open stablecoin a kind of consortium
[30:08] that got announced. I think they have
[30:09] like MasterCard and Visa and Stripe and
[30:12] a bunch of people all participating in
[30:14] this. And it just feels like some of
[30:17] this is, "Hey, how do I get into the
[30:18] stablecoin game?" But a lot of it is,
[30:21] "The agents need money." And what money
[30:24] are they going to use? How are they
[30:25] going to use this? What whatever. Um do
[30:28] you see anything on that front, or are
[30:30] you hearing anything from fund managers
[30:32] that suggest they're now starting to
[30:35] move some of their funds into whether
[30:37] stablecoins or Bitcoin, or is this more
[30:40] so you think we're pretty early in this
[30:41] reversal and you're you're certain to
[30:43] buy now because, you know, you think
[30:45] that's what's people are going to be
[30:46] doing 3 months from now.
[30:49] >> So price leaves nar- leads narrative.
[30:51] The first thing that always happens in a
[30:52] bottom is you start getting short
[30:54] covering because people are frustrated
[30:55] that this has gone on. I think Bitcoin
[30:57] has been a funding side of the AI trade.
[31:00] I think um it started when software got
[31:03] bludgeoned and all of the software names
[31:06] got hit hard
[31:07] uh and Bitcoin was part of it. I would
[31:09] say anything built on code, you couldn't
[31:11] get away from the correlation. So if you
[31:13] wanted to hedge your semiconductors,
[31:16] which now had high beta, you needed to
[31:18] find high beta shorts. And high beta
[31:19] shorts ended up being salesforce.com,
[31:22] Adobe Adobe, Workday, and Bitcoin. And I
[31:25] think all of those things became part of
[31:27] the momentum trade. So it's not a
[31:29] coincidence that as momentum has gone
[31:31] down sharply that Bitcoin has a bid. So
[31:34] for people who want to be bearish, like,
[31:35] "Well, this is just short covering." All
[31:38] rallies that last for a year start with
[31:40] short covering. Every single one because
[31:41] it means
[31:43] the narrative that was existing before,
[31:45] the AI infrastructure trade, is no
[31:47] longer as strong as it's going to be.
[31:49] And it's going to take a while for that
[31:50] re-leveraging to happen. As we go
[31:52] through earnings, like I said, I think
[31:53] you're going to have disappointments
[31:55] before
[31:56] every single name was working. I don't
[31:58] think that's going to be the case
[31:59] anymore in the AI trade. I also don't
[32:00] think you can get five-baggers and
[32:02] six-baggers over the course of the next
[32:03] year. I think a lot of these things are
[32:05] priced well. Dylan Patel talked about it
[32:07] today and he said you're going to see
[32:08] some surprises on the CPU side. And the
[32:10] surprise is going to be that we've kind
[32:11] of gone a little too far. On the optic
[32:13] side, we've gone a little too far.
[32:15] You're going to He talked about this. I
[32:17] think what all that means is that that
[32:18] place is priced in a way where
[32:21] if you can get 40% over the next year,
[32:24] that's great. I think Bitcoin can do
[32:26] more than 40% based on the fact that
[32:28] just to get back to all-time highs,
[32:30] you're talking about a double. And
[32:32] getting back to all-time highs to me
[32:34] we'll be left with the narrative that I
[32:35] just talked about. So the first stage is
[32:37] short covering, the second stage is for
[32:39] the momentum to shift once we get above
[32:41] the 200-day moving average, which I've
[32:43] said that is the critical line. We are
[32:45] still way below it right now. We still I
[32:47] think it's around 76, 77,000. So, we
[32:50] still have quite a ways. We have 20%
[32:52] before we get up there. So, this will be
[32:54] a short covering rally. We'll see how it
[32:56] goes. And like I talked about and for
[32:58] people who who heard this before and
[32:59] still reach out to me, are we still
[33:02] watching Doge? Yes, I'm still watching
[33:04] Doge for the energy side of crypto, but
[33:07] I do think there will be uh surprising
[33:09] news on what you said over the course of
[33:12] the year where when people start to
[33:14] realize that the 40-name index that I've
[33:17] created
[33:19] which goes through eight separate
[33:21] verticals or sectors is a direct overlay
[33:24] with Bitcoin that Bitcoin just
[33:25] represents the ecosystem and I think the
[33:27] ecosystem is going to benefit from the
[33:29] stable from the agentic side and that's
[33:31] going to flow to Bitcoin as well.
[33:33] >> Michael Saylor the first Bitcoin sale
[33:36] was like a dip in the you know, dip your
[33:38] toe in the water.
[33:39] The second one was like a cannonball in
[33:41] the deep end.
[33:42] The first time it sold off, the second
[33:44] time it sold off a little bit, but it
[33:46] almost seemed like the market was like,
[33:47] "Yeah, whatever." And they really didn't
[33:48] care. So, you know, maybe he should get
[33:50] some kudos for the way that he did this
[33:53] um and it's minimal impact on the market
[33:55] with the big sale. But do you take or
[33:57] read into that sale or or the response
[34:00] from the market?
[34:02] >> Uh a year from now this stuff is going
[34:04] to be noise. Um
[34:06] it I'm actually a little embarrassed to
[34:08] be associated with something where we
[34:10] have to even talk about this ridiculous
[34:12] thing. I I just the Michael Saylor thing
[34:15] to me um he's even become like people
[34:17] are blaming him for why Bitcoin's not
[34:19] going higher.
[34:20] >> [gasps]
[34:21] >> Bitcoin as we talked about um
[34:24] in every asset you have corrections. I
[34:27] don't get into four-year cycles. I don't
[34:28] get into any of this stuff, but
[34:31] the reality is what he has done and what
[34:34] he has been able to do. I'm more focused
[34:36] on the quantum side of crypto and the
[34:38] solutions that are coming out and some
[34:40] of the themes about should we just
[34:42] basically deal with these tokens that
[34:45] are dormant and should we
[34:48] The man owns an enormous amount of
[34:50] supply of an asset that I believe the
[34:53] world is going to acknowledge is a real
[34:55] asset for collateral. And as we get into
[34:57] tokenization and we start getting into
[34:59] the dormant asset side, I think people
[35:01] will start to understand the value of
[35:03] collateral again. Like they'll start to
[35:04] understand what he has always talked
[35:06] about. So, I never bought into this
[35:08] thing. I do agree that the most
[35:09] important thing is that it's actually
[35:12] higher than when he made his biggest
[35:13] sale. That would be a sign that if you
[35:16] would have said to people, "Hey, what
[35:17] would happen if a week after he makes
[35:19] his biggest sale,
[35:21] the unthinkable that he said he would
[35:22] never do publicly,
[35:24] and even though this is for a different
[35:26] reason, so it doesn't count, but he did
[35:28] do it, what would happen?" Everyone
[35:30] would have said the thing would be down
[35:31] big. So, the fact that it's higher, I
[35:33] think is a Apple type news item where it
[35:37] happened and once you don't sell off
[35:39] after something like that, it actually
[35:41] is more of a positive than a negative.
[35:43] So, I don't think this will ever be
[35:44] talked about again in a big way.
[35:46] >> Today's episode is brought to you by
[35:48] Uphold. Are you someone who's tired of
[35:50] juggling multiple apps just to trade,
[35:52] earn, stake, or spend your crypto? Then
[35:54] listen up because Uphold does all of
[35:56] that in one single, unified platform.
[35:58] You can access 300 plus crypto and fiat
[36:00] currencies with an interface that works
[36:02] for you, whether you're a beginner or an
[36:04] expert. [music]
[36:05] Uphold also features any-to-any swaps
[36:07] where you can swap crypto to fiat, fiat
[36:10] to metals, and tokens to tokens directly
[36:12] in the app. If you prefer self-custody,
[36:15] Uphold's vault gives you multi-sig
[36:16] security, key recovery, [music] direct
[36:18] trading access, and peace of mind
[36:20] without giving up convenience. Uphold is
[36:22] also 100% reserved, meaning no
[36:24] fractional practices, and proof of
[36:26] reserves are updated every 30 seconds,
[36:28] so you can verify your assets anytime.
[36:31] For US users, you can even earn yield on
[36:33] dollars with the USD interest account.
[36:35] No fees, no minimums, and funds are
[36:37] insured up to $2.5 million
[36:39] through the atomic cash sweep program.
[36:42] If you want one app for your entire
[36:44] digital asset life, check out Uphold
[36:46] today.
[36:47] >> [music]
[36:47] >> Go to uphold.com to learn more.
[36:50] uphold.com, go check them out today.
[36:53] Today's episode is brought to you by
[36:54] [music] Mogul. $4 trillion, that's
[36:57] how much real estate Deloitte says moves
[36:59] onto the blockchain over the next 10
[37:00] years. That's good news for the rest of
[37:01] us, because this year the VIX, Wall
[37:04] Street's fear gauge,
[37:05] >> [music]
[37:05] >> it ripped past 30. The S&P went red, and
[37:07] stocks reprice every time a politician
[37:10] tweets. Real estate doesn't do that.
[37:12] Tokenization is the key to giving
[37:13] everyone access to real estate. Here's
[37:15] what most people don't [music] realize.
[37:16] You can invest in tokenized residential
[37:18] real estate right now, thanks to Mogul.
[37:20] Mogul's founders come from Goldman
[37:22] Sachs, where they deployed billions into
[37:23] real estate for institutional clients.
[37:25] Now they handpick the properties and
[37:27] handle the management for you.
[37:28] >> [music]
[37:28] >> Historically, an 18% average annual IRR
[37:32] is 3x better than their competitors. You
[37:34] get targeted yields over 8%, monthly
[37:36] rent payouts, tax benefits from
[37:38] depreciation, and zero 3:00 a.m. tenant
[37:41] calls.
[37:41] >> [music]
[37:41] >> Over 100 properties have already
[37:43] launched on Mogul. More than 30,000
[37:45] people have already signed up, too. Do
[37:46] you want in plus get a special [music]
[37:48] offer? Head over to mogul.club/pomp.
[37:52] Again, that is mogul.club/pomp.
[37:55] You can see important disclosures in the
[37:57] description,
[37:58] >> [music]
[37:58] >> and head over to mogul.club/pomp.
[38:02] >> When [snorts] you look at the market
[38:04] right now, um we've talked a lot about
[38:05] AI, we've talked a lot about crypto. Are
[38:07] there are there any areas that you're
[38:09] really excited about, or that uh you're
[38:11] you're starting to deploy capital into
[38:13] that don't fall into AI and crypto, or
[38:15] do you think those are like the two
[38:16] games in town?
[38:17] >> So, I am looking into the insurance
[38:19] side. I'm looking into the regional bank
[38:21] side. I've talked a lot about Eli Lilly,
[38:23] which is going to remain
[38:26] Eli Lilly, I I don't think, even if it
[38:28] were to go up
[38:30] even if it were a three bagger from
[38:31] here, I wouldn't get out of it. I I have
[38:33] so much belief in what is happening on
[38:36] the health side and the compounding side
[38:39] of
[38:40] curing diseases and I want to be
[38:41] associated with
[38:43] the group that's already done sovereign
[38:45] AI. And I want to make sure people hear
[38:47] that. All this talk you're hearing from
[38:49] people, well, you should own your own
[38:50] models, you should go through this.
[38:51] Remember what Anthony and I talked about
[38:54] with Lilly Pod and the fact that on
[38:57] their campus, they have a thousand
[38:59] Blackwells. They have their own
[39:01] sovereign AI. They are using that with
[39:05] biotech companies. So, I'm going to keep
[39:07] saying that healthcare is a big thing.
[39:09] On my video last week, I did go through
[39:11] some of the insurance names, which are
[39:13] now starting to see. I think we're
[39:14] actually going into how important agents
[39:18] are for companies that were already
[39:20] using AI.
[39:22] And that means a lot of the banks are
[39:24] going to do well and you've seen a lot
[39:26] of the banks make new all-time highs.
[39:28] The regional banks have made new
[39:29] all-time highs. Regional banks fit into
[39:31] a very interesting thing, which is
[39:34] it's very hard to buy a bank and take
[39:36] them private because of all the
[39:38] regulations, which means takeovers
[39:41] within side the regional banks for
[39:42] mid-size banks is probably going to
[39:44] happen. Because as people put AI into
[39:48] their model, they can go buy companies
[39:49] and then infuse them with AI and get rid
[39:52] of head count. So, it becomes very
[39:53] accretive immediately. So, I think
[39:55] you're going to see a lot of
[39:56] consolidation within side the financials
[39:58] and I think the healthcare side and the
[40:00] insurance side has already been spending
[40:01] money on this and are already advanced
[40:03] on it. They're going to start seeing the
[40:04] benefits over the next year.
[40:06] >> What about outside of the public
[40:07] markets? So, are there anything in terms
[40:09] of commodities or anything else that
[40:11] you're starting to to get excited about?
[40:13] >> Well, if I'm right about um
[40:16] Bitcoin, then silver's going to do and
[40:18] gold's going to do well. It's not going
[40:20] to be a resurgence of the um of the
[40:22] debasement trade.
[40:25] But it is going to be a hey, these
[40:27] things were impacted more than we
[40:29] thought from AI sucking the dollars out
[40:31] of everything. And then secondly, the
[40:34] reality that we have hikes built in. So,
[40:39] the run at hot thing turned into hey, we
[40:41] don't have any need for this. I I wrote
[40:44] about this. I talked about it. We still
[40:47] have a budget deficit that's massive.
[40:50] And the reason that's important
[40:53] is I don't think the Fed can raise
[40:54] rates. So, one other angle is they raise
[40:56] rates in July.
[40:59] This could be a Michael Saylor type
[41:00] moment as well. Because if they raise
[41:02] rates, in my opinion, without any
[41:04] question, it'll be a one and done and
[41:05] it'll be a fret a Fed credibility hike.
[41:09] The long end's going to come down,
[41:10] meaning yields are going to come down.
[41:12] And that would be the reason that they'd
[41:13] be okay doing it. Every single macro
[41:15] person I talked to when we've had good
[41:17] conversations on this believes this. So
[41:18] do I. That if they were to tighten in a
[41:21] surprising manner in July as a
[41:23] credibility hike, the long end would
[41:25] come down.
[41:27] I think at that point people would treat
[41:29] it as a one and done. And for the end of
[41:31] the year, we have another 50% of a hike
[41:34] built in. And so, it could actually act
[41:36] as a positive if it's a dovish hike
[41:39] where he says, "We did this for the
[41:41] credibility side. We are committed to
[41:43] this, but we think now we're in a
[41:44] position where we can sit back and
[41:46] wait." If that were going to happen,
[41:47] which I think is going to happen in
[41:49] terms of if they did it, they're going
[41:51] to do that way.
[41:52] I think that would be positive as well.
[41:54] So, I actually think gold, silver, and
[41:57] Bitcoin that three grouping, the
[41:59] debasement trade, will start to research
[42:02] see a bottoming and a a rise as well.
[42:04] >> Now, sometimes people will listen to
[42:07] this and they should take away signal
[42:10] from the things we talk about.
[42:12] There's also times where they should
[42:13] take away signal for the things we don't
[42:14] talk about. And
[42:17] the Iran war
[42:19] has been on again, off again, on again,
[42:21] off again. Off for real. No, not
[42:23] anymore. And uh this past week, we got
[42:26] uh I think whiplash. Maybe it was on and
[42:29] off twice.
[42:31] I don't And this sounds crazy, but I
[42:32] don't think that the investment
[42:34] community really cares anymore about the
[42:36] Iran war's impact on their investments.
[42:40] Obviously, there may be difference in
[42:41] terms of the people who live in Iran or
[42:43] the the military components, the
[42:44] geopolitics, all that. But like from an
[42:46] investor standpoint, I have not heard
[42:48] someone talk about Iran impacting
[42:51] anything in their portfolio in the last
[42:53] 2 weeks. Maybe you have, but what's your
[42:55] take there?
[42:57] >> Um no, no one's really talked about it
[43:00] and I I don't think this is uh
[43:04] I don't think this is people just not
[43:06] being aware of it or not being
[43:09] uh I don't know, scared of the fact that
[43:12] it's started and might I think the
[43:14] reality is oil doesn't budge. And even
[43:16] though we had a rally, you know, we we
[43:18] jumped,
[43:21] if if oil and this is the beauty of
[43:23] markets,
[43:25] we heard the doom and gloomers say it
[43:26] should go to 200.
[43:29] Let's haircut that and say it should
[43:30] have gone to 150.
[43:32] The problem is it's still, despite
[43:35] everything, and it's not just the front
[43:38] part of the curve, it's the back end.
[43:40] There are no issues. The thing that I
[43:44] take from this is is plain and simple.
[43:46] There, we were supposed to have
[43:49] gas prices surging. Now again, crack
[43:51] spreads are wide, and the reason crack
[43:53] spreads are wide, and so people
[43:54] understand what crack spreads are, it's
[43:56] the differential between the gas price
[43:58] and the oil price, the conversion from
[44:00] oil into gas. That is still high. So,
[44:04] the ability of having gasoline is the
[44:07] issue that is still still there. But oil
[44:10] as a whole is not going higher and I
[44:12] think everyone would treat it as a
[44:13] short-term thing. They don't think this
[44:15] is going to not end, meaning this is a
[44:18] flare-up. I think everyone had built
[44:19] into their expectations this. So, I'll
[44:21] just say to everyone here, unless we see
[44:24] surprising movements in oil where it
[44:26] actually goes higher, it still acts in
[44:29] the market as something that has a
[44:31] positive negative and I hate say this,
[44:33] it acts like there's a glut. It
[44:35] literally acts like when this ends, the
[44:37] price is going to head back to 50. So,
[44:40] I'm not saying that is going to happen.
[44:41] I'm just saying the way the market looks
[44:43] and the way it acts, it just doesn't act
[44:45] like something that's going to go
[44:46] significantly higher. So, I think we're
[44:48] stuck between 65 and 80 and unless we
[44:52] get above 80 and the Iran situation is
[44:55] worsening where there's no ships going
[44:57] through Hormuz, I just don't think it
[44:58] matters for the markets.
[45:00] >> Last thing I want to talk about for a
[45:01] couple minutes is um I don't know if you
[45:03] saw the 1X robotic hand demo?
[45:07] >> No.
[45:08] >> So, um maybe the best way I can describe
[45:10] this is the hand has been the hardest
[45:13] part of robotics for a very long time,
[45:15] 30, 40 years people have been working on
[45:16] this. And most people what they
[45:17] essentially have done is they have used
[45:19] robotic components to build the hand.
[45:21] But if you think of your hand, it is all
[45:24] operated off of tendons. And so, 1X
[45:27] essentially has created a synthetic
[45:30] version of a tendon-based robotic hand.
[45:33] And in this demo, you know, they can do
[45:35] very kind of ambidextrous, you know,
[45:36] type of type things or very dexterous
[45:38] type things. But more importantly is it
[45:41] can screw on a lightbulb. It can pick a
[45:44] cherry without, you know, going and
[45:46] smashing it. Or it can do things that I
[45:48] think a lot of people when they saw saw
[45:50] this demo, the video got millions of
[45:51] views online, were like, "Whoa, hold on
[45:53] a second here. This is a significant
[45:56] leap forward."
[45:58] My takeaway from seeing this is that
[46:01] everything that you and I are excited
[46:03] about from an AI perspective, I would be
[46:05] shocked if that robotic hand was not
[46:09] somehow either ideated or developed
[46:12] using AI. And so we now are getting the
[46:15] like, you know, second boost to this
[46:17] robotic stuff where people are having
[46:19] this super intelligent software to help
[46:22] them make these advancements, and it
[46:23] wouldn't shock me at all if we start to
[46:25] see some pretty crazy stuff happen where
[46:27] these robots start to be as human-like
[46:30] as maybe we all feared that they could,
[46:31] you know, eventually get to.
[46:32] >> Mhm.
[46:33] So, I'm going to
[46:35] And this is This is a Finnish company,
[46:37] right? Or a Norwegian company?
[46:41] >> I think the fact I think that it is made
[46:43] in the United States, uh but I think the
[46:45] founders might be from there. I I can't
[46:47] remember. There's something that is like
[46:48] a Nordic relationship, uh but I don't
[46:50] think that the company is based I think
[46:52] it's actually a US company, but maybe
[46:54] the founders are from there.
[46:55] >> So, I'm going to I agree with what you
[46:57] said, and and so people hear this, um
[47:00] I'll start with the statement, and then
[47:01] I'm going to work a little backwards
[47:03] towards why this isn't, you know, why
[47:05] ROIC for for hyperscalers is not like
[47:08] fracking and oil um in the in in this
[47:11] context. Uh so, first of all,
[47:14] Anthony's point is really important,
[47:16] which I agree with. We're
[47:18] underestimating that as we get to 100
[47:21] We're above 140 IQ now. As we get above
[47:23] 160, and as we get above 200, we will be
[47:26] solving problems that we haven't been
[47:28] able to solve. We'll take angles that
[47:30] will allow us to produce these things,
[47:32] and then once the humanoids are
[47:35] available,
[47:37] then we also don't have to do the
[47:38] physical work of testing it out. They
[47:40] can build them for us. So, the hand is
[47:43] critical because it's probably the worst
[47:45] thing we have for actually doing
[47:47] intricate work that would be necessary
[47:50] for robots to be making robots or robots
[47:52] to be solving the problem. So, if the
[47:54] intelligence is there, we do need the
[47:56] hands to be able to have them go. And
[47:58] this was on a moonshots where
[48:00] Dave Blundon basically admitted that he
[48:03] underestimated how quickly we will get
[48:05] billions of robots because he he saw at
[48:09] Tesla
[48:10] Oh my gosh, the robots will be building
[48:12] the robots at some point. This is the
[48:14] compounding side that goes with Opus 4.5
[48:16] leads to 4.6 which leads to 4.7 and this
[48:19] whole thing of recursive
[48:20] self-improvement. Now, the reason I want
[48:22] to bring this back is to oil. I was
[48:25] asked by someone this week that isn't
[48:26] the ROIC risk for the hyperscalers the
[48:29] same as what happened with fracking.
[48:32] With the reason being in a a bunch of
[48:34] people on Vex have posted this. They've
[48:35] scared the hell out of some of my
[48:37] subscribers. They've reached out. Here's
[48:39] the reality, guys. This is not
[48:41] intelligence is not the same as oil.
[48:43] Like I don't I don't know how to like
[48:44] make that clear. Oil is oil.
[48:47] Intelligence is everything. Meaning
[48:50] the price of tokens going down, the
[48:53] usage goes higher.
[48:55] >> [snorts]
[48:55] >> Oil usage only grows in a linear fashion
[48:57] related to nominal GDP.
[49:00] It is a physical constraint to get more
[49:02] of it.
[49:03] The problem is for intelligence, it's
[49:06] the usage is growing exponentially. Oil
[49:09] never grows exponentially demand-wise.
[49:11] The only thing that slows down AI is the
[49:16] data centers. It's the gas turbines.
[49:18] It's the physical stuff which cannot be
[49:20] made quickly. So, to Anthony's point and
[49:22] for all of you that are not
[49:23] understanding how the compounding will
[49:25] go towards solving the energy problem
[49:27] and why I always have one eye on crude
[49:30] because eventually crude will go to
[49:32] zero. It will go to zero because we will
[49:34] solve energy at the fission, fusion,
[49:37] name anything you want. We will come up
[49:39] with so many solutions going forward.
[49:41] And so, when you listen to Dario and you
[49:43] listen to Sam Altman and you listen to
[49:45] Greg Brockman,
[49:46] they're all saying the same thing which
[49:48] is the intelligence in getting us to a
[49:50] point where math is being solved. Once
[49:52] math is being solved, then cancer can be
[49:54] solved. Then energy can be solved. These
[49:57] are all the science problems. These are
[49:59] all the physics problems of getting us
[50:01] to the point where everything can
[50:02] compound. So, in the same way we're
[50:04] talking about the compounding of
[50:05] intelligence and models for coding,
[50:07] which is language, we're about to get
[50:09] into it with consumer agents, and we're
[50:11] also about to get it into science. And
[50:13] that's where the robotics thing in the
[50:14] humanoids gets so important for your
[50:16] investment decision-making, and it also
[50:18] feeds in why the financial guardrails
[50:20] are necessary to deal with this type of
[50:21] speed.
[50:23] >> I think that's a great place for us to
[50:24] end it. What
[50:26] What are you going to cover in your
[50:27] YouTube video this Sunday?
[50:29] >> I'm I'm going to harp on the fact number
[50:32] one that the AI mid-cycle slowdown is
[50:34] over in the face that I think the
[50:35] sentiment has dropped off enough, the
[50:37] volatility's gone high enough that we
[50:39] are in the bottom end of the
[50:40] consolidation, and you should not should
[50:42] now be thinking that instead of is this
[50:45] going lower now, you should be thinking
[50:47] with the AI trade, which I couldn't say
[50:49] when Micron was at 1,300, but at 950,
[50:53] will it be higher 6 months from now?
[50:54] Yes, it will. Once we get through two
[50:56] earning cycle, it will be higher. So,
[50:59] the 6-month look forward from this,
[51:01] you're at the bottom end of the range,
[51:02] not the top end of the range. Second
[51:04] thing is how important consumer agents
[51:06] are to compute. For all of you scared
[51:08] about compute thinking there's a chance
[51:10] there's excess compute,
[51:12] if you guys want to hammer me in the
[51:14] future, I will say this again. We have
[51:16] insatiable demand for compute. We do not
[51:19] have enough supply, and the argument I
[51:21] just gave you, compute demand is
[51:23] exponential because intelligence usage
[51:25] goes up dramatically with every model
[51:27] advancement, but the physical supply is
[51:30] what constrains us. So, this is a
[51:32] supply-demand mismatch, and I'm going to
[51:34] go through why that means you should be
[51:35] invested there. And then thirdly, I will
[51:38] go in more detail on this Bessemer
[51:39] speech, and I will focus again on the
[51:41] importance of Bitcoin because I do
[51:43] believe that a year from now we will be
[51:45] talking about crypto in the same way
[51:47] that we're talking about memory and I'm
[51:48] going to
[51:49] finalize with what you said. You asked
[51:52] me the question, are fundamental people
[51:54] starting No, they are not focused on
[51:56] this right now. Number one, because
[51:57] they're still focused on AI both from a
[52:00] loss perspective and what's happened the
[52:01] last 4 weeks, but also from what should
[52:03] we be buying down here. Once they
[52:06] recognize and get through the Fed and
[52:08] they get through all of the changes,
[52:09] they are going to need to focus on
[52:11] crypto and the financial guardrails of
[52:12] the future. And 6 months from now, as we
[52:14] go into next year, crypto will be the
[52:17] memory for next year.
[52:19] >> Man, that got me hyped up, Journey. I'll
[52:21] be watching.
[52:22] >> [laughter]
[52:23] >> We'll talk again next week.
[52:25] >> All right, man.