Bitcoin’s Turn Is Next — The Easy Money AI Trade Is Over?

[0:00] Again, I'm more focused on the second
[0:02] half of the year and the fact that at
[0:05] some point here the agentic side becomes
[0:07] really important for for crypto. That's
[0:09] not going to be today's business. The AI
[0:11] trade's over. The ability of getting
[0:13] seven, eight times your money in that is
[0:15] over. I think the things people need to
[0:16] watch, and I've started to post more and
[0:18] X on this,
[0:19] >> What's going on, guys? Today we got a
[0:21] great conversation with Jordy Visser. In
[0:22] this conversation, we go deep into the
[0:24] rabbit hole of AI, what's going on with
[0:26] all the open-source models. Does this
[0:28] actually present a risk to all of the
[0:30] American large language model labs? And
[0:33] then we also get into what's going to
[0:34] happen to the portfolios, what companies
[0:35] are going to be winners, what companies
[0:36] are going to be losers. Are companies
[0:38] like Google betting their entire company
[0:40] on winning the AI race? What's going on
[0:42] with the war in Iran, interest rates,
[0:43] the Fed, and of course Bitcoin and
[0:45] crypto? All that much more in this
[0:46] week's conversation with Jordy Visser.
[0:49] All right, Jordy, all hell has broken
[0:51] loose. People are very worried about the
[0:52] Chinese open-source models. There are
[0:54] people who are advocating that the White
[0:56] House bans them. We now have an open
[0:58] letter coming from Nvidia and a bunch of
[0:59] other companies saying, "No, we should
[1:01] not ban them. We should encourage
[1:02] competition." What is your take as it
[1:05] the impact on investors' portfolio as
[1:08] the AI industry tries to figure out what
[1:09] to do with these open-source,
[1:10] open-weight models?
[1:13] >> It's just a uh
[1:15] uh
[1:17] I guess the best way to put it. So, this
[1:18] week in the video I'm I'm going to do
[1:20] over the weekend,
[1:21] uh I'm going to go through just the ups
[1:24] and downs that are going to come with
[1:25] AI. We we've talked about them a lot.
[1:28] Open source is just another one that
[1:30] causes conversation, and it's really
[1:33] related to how fast this is moving. So,
[1:37] open source was always going to be a
[1:38] part of the puzzle,
[1:40] uh always. Um and I say that with
[1:43] certainty because
[1:45] intelligence, again, will all be
[1:47] different. You'll be using different
[1:48] models. There'll be a race to lower
[1:51] boundary. I mean, Satya Nadella on an
[1:53] interview with DoorDash Patel talked
[1:54] about the fact that AI would be
[1:56] commoditized. So, you get to the point
[1:59] where you know this is going to happen
[2:01] to some degree where there's going to be
[2:02] open source. So, I just think this is
[2:05] another um spot where everyone, because
[2:07] of how fast it's moving, we migrate from
[2:11] Anthropic's ARR is going parabolic. What
[2:14] does that mean for every single company?
[2:17] Anthropic's going to be involved in
[2:18] every company in disrupting them.
[2:20] To, "Hey, we don't want Anthropic to
[2:23] have our data. We don't want OpenAI to
[2:25] have our data. We need to have open
[2:26] source. We need to have AI sovereignty."
[2:28] This has all been headed this direction.
[2:30] It's not like the Chinese open source
[2:32] models are a new thing. I just think
[2:35] we've reached a point where
[2:37] the Chinese open source models have
[2:39] caught up in a very, very big way. Uh
[2:42] and, you know, the the point of what I'm
[2:44] trying to go through is this needs to
[2:46] happen for adoption to lift off. And the
[2:49] cheaper the models get and the more that
[2:51] open source drives the prices lower,
[2:54] the faster adoption happens and the
[2:56] faster we get into a world that we
[2:58] discussed last week, which is based on
[3:00] routers. Where, just like in in a firm,
[3:03] you have different intelligence levels.
[3:04] Not every firm has all 150 IQ people,
[3:07] because there's not that many 150 IQ
[3:09] people. So, I think the front-tier
[3:10] models will stay the way they are um
[3:13] with very high cost. And what the
[3:15] Chinese models are going to run into,
[3:17] and this is the other part that isn't
[3:19] getting discussed, but this is the If
[3:21] you put everything together, the major
[3:23] story is
[3:25] even for Kimmi
[3:27] K3, which started this whole thing,
[3:29] you saw they have they ran out of
[3:31] compute
[3:32] immediately.
[3:34] And they're going to have to go IPO. So,
[3:36] Eric Schmidt said uh at the end of last
[3:40] year, and he's repeated it this year,
[3:43] "Energy's not going to stop AI.
[3:46] The biggest risk to stop AI is capital.
[3:48] And the capital is need for the compute.
[3:50] And now we're seeing all of this stuff
[3:52] lead to, "Hey, maybe it's not the
[3:55] commoditization means adoption is going
[3:56] to go faster. Maybe we don't have the
[3:58] capital to fund all the compute that we
[4:00] need." And the markets and bonds and
[4:02] everything are starting to react that
[4:03] way.
[4:04] >> What I find interesting is um
[4:07] at the same time that the large
[4:09] open-source or the large LLMs are
[4:12] complaining about the Chinese models
[4:14] using distillation and you know, kind of
[4:16] drafting off of their innovation,
[4:18] they're also settling lawsuits.
[4:20] Uh Anthropic just settled a $1.5 billion
[4:22] lawsuit where uh they basically had to
[4:25] pay out to a bunch of authors and and uh
[4:27] publishers for having trained on their
[4:29] information. And so, it does feel like
[4:33] uh in the AI world, everyone is doing
[4:35] the same thing just in different
[4:36] flavors. But then also, Bill Gurley from
[4:38] Benchmark, he's been tweeting uh the
[4:40] screenshot of I think his name is uh Jim
[4:43] Far- uh Farley, who's the Ford CEO,
[4:46] talking about that Ford has been buying
[4:48] Chinese EVs and then basically taking
[4:51] them apart, figuring out how they built
[4:52] them, and then trying to improve their
[4:54] own EV. And so, you know, hit the point
[4:56] he's trying to make is like, "Dude, this
[4:58] is just how business works." Cuz
[4:59] everyone tries to take apart their
[5:00] competitor's thing and then rebuild
[5:03] theirs to be better, faster, cheaper,
[5:05] more efficient, more effective,
[5:07] whatever.
[5:08] And so, is this all just like a big
[5:09] nothingburger?
[5:11] >> No, this is something much more complex
[5:13] than that and much more important and
[5:14] something that I'll continue to say.
[5:17] Behind every great idea,
[5:20] well, let's start at the higher end.
[5:21] Behind every company that is um runs
[5:25] into competition,
[5:26] is some sort of intelligence that drove
[5:28] it there. Intelligence is behind the
[5:30] ideas, intelligence is behind uh
[5:32] figuring out how to market them, how to
[5:34] get it out there.
[5:36] What has changed is time.
[5:38] There are no moats anymore because the
[5:41] one thing that prevents a moat from
[5:44] happening is the speed of competition.
[5:46] AI is arguably the most complex
[5:48] innovation we've ever seen. It's
[5:50] certainly the most important.
[5:53] The Chinese have been able to stay neck
[5:55] and neck with the US regardless of how
[5:56] it's done. I don't care whether it's
[5:58] distillation, I don't care whether it's
[5:59] algorithmic efficiencies, they don't
[6:01] have the compute and they've been able
[6:03] to do it. xAI wasn't even a company
[6:07] and Grok is right with it. So, I I just
[6:11] think as you go through this, you have
[6:13] to start to realize that if intelligence
[6:15] is abundant and commoditized and free,
[6:19] how will there ever be an idea that has
[6:21] scarcity? How will there ever be an
[6:23] idea, and this is one of the reasons why
[6:25] number one, you and I started doing this
[6:27] and the reason the name of my company or
[6:29] the the research I do is AI macro nexus.
[6:32] We are at the point where the linear
[6:34] meets the exponential and in this
[6:37] exponential world, what we're talking
[6:40] about right now is the most important
[6:42] thing which is to go from intelligence
[6:44] and idea, a thought, and now have it at
[6:46] the snap of your finger means there is
[6:48] no way to run a business and not have
[6:50] competition come up extremely quickly.
[6:52] So, what Anthropic and OpenAI are going
[6:54] through
[6:56] is something that no one thought could
[6:58] happen 3 months ago, which is okay,
[7:00] there's now a duopoly. These guys have
[7:03] separated themselves, they're going to
[7:04] reach AGI and this is going to take off
[7:06] and what has happened is we're now
[7:08] having a conversation that open source
[7:10] is neck and neck with closed. What does
[7:12] that mean? And I think that just means
[7:13] that intelligence and this whole moat
[7:15] theme 5 years from now, I don't think
[7:17] people have thought through what it
[7:18] means with AGI.
[7:20] >> Now, there's an interview that
[7:23] the let's see, the DeepMind CEO did. And
[7:26] in this interview, there was a couple of
[7:28] very interesting things that he talked
[7:30] about, but I want to read to you one of
[7:32] the takeaways. There's a a guy on
[7:35] Twitter, maybe we can pull it up to show
[7:37] this, but he says that DeepMind was
[7:39] never created to maximize profits, IPO
[7:42] or dominate the market. Instead, its
[7:44] sole mission is to achieve AGI or
[7:46] artificial general intelligence, which
[7:48] he believes will be one of humanity's
[7:49] most important technological
[7:50] breakthroughs. Everything else, consumer
[7:52] products, enterprise APIs, revenue and
[7:55] market share, is simply a byproduct of
[7:57] pursuing that mission. Now, this entire
[8:00] interview and the analysis of the
[8:02] interview basically is this idea that
[8:04] you almost have a missionary in
[8:05] DeepMind, which is saying, "Look, we
[8:07] actually don't want to make as much
[8:08] money as possible. We simply want have
[8:10] the technical breakthrough to get to
[8:11] AGI. Whoever controls AGI, that's going
[8:13] to be a significant, you know, advantage
[8:15] in the future."
[8:16] And he's almost counter-positioning
[8:18] DeepMind to OpenAI, Anthropic, etc., who
[8:22] are doing everything from creating, you
[8:23] know, hardware to doing enterprise, you
[8:25] know, services, etc.
[8:28] Is that the like that difference of
[8:31] positioning or that kind of root North
[8:33] Star going to ultimately negatively
[8:37] impact the large language models? Like
[8:39] if you have somebody who doesn't care
[8:40] about making money,
[8:41] doesn't that drive significant financial
[8:44] impairment for the companies that are
[8:45] trying to make money?
[8:48] >> I mean, I've written papers about the
[8:51] end of capitalism as we know it, and
[8:52] that's what you're talking about. So,
[8:54] let me just let me just give people a
[8:57] little bit of details cuz I did a video
[8:59] on this this week for subscribers on
[9:01] what China did this week cuz China
[9:02] stepped into their stock market. Um
[9:05] I mentioned that Eric Schmidt said the
[9:07] only thing that can slow the AI race or
[9:10] end it is capital. You know, it's more
[9:13] important than energy. And over time,
[9:16] that's become more evident. Now, he
[9:17] specifically said that the Chinese are
[9:20] at a huge disadvantage because they
[9:22] don't have access to the global capital
[9:24] markets the way the US does. US has the
[9:26] deepest capital markets. We've talked
[9:28] about Google raising money, OpenAI
[9:30] raising money, SpaceX raising money,
[9:32] Anthropic, all the bonds being sold,
[9:34] everything in there. So, DeepMind has a
[9:37] very unique situation. Everyone knows
[9:39] Deep Seek because of last January.
[9:41] They still may not remember or know that
[9:45] it started out of a hedge fund.
[9:48] Well, that quant fund was down 16% last
[9:51] week. Yes, that's last week.
[9:54] They did an IPO or they raised capital I
[9:56] and
[9:58] the
[9:59] only one buyer of the capital that they
[10:02] issued got voting rights. And that was
[10:05] the Chinese government. So, the Chinese
[10:06] government is part of Deep Seek.
[10:10] Last week, Xi Jinping gave an or gave a
[10:13] speech at in Shanghai
[10:16] that is basically the same thing as what
[10:17] the Deep Seek founder said. So, let's
[10:21] just get to the point that this is not
[10:23] about Deep Seek. This is not Sam Altman
[10:25] versus the founder of Deep Seek. This is
[10:28] the difference in philosophy of what
[10:30] China is trying to do versus what the US
[10:32] is. China has awfully talked has also
[10:35] talked about possibly not allowing their
[10:37] best models to be out in the world. So,
[10:40] this is all strategic. There is no doubt
[10:42] in my mind based on the fact that the
[10:44] Chinese government has told insurance
[10:46] companies and asset allocators to be
[10:49] buying more money in stock because
[10:52] if people put their money into bank
[10:53] accounts and housing, that doesn't help
[10:55] the capital markets. They actually need
[10:57] the money in the stock market and they
[10:58] need the stock market to be going
[11:00] higher. Now, that takes a lot of cash.
[11:02] It was said that $9 is what is what was
[11:06] put in this week, which happened to be
[11:08] the lows. They have a an innovation
[11:10] index there called the Star 50,
[11:13] which had its biggest daily performance
[11:16] since the lows in the China stock market
[11:18] in 2024. So, I think reading a little
[11:22] bit below this, this is still a race
[11:24] between China and the US. This is
[11:26] definitely them trying to make sure that
[11:27] they realize they need capital because
[11:30] eventually if they don't get more
[11:31] capital for compute, they will run into
[11:34] problems because the computer is not
[11:35] just for the intelligence, it's to allow
[11:37] people to use it, which is why Alibaba's
[11:40] raising money and Kimi K3 is raising
[11:42] money. They're all raising money, and
[11:43] the reason they're raising money is
[11:44] because this is on the same path as the
[11:46] US. I happen to do I do believe though
[11:49] that the end game is they're looking
[11:51] forward and so has Microsoft in a very
[11:53] similar way believing that these models
[11:56] are going to be commoditized anyway, so
[11:58] we don't want to be involved in building
[11:59] the models and trying to go through it.
[12:01] So, I do think there's different
[12:02] approaches here. In the end, I think
[12:04] we're headed towards abundance
[12:06] eventually because intelligence is
[12:08] basically the key to abundance.
[12:10] >> It feels like these companies are
[12:13] willing to bet the farm on being a
[12:16] leader in the AI sector. There's a chart
[12:18] that has been going viral online.
[12:20] I mean, cuz I tweeted it, but
[12:22] the chart basically shows Google's free
[12:24] cash flow, and they just put in the
[12:27] first negative free cash flow quarter.
[12:29] To be clear, they have a lot of cash, so
[12:32] they can do this for you know, maybe a
[12:34] year, year and a half and not run out of
[12:37] money, but
[12:39] people look at that chart and they say,
[12:40] "Wait a second. This is obviously a very
[12:43] intentional strategy where they believe
[12:45] that there is some massive ROI." And
[12:47] that brings the question of, "You know,
[12:48] will the ROI come or not?" Is this like
[12:51] if you do not make that investment, you
[12:53] pretty much will be left behind? Is it
[12:55] almost like an existential everyone's
[12:57] game theory is you have to basically bet
[12:59] the farm to try to be a leader in this?
[13:25] >> in the last year has gone from a hundred
[13:27] billion to 514 billion.
[13:30] So, they have the revenue sitting there.
[13:32] They just can't monetize them yet. And
[13:34] the reason they can't monetize them yet
[13:36] is because the data centers aren't built
[13:38] yet. So, I go through all of these
[13:41] details and just kind of say that
[13:42] whenever you see these charts of free
[13:44] cash flow, you said something important.
[13:46] Number one,
[13:48] they have massive cash still.
[13:50] They just raised 80 billion in equity.
[13:53] They're issuing bonds. So, they have
[13:55] access to the capital market. So, they
[13:57] can continue to go out. Yes, rates are
[14:00] moving higher, but this doesn't change
[14:02] the fact that
[14:04] something you said is important.
[14:06] These companies to keep the business
[14:08] they have,
[14:09] they have to get AI to be a
[14:13] revenue-producing part of their
[14:14] business.
[14:16] And it's not easy to do other than the
[14:18] fact that they have a massive business
[14:20] in advertising right now. Well, I think
[14:23] they're
[14:24] believing, which I think most people
[14:26] are, that at some point advertising will
[14:28] not be a great business. But they still
[14:30] have YouTube. And YouTube is competing
[14:33] with Netflix, but they have AI and all
[14:36] of these things get wrapped into the
[14:38] same thing that when you tear apart
[14:40] Google's business to Waymo and to all of
[14:42] these different components, all of those
[14:44] businesses to some degree are built on
[14:46] code and with intelligence, the only
[14:49] thing that you really have is the
[14:50] distribution, you have the brand name,
[14:52] you have all these things. But as we've
[14:53] seen with other businesses, it can come
[14:56] under attack very, very quickly if
[14:58] there's something cheaper or something
[14:59] better or new that's happened. And I
[15:01] think that they're going to have to
[15:02] spend just like all the other companies.
[15:04] So, anyone who's worried about the CapEx
[15:06] side,
[15:07] uh
[15:08] Google has basically a problem right
[15:11] now. If they didn't spend the CapEx,
[15:13] they're not going to get the 400 or the
[15:15] 514 billion in revenues. 514 billion, I
[15:19] mean, we're talking they have contracts
[15:21] bigger than the than, you know, 90% of
[15:23] the S&P 500 companies in terms of their
[15:26] market cap. So, you're you're you're at
[15:28] a point where they don't really have a
[15:30] choice, and so I think the CapEx is
[15:31] going to continue, uh, and we'll
[15:33] probably have some companies that do run
[15:35] into problem. The bond market is making
[15:37] it more difficult for certain companies
[15:39] to raise money like OpenAI and and
[15:41] Anthropic because these are the
[15:43] hyperscalers with the best balance
[15:44] sheets. It will be harder for other
[15:46] companies that don't have the same
[15:47] balance sheet.
[15:48] >> Today's episode is brought to you by
[15:50] Arch Public. Arch Public has just
[15:51] expanded its agentic trading platform
[15:53] beyond crypto. So, pay attention. This
[15:55] is a big one. Now, they are automating
[15:57] strategies across stocks, commodities,
[15:59] and ETFs, and I think that this is going
[16:01] to be huge. You can now automatically
[16:02] take profits when one market hits new
[16:04] all-time highs and rotate that capital
[16:06] into other markets showing more
[16:08] opportunity. Whether you're rotating
[16:09] capital into AI stocks, gold, if you're
[16:11] investing in the S&P 500, or you're
[16:13] accumulating Bitcoin, Arch Public brings
[16:15] real discipline and automation to your
[16:17] investment strategy. Additionally,
[16:19] they've launched a powerful new tax loss
[16:20] harvesting tool. With crypto being so
[16:22] volatile and its exemption from the wash
[16:24] sale rule, Arch Public can offset gains
[16:26] with losses without compromising your
[16:28] long-term positions. It's exactly what
[16:31] every serious investor does.
[16:32] Institutional-grade automation that
[16:34] works across every major asset class.
[16:36] There's no more emotional trading, no
[16:37] more missing tax opportunities, just
[16:39] smarter, hands-free execution of your
[16:41] preferred strategies. Go to
[16:43] archpublic.com
[16:44] right now. Connect with their team, set
[16:46] up a time, bring your account if you'd
[16:48] like, then you can learn what automated
[16:50] trading can do for you. archpublic.com
[16:53] >> I saw a chart that shows Anthropic's
[16:55] revenue growth is slowing. So, it's
[16:57] still growing, but, you know, obviously,
[16:59] the growth rate's very hard to keep, uh,
[17:01] keep that pace. Um,
[17:03] and I think that the Anthropic revenue
[17:06] growth slowing is happening at the same
[17:09] time that you get Meta, you get Grok
[17:12] coming into the market and specifically
[17:14] positioning themselves with a lower
[17:15] cost, you know, per token or or more
[17:18] efficient uh type model. At the same
[17:20] time, we see that Stripe is thinking
[17:22] about buying open router. Ramp just
[17:24] announced a model router. Uh us with
[17:27] Silvio, we've created our own model
[17:29] router internally to try to keep costs
[17:30] under control and and uh increase
[17:32] performance. It just feels like there's
[17:34] been this shift from like, who cares how
[17:36] much this cost, to now all of a sudden
[17:38] it's all about efficiency and total cost
[17:40] per task or or kind of per outcome.
[17:43] Would Could we see Anthropic's growth
[17:45] rate plateau? And some of this is the
[17:48] DeepSeek and the Chinese open source
[17:49] model, some is the model router, etc.
[17:51] But like, could OpenAI and Anthropic be
[17:53] in be in a little bit of trouble here?
[17:55] >> No, but I I think it was foolish and and
[17:58] I know I said this last week, but I I
[18:00] think it was foolish for people to
[18:01] extrapolate, you know, Anthropic's
[18:04] growth rate, that it wouldn't reach a
[18:07] point. I think plateau is taking it too
[18:09] far. I think the growth rate slowing is
[18:12] happening.
[18:13] Uh I still believe that one of the best
[18:16] uses and and and the highest margins
[18:18] available for this is the ability of
[18:20] giving this for solving some of the
[18:22] world's biggest problems, and I think
[18:23] that's what
[18:25] >> [gasps and sighs]
[18:25] >> DeepSeek, uh not DeepSeek, um Google
[18:28] DeepMind and also uh Anthropic and even
[18:31] OpenAI to some degree, they're they're
[18:33] focused a lot on research and on the
[18:35] bigger picture for things, where they
[18:37] probably can get more money for curing
[18:39] diseases and focusing on that part. Uh
[18:42] but, we're also at the stage where it
[18:45] would be foolish to not take into
[18:46] account Jevons' paradox and just realize
[18:48] that as compute gets cheaper, it means
[18:51] more
[18:52] companies go from token maxing to token
[18:54] efficiency. You're describing token
[18:56] efficiency. So, we went through token
[18:58] token maxing, which I would just say to
[19:00] people, having someone that's used AI
[19:02] all the time,
[19:04] you're trying to figure out how to
[19:05] incorporate it into your business, how
[19:06] to get it into your workflows. You're
[19:08] going to max out on it. The router part
[19:10] is going to make it easier once you've
[19:13] you know, now that you're you're through
[19:14] the first phase. So, Anthropic's going
[19:16] to grow because we're at the early stage
[19:18] of
[19:20] enterprise agents. We still have
[19:22] consumer agents to come. That'll be less
[19:24] focused for them and probably more
[19:25] focused for
[19:27] for OpenAI if they can get the hardware
[19:29] side out. But, I will say that for
[19:31] Anthropic, I wouldn't bet against the
[19:32] fact that they're going to continue to
[19:33] grow the revenues. I just think this
[19:35] concept of growing at the pace they were
[19:38] is over. So.
[19:40] >> Now, we've also seen I think a lot of
[19:42] talk what I'll consider in a generalized
[19:44] way Wall Street is very focused on
[19:47] memory as a huge shortage, compute,
[19:50] chips, you know, all of these things
[19:52] that we've been talking about I don't
[19:53] know 6 months, 12 months now.
[19:55] And most of that conversation is what
[19:58] are the publicly available companies
[19:59] that I can go and I can invest in. So,
[20:00] whether it's SK or Micron or whatever,
[20:03] that's really been most of the the
[20:04] public market conversation. At the same
[20:06] time, venture capitalists, they don't
[20:07] invest in the public market. They've
[20:09] been funding you know, brand new
[20:10] competitors to the market. And so, we've
[20:12] seen companies like Etched and a couple
[20:14] of others show up and say, "Hey, I have
[20:17] kind of a better mouse trap. I've got
[20:18] the ability to whether it's compute,
[20:20] memory, etc. do this in a better way."
[20:24] What I found interesting is Etched as
[20:26] one example just announced a brand new
[20:28] fundraising round and they had the usual
[20:30] players. I think Sequoia and Andreessen
[20:32] Horowitz, etc. But, SK actually invested
[20:35] as well. And so, you're starting to see
[20:37] this kind of cross-pollination where the
[20:40] major companies, whether it's the
[20:41] Nvidias, the SKs, etc. They also are
[20:44] helping to fund competitors or potential
[20:47] partners or potential acquisition
[20:49] targets of the future where it's almost
[20:51] like they're outsourcing their
[20:52] innovation and they understand to solve
[20:54] this bottleneck, maybe they can do it
[20:56] internally, but there's also going to
[20:58] have to be more competition on the
[20:59] playing field.
[21:02] >> Yeah, you've got a big problem with
[21:03] competition catching up. It's one thing
[21:06] to come up with an idea.
[21:08] Again, I'm going to say it over and over
[21:10] again. We got caught in this world where
[21:13] software you could compete. Physical
[21:16] hardware is different. So, let's assume
[21:18] a new company comes out and and says
[21:20] we've developed a better chip that will
[21:23] do XYZ. Well, that's great. Now, all you
[21:25] got to do is get the scale. Who's going
[21:27] to make it at the size that you need to
[21:29] supply the demand that's way too big.
[21:32] So, you know, I spent a lot of time
[21:34] creating the logo for my
[21:36] for my company. And when I was working
[21:39] with with Fable 5 on this and kind of
[21:42] going through it, I just kept saying
[21:45] I want you to think Nexus. I want you to
[21:47] think the world is changing. And so, the
[21:49] logo that came up was a V where the left
[21:51] side is a linear line and the right side
[21:54] is a parabolic line or an exponential
[21:57] line. And that's what fits into this. I
[21:59] think every time we run into a situation
[22:01] right now, it's a hardware problem. And
[22:04] until we get out of that, the
[22:05] hyperscalers had a bad week this week.
[22:07] There was nothing wrong with Google's
[22:08] numbers, but I think at this point the
[22:10] spenders keep coming under attack even
[22:13] if they get a rally for short amount of
[22:14] time. No one can get past the point, and
[22:17] I think it's justifiable because you and
[22:20] I talked about this concept of 3 years
[22:22] from now we don't know what terminal
[22:23] value is. One thing I know is if you're
[22:25] spending lots of money,
[22:28] you're taking risk. So, I went through
[22:31] Fable 5 and basically said, "Okay, let's
[22:34] go through Google's earnings reports.
[22:36] You're an expert forensic accountant
[22:38] with experience as a CFO at variety of
[22:41] big tech enterprise companies. Based on
[22:44] Google spending money, the backlog,
[22:47] every single detail, what's the
[22:49] probability of this working out for
[22:51] them?"
[22:52] The answer wasn't 100%.
[22:54] Um the answer was legitimately 75%. And
[22:58] when I went through the the 25 to 30%
[23:02] that outcome that wasn't good. These are
[23:04] all things people know. So, I bring this
[23:06] up just because the problem for software
[23:08] companies
[23:10] the competition's coming and the more
[23:11] that intelligence moves forward in the
[23:13] capabilities, the worse it gets for all
[23:15] these companies. If you're spending lots
[23:17] of money, you're running into the
[23:18] physical hardware side. If Google had
[23:21] all of these data centers already built,
[23:22] they'd have $514 billion
[23:25] from Google Cloud and they would have a
[23:26] dominant market share and all that would
[23:28] be great. Instead, they're having to
[23:30] rent stuff from other people. So,
[23:32] whenever you hear these stories people
[23:33] and oh, we're going to catch up, you
[23:35] just have to ask yourself a question and
[23:37] put it into the best model and say, is
[23:39] this a physical problem and will they
[23:40] have the scale to bring this out quickly
[23:43] or is this an algorithmic coding thing
[23:44] that can easily be rolled out to
[23:47] everything with the existing chips? And
[23:49] the reality is almost every single time,
[23:51] it's a hardware issue and the hardware
[23:53] issue is is is not solved easy. And I
[23:56] just I I've been using an analogy now
[23:58] with with um groups that have asked me
[24:01] the question about why memory is such a
[24:04] big problem.
[24:06] Do you remember the movie First Dates
[24:08] with Adam Sandler?
[24:10] >> Yes.
[24:11] >> Where she couldn't remember anything at
[24:13] Okay.
[24:13] >> Amnesia.
[24:15] >> And
[24:16] daily amnesia where every day she forgot
[24:18] what happened the prior day. Well,
[24:20] that's basically like a context window.
[24:22] And the problem is to get agents to
[24:24] remember everything, it's an enormous
[24:27] amount of memory and the reason we
[24:29] haven't solved this is cuz this is a far
[24:31] more complex problem than just going
[24:33] into a context window and then going
[24:35] back theoretically to that day and
[24:37] having the same conversation knowing the
[24:39] context is in there. This is an
[24:41] incredibly huge amount of compute
[24:44] huge amount of memory that needs to come
[24:47] through and the problem is we know how
[24:49] to make the memory, we just can't make
[24:51] it fast enough and that's why prices are
[24:53] going higher and that's why GPU
[24:54] availability continues to go down. We
[24:56] still have a compute shortage and it's
[24:58] going to remain for a long time.
[25:00] >> I'll take it one step further. There's
[25:01] absolutely the memory and the compute
[25:03] shortage etc. I also
[25:05] and not because I knew to do this but
[25:08] the engineering team at Salvia they
[25:10] actually built an entirely proprietary
[25:13] agentic file system specifically focused
[25:15] on spinning up a specific file system
[25:18] for each individual user that helps to
[25:20] increase memory, reduce latency, you
[25:23] know, solve all this stuff. And so what
[25:25] it feels like is there's almost like a
[25:26] sandwich. You have the hardware or like
[25:29] the bottlenecks of memory, compute, etc.
[25:31] But then also you can't just use like
[25:33] traditional databases. The whole
[25:35] agentic, you know, memory or agentic
[25:37] file systems are going to be rebuilt and
[25:40] you're going to start to get different
[25:41] architecture, you're going to start to
[25:44] you know, have to take a lot of
[25:46] different information and be able to put
[25:48] it into these file systems, be able to
[25:50] recall it in you know, very specific
[25:53] ways, have the agents be able to
[25:54] coordinate between each other. Like it
[25:56] is a software problem as much as it is a
[25:58] hardware.
[25:59] The nice thing about the software is if
[26:01] you throw enough people at it like you
[26:02] can figure it out. You can't do that
[26:04] with the hardware which I think is kind
[26:05] of your point is the physical world just
[26:08] has these limitations that it kind of
[26:10] doesn't matter how much money you have
[26:11] or how smart your people are. There's
[26:13] just no way to to solve that piece of
[26:14] the stack.
[26:16] >> Yeah and I I don't see how that solves
[26:17] until we get to AGI. I I I think we need
[26:20] solutions that are so far above the
[26:22] things that people are doing
[26:24] and as a really smart person who's in
[26:26] the tech side from the hedge fund world
[26:28] said to me
[26:29] the smartest people in the world are
[26:30] working at all of these big companies by
[26:32] now and they're trying to solve the same
[26:34] problem because if they can solve it
[26:35] then they their margins go up
[26:37] dramatically because then they don't
[26:38] have to go pay for like there's just no
[26:40] way to get through this problem so they
[26:42] can make the chips. So they're all
[26:43] getting into the chip business but the
[26:46] memory business and this is why memory
[26:48] is really AI equals memory is a paper
[26:51] that I wrote recently and at the end of
[26:53] the day you have to watch those stocks
[26:55] because until those stocks are bottoming
[26:56] and rally and we did see, you know, I
[26:58] mentioned that I bought Micron last
[26:59] week, it did see a rally this week but
[27:02] it's not going right back up to the
[27:03] highs. I don't expect the AI worries to
[27:06] end anytime soon because all the things
[27:07] we've talked about in here,
[27:10] they're all viewed in a negative lens
[27:12] because there's so much uncertainty and
[27:13] I think that's what AI does to
[27:15] everything. Um I highly recommend if
[27:17] people get the chance Emad Mostaque,
[27:19] have you had Emad before? So
[27:21] >> Um no, I've talked to him but I don't
[27:23] think he's ever been on the podcast.
[27:24] >> He's he's a good interview for you to
[27:26] have on. I mean Emad, the reason I like
[27:29] him so much is because he was in the
[27:31] hedge fund world, he understands
[27:32] economics and he wrote a book recently
[27:35] titled The Last Economy and it basically
[27:38] again is kind of the same as AI macro
[27:40] nexus. It's like it's the last economy,
[27:43] so the end of capitalism for me, the
[27:45] heading into a world of abundance. He
[27:47] and I are
[27:48] mostly on the same page with everything.
[27:51] Um I I I think this is a really
[27:53] important thing for people to start to
[27:55] put into their head now. I think 2 years
[27:57] ago when I talked about it I didn't tell
[27:59] people or suggest to people they should
[28:01] read this type of stuff, but we're only
[28:03] 3 years away from AGI and this is
[28:05] becoming a centralized number that
[28:07] everyone is coming up with and because
[28:10] of the open source catching up and
[28:12] because of humanoids and the rise that
[28:14] seems to be happening at the exact same
[28:16] time and you've got these bigger players
[28:18] getting involved in actuators and all of
[28:20] these things, I think it's getting
[28:22] closer and closer to this world that
[28:23] people just haven't thought too much
[28:26] about. It has huge implications for your
[28:28] portfolio but also I think it has huge
[28:30] implications for crypto and when you go
[28:32] back to that first dates analogy I gave
[28:34] where I used it, Sandy Call wrote a
[28:36] great paper. I don't know if you've
[28:38] you've had Sandy on. Sandy was at
[28:39] Citibank. Okay, she she's awesome. She
[28:42] was really one of the original people in
[28:43] digital assets that got me focused on
[28:45] this but this whole thing of the agentic
[28:47] AI economy that I've talked about, she
[28:50] wrote a paper this week that people
[28:52] should read as well, which I'll
[28:53] highlight on the video, but I think
[28:55] we're at that stage now where you have
[28:56] to start thinking how the world is going
[28:57] to look different by the time we get to
[28:59] 2030.
[29:00] >> I want to talk about the war in crypto,
[29:02] but before we do that, one other aspect
[29:04] of this like AI industry is Travis
[29:06] Kalanick, who was the founder of Uber.
[29:09] He has been building a company for the
[29:11] last eight or nine years in stealth.
[29:14] And he kind of came back. He did the
[29:15] Michael Jordan, you know, I'm back, and
[29:18] he put the world on notice. Now, the
[29:20] company I think is like a 15 or 20
[29:22] billion-dollar business. So he's been
[29:24] working very hard. And [snorts]
[29:26] my understanding of what he had been
[29:28] building previously was pretty much
[29:30] focused on cloud kitchens. That was kind
[29:32] of the US consumer business.
[29:35] And the only thing I knew was that the
[29:37] actual legal entity was called city
[29:39] storage
[29:40] systems, which is kind of a weird, you
[29:42] know, name. Well, turns out that he now
[29:44] is reporting that they have hundreds of
[29:47] facilities in 30-plus countries around
[29:50] the world. They've, you know, expanded
[29:51] this thing substantially when it comes
[29:53] to food. And my understanding of the
[29:55] business was basically it's ghost
[29:58] kitchen. It's, you know, a dark kitchen.
[30:00] They basically have the WeWork model but
[30:02] for kitchens, and that was it.
[30:04] I heard him talk with Ben Horowitz and
[30:07] Eric Tornberg from A16Z, who was his big
[30:10] backer in this round. And
[30:13] he opened my eyes to the merging of AI
[30:17] and the physical world in a way that no
[30:18] one else has. I'm going to tell you what
[30:20] he said, and I'd love for you to react
[30:21] to it.
[30:22] He basically talked about the idea that
[30:24] he has recognized the physical world
[30:27] operates the exact same way that
[30:29] computing and networking works. And so
[30:31] he talked about the kitchen is
[30:33] manufacturing. He talked about like
[30:36] the IP packets or TCP/IP and how all of
[30:41] the logistics and the movement of
[30:43] information and the movement of the food
[30:45] and all this stuff, right? So, he has
[30:47] this kind of like mental framework of if
[30:48] you take a computer network that you
[30:49] would learn in my computer science class
[30:51] and you overlay it onto the physical
[30:53] world, it's very similar. And so, he's
[30:55] basically been eating at this stack. And
[30:57] so, one of the things he talked about in
[30:58] the kitchen is
[31:00] um
[31:01] it started out as a real estate play.
[31:03] But then they realized, well, if we want
[31:05] more restaurants, we have to make sure
[31:07] the restaurants don't go out of
[31:08] business. So, how do we do that? Well,
[31:10] we got to bring them customers. Well,
[31:12] could we also help them on their cost?
[31:14] Let's use robotics. And he just kept
[31:16] eating into all of their problems and
[31:18] trying to make sure that he could get
[31:20] them more orders, keep them in business,
[31:22] drive this, you know, huge growth.
[31:24] I say all of that because the business
[31:26] that he unveiled is basically that same
[31:28] concept. He has it in food. He's now
[31:31] doing it in the mining industry. He's
[31:34] got an entire kind of like autonomous,
[31:36] you know, mining facility uh or uh
[31:37] vehicles, etc. And it sounds like his
[31:41] business is he essentially wants to
[31:43] digitize the physical world.
[31:45] And all he kept talking about was AI and
[31:49] robotics and automating this stuff. And
[31:52] you know, when you hear this, it it kind
[31:54] of feels like the Silicon Valley pie in
[31:55] the sky. Like, hey, this sounds, you
[31:57] know, awesome, whatever, it's cool.
[31:59] But when he talks about the concrete
[32:00] impact, for example, with the food, he
[32:03] would like to make it cheaper for you to
[32:06] order food than to go to the grocery
[32:08] store.
[32:09] And his point is, if I can make a better
[32:12] meal that is healthier and cheaper, of
[32:15] course you're going to go and do that.
[32:18] The challenge is how do you do it? How
[32:20] do you do the logistics network, etc.?
[32:21] And so,
[32:23] we spend so much time talking about
[32:24] software, but when I hear something like
[32:27] this,
[32:28] I start to say, hey, this may actually
[32:29] be the thing that impacts so many
[32:32] people's everyday lives way before some
[32:35] chatbot or, you know, some other kind of
[32:37] software AI product that they may or may
[32:39] not download, right?
[32:42] >> Yeah, so there I let's start at the
[32:44] beginning. Um
[32:46] So
[32:48] I'm very big on on talking about the
[32:50] importance of being a systems thinker
[32:52] right now. What you just described is
[32:55] Travis kind of looking at a problem and
[32:57] then breaking down each of the vertical
[32:59] pieces that roll up to one and trying to
[33:01] solve each piece with the end goal being
[33:03] very simple, which is
[33:05] how do I reduce the overall cost for
[33:06] people, get them to be ability to to get
[33:09] food, and let's just use the restaurant
[33:10] example, which is cheaper than what they
[33:12] can do in their daily lives where they
[33:14] don't have to go buy it, they don't
[33:15] We're going to save them the time on it.
[33:17] So right off the bat, that sounds a lot
[33:19] like how Amazon has disrupted the the
[33:21] retail world. Um let's take out the
[33:24] driving, let's take out the gas, let's
[33:25] take out It'll just get to you and it
[33:27] and with drones, which is the next layer
[33:29] for Amazon,
[33:31] think about what Amazon has done for
[33:33] each parts of the system of getting that
[33:35] there. So it starts as a book place, but
[33:37] then it eventually gets into everything.
[33:40] And now it's unbelievable how many
[33:42] things show up. And anyone who's lived
[33:43] in an apartment complex in Manhattan,
[33:46] and you just go to the package room, and
[33:48] you see how many packages are coming in
[33:49] every day, and how it's non-stop
[33:51] multiple times a day, you realize how
[33:53] this could work. So that's the first
[33:54] thing is the systems thinking. The
[33:56] second thing is this is the beginning of
[34:00] what I call the end of capitalism.
[34:02] When you start breaking down every
[34:03] component of cost to try and get it
[34:06] down,
[34:08] especially with something like food,
[34:09] which is something we all need to have,
[34:12] you're now getting to the point of
[34:13] people looking at abundance. How can I
[34:15] use intelligence to replace the cost?
[34:17] But it's the merging of humanoids and
[34:21] robots, anything intelligence-wise with
[34:24] humans. So this is all based on the fact
[34:27] of how do I reduce the cost of making
[34:29] the product, make it I mean, again, if
[34:32] it's made and the easiest visual for
[34:34] people on this is to look at Chipotle
[34:36] cuz that's kind of what he's doing is
[34:37] saying, "Okay, how do I create this
[34:39] where it's like Chipotle?" And Chipotle
[34:41] did the first version of this. How do
[34:43] people get through this really quickly?
[34:44] They can see all the choices. They can
[34:46] pick the stuff as they kind of go
[34:47] through. Subway did the same thing. All
[34:49] of these different stores with this this
[34:51] concept. I just think that the merging
[34:54] of the
[34:56] intelligence
[34:58] the physical humanoids, which is the
[35:00] part you're bringing in. Use Amazon for
[35:02] just kind of a an early description of
[35:04] how that'll look like. But now you're
[35:06] getting into something related to food
[35:08] and less about something related to
[35:10] discretionary purchases and more, can
[35:12] you satisfy this for me and get it to me
[35:14] in a way that's very simple to do? I
[35:16] just think this is the world we're
[35:17] headed towards, but I think this is a
[35:18] merging of the physical world that uh
[35:22] let's break AI into two parts. Embodied
[35:24] AI, the intelligence part, the software
[35:26] part, and then merging that physical
[35:29] that into a physical world and then
[35:30] combining it with the way human beings
[35:32] go today
[35:34] with the end goal eventually being
[35:35] abundance. 15 years from now, 20 years
[35:38] from now, whatever the case is. We will
[35:39] hit that point in my mind and capitalism
[35:42] and the way that you invest in companies
[35:44] will be disrupted. So think about how
[35:45] many businesses that would not only
[35:46] disrupt but then how many
[35:49] businesses could be a startup to compete
[35:51] with other places like you mentioned
[35:53] mining and all of that. It starts to get
[35:55] very scary that any business has
[35:57] no moat now and can't grow anymore.
[36:00] >> Today's episode is brought to you by
[36:01] TikTok for business. If you run a
[36:02] company, you are probably already
[36:04] spending a bunch of money on ads,
[36:06] search, X, maybe meta. But there's a
[36:08] platform your competitors are sleeping
[36:10] on and the window to move first is
[36:12] closing. TikTok has over 200 million
[36:15] monthly active users in the US alone and
[36:17] 51% of that reach is unique, meaning
[36:19] that those users aren't reachable on any
[36:21] other major platform where you are
[36:23] already buying. Unlike follower-based
[36:26] platforms, TikTok's algorithm rewards
[36:27] strong creative, not audience size. That
[36:30] means a well-made video from a brand no
[36:31] one's heard of can reach millions of
[36:33] relevant users. For companies trying to
[36:35] acquire users cost-effectively, that's a
[36:37] fundamentally different model. TikTok
[36:40] for business gives you full-funnel tools
[36:42] built for performance. Whether you're
[36:43] looking to drive awareness, introduce
[36:45] your products, or convert new audiences,
[36:47] TikTok is becoming an important
[36:49] discovery engine for businesses and
[36:51] cryptocurrency alike. If you're a
[36:53] founder, a CMO, or growth lead, and you
[36:55] haven't run a TikTok campaign yet, now's
[36:58] the time to change that. Click on the
[37:00] link in the description or go to
[37:02] anthonypompliano.splashthat.com.
[37:05] Again, click on the link in the
[37:06] description and go check out TikTok for
[37:08] business today.
[37:10] >> Today's episode is brought to you by
[37:11] Uphold. Are you someone who's tired of
[37:13] juggling multiple apps just to trade,
[37:15] earn, stake, or spend your crypto? Then
[37:17] listen up, because Uphold does all of
[37:19] that in one single unified platform. You
[37:22] can access 300-plus crypto and fiat
[37:24] currencies with an interface that works
[37:26] for you, whether you're a beginner or an
[37:28] expert. [music] Uphold also features
[37:30] any-to-any swaps, where you can swap
[37:32] crypto to fiat, fiat to metals, and
[37:35] tokens to tokens directly in the app. If
[37:37] you prefer self-custody, Uphold's vault
[37:39] gives you multi-sig security, key
[37:41] recovery, direct trading access, and
[37:43] peace of mind without giving up
[37:45] convenience. Uphold is also 100%
[37:47] reserved, meaning no fractional
[37:49] practices, and proof of reserves are
[37:50] updated every 30 seconds, so you can
[37:53] verify your assets anytime. For US
[37:55] users, you can even earn yield on
[37:57] dollars with a USD interest account. No
[37:59] fees, no minimums, and funds are insured
[38:01] up to $2.5 million
[38:03] through the Atomic Cash Sweep Program.
[38:06] If you want one app for your entire
[38:08] digital asset life, check out Uphold
[38:10] today. Go [music] to uphold.com to learn
[38:13] more. uphold.com.
[38:15] Go check them out today.
[38:16] >> It It's fascinating because on one hand
[38:18] you're like um
[38:20] what Uber has done uh in terms of uh the
[38:24] taxi industry or you know private cars
[38:26] whatever. On the other hand they
[38:27] basically turned every single individual
[38:29] driver into a small business owner. So
[38:31] there's kind of you know creative
[38:32] destruction and and then obviously
[38:34] there's the creation component.
[38:36] Restaurants very similar, right?
[38:39] You hear robotics and you just saying oh
[38:40] my god this is bad. But I think their
[38:43] argument would be well we're helping
[38:44] these restaurants not go out of
[38:45] business. We're actually keeping them in
[38:47] business and we're helping them grow
[38:49] orders and we're helping them be more
[38:50] efficient and we're driving down the
[38:52] cost. It's very like deflationary for
[38:54] the end consumer.
[38:56] And so it does feel like the robotics
[38:59] component. The other thing that he
[39:00] talked about that I thought was pretty
[39:01] interesting is he talked quite a bit
[39:03] about everyone has almost over rotated
[39:06] on humanoid robots. And his point is
[39:09] he was watching the Chinese
[39:12] kind of robotics like marathon. Remember
[39:15] when they had the humanoids running the
[39:16] marathon? And he said all he kept
[39:18] thinking about why don't we put wheels
[39:20] on them?
[39:21] Right? He was like like why are we
[39:23] trying to use the humanoid form factor
[39:25] for something where obviously putting
[39:26] wheels would be such a better use case.
[39:28] And so he is very into like specialized
[39:31] robotics over humanoids. He doesn't
[39:33] think humanoids are not going to be
[39:34] valuable. He's like there's a time and a
[39:36] place for them to be able to do that.
[39:37] But I think he actually looks at it as
[39:39] like humans are really bad at doing
[39:41] specialized work. We're very good at
[39:43] doing general purpose low volume work.
[39:45] But if you need specialized high volume
[39:47] work then you want a specialized robot.
[39:49] And I think you and I you know at least
[39:50] beat around this
[39:52] in the past when we talked about
[39:53] robotics of there may be this explosion.
[39:56] It just may not be what people think it
[39:57] is.
[39:58] >> Yeah, I agree. I again we're we're
[40:00] talking about something that most people
[40:01] haven't
[40:02] put into their minds and so it can go a
[40:05] bunch of ways. I'll just give people
[40:07] again two things to think about with
[40:09] regards to their portfolios and and and
[40:11] what is happening now. So we're going to
[40:14] get the Fed this week.
[40:16] And the Fed has to make a decision on on
[40:18] rates. Um last week it had gone down to
[40:21] about 15% chance of a rate hike in July
[40:26] after the CPI and PPI numbers came in.
[40:28] Well, oil goes higher. You see the long
[40:31] end yields moving higher. And all of a
[40:33] sudden, as we go into next week, we're
[40:35] at 35%. Now,
[40:37] Warsh has said he doesn't want to have
[40:38] public guide or forward guidance
[40:40] anymore. So,
[40:42] every single
[40:43] Fed day is going to be like this unless,
[40:46] you know, something is is structurally
[40:48] changing, which I'm not really sure it's
[40:50] going to be. Um the inflation side is
[40:53] not built in. So, I bring this up
[40:55] because he's talked about the fact that
[40:57] we have to change the way the Fed thinks
[40:59] about raising rates or doing monetary
[41:01] policy because it can't be based on the
[41:04] last month's CPI. It has to have a
[41:07] forward-looking side, the same way the
[41:09] stock market moves. Meaning, we discount
[41:11] a multiple like we're doing with Google.
[41:14] It's down to below 20 PE, and the reason
[41:16] is cuz we don't know if they're going to
[41:17] make it or not. And it's one of the
[41:18] biggest companies in the world. So, with
[41:21] robotics, you have to think about the
[41:22] fact that 3 years from now we could have
[41:24] an incredible deflationary pressure, and
[41:26] shouldn't the Fed be thinking about that
[41:28] rather than the other side. The second
[41:30] part is, and again, this gets into it,
[41:33] once we start talking about robotics in
[41:34] a real way,
[41:36] it's still early.
[41:38] Every business is up for risk now
[41:40] because a startup can come in
[41:43] and they don't need the human labor. So,
[41:45] right now to find electricians and
[41:47] plumbers, impossible. To be able all of
[41:50] these physical jobs, it's really, really
[41:52] hard. That's going to change in the
[41:56] future. Uh and I do want to end on on
[41:58] one point here cuz now everyone is
[42:00] swinging the other direction on the
[42:01] labor side, which to me is insane.
[42:05] AI is having an impact on the jobs
[42:07] market. Um it's not destroying jobs,
[42:10] which I didn't think it would, But, last
[42:12] month, we had a massive fall again in
[42:15] labor participation rate. So, the labor
[42:17] force participation rate has been going
[42:20] down
[42:21] since about 2000, and it's just going
[42:24] lower. Some of this is demographics, but
[42:26] it's demographics plus the household net
[42:28] worth in the country. And the household
[42:30] net worth in the country is being driven
[42:32] by people owning effectively profit
[42:34] margins. They're all long profit margins
[42:36] because they're long the stock market.
[42:38] And if the stock market's going higher,
[42:40] they're making money, and it allows them
[42:41] to retire earlier or not look for a job
[42:43] the way they would have. So, I bring
[42:45] that up because part of this thing when
[42:46] you connect everything,
[42:48] no one's losing their jobs overnight
[42:50] because this humanoid story is going to
[42:51] take time. But, what will happen is the
[42:54] pressure on all of these, the
[42:55] forward-looking side, the way that the
[42:57] market deals with things, the reason
[42:59] that the inflation swaps market hasn't
[43:01] budged, guys, as oil's gone back up to
[43:03] the highs. Um we haven't done that at
[43:05] all, and it's because we're starting to
[43:07] look through all of this and start
[43:09] looking out 2 to 3 years. That is why
[43:11] the multiple in the market is
[43:12] compressing, in my opinion, and that is
[43:14] why you have about 2 to 3 years to
[43:17] really start to make sure your portfolio
[43:19] is shaped more for the future of
[43:20] robotics, and less about the capitalist
[43:23] world that's ending.
[43:25] >> Let's talk about Bitcoin and uh and
[43:26] crypto. Um Bitcoin has had a little bit
[43:28] of strength uh this week, but I don't
[43:31] think that anyone is uh is kind of
[43:32] cheering. It's still down, you know,
[43:34] almost 50% from its all-time high. Uh do
[43:37] you have any different views or or
[43:39] anything that happened this week that
[43:40] really caught your attention?
[43:42] >> No, I think um I think it's held in
[43:44] well. I mean, as of when we did this, it
[43:46] was unchanged for the week. I think
[43:48] there's a tremendous now amount of hope
[43:50] that's being built into the Clarity Act
[43:52] being passed, even though the
[43:54] probabilities in Polymarket and Kalshi
[43:57] is still below 50%. Uh
[43:59] if it did pass, um it'll be interesting
[44:02] to see see the reaction uh not only day
[44:04] one, but but 2 weeks out.
[44:08] Again, I'm more focused on the second
[44:10] half of the year and the fact that at
[44:13] some point here the agentic side becomes
[44:15] really important for for crypto. That's
[44:17] not going to be today's business.
[44:19] Um I do think the AI,
[44:21] you know, what we talked about
[44:22] Anthropic,
[44:24] the AI trade's over. Like and let me re
[44:27] you know, as someone who basically has
[44:29] subscribers that pay attention to AI,
[44:32] it's not that AI is over, it's that the
[44:35] easy money of it is done. The ability of
[44:38] getting seven, eight times your money in
[44:40] that is over. Now you get into the grind
[44:42] of oh my gosh, 30% a year is horrible.
[44:44] No, 30% a year in the AI infrastructure
[44:46] trade's going to be great, which is why
[44:48] I got interested in Micron again. I'm
[44:49] still got a very small position in
[44:51] Micron cuz I still think there's a
[44:52] chance there could be another leg lower
[44:54] in this panic side. think it'll stay
[44:55] down there very long, so I want to be
[44:57] ready to put money in. Same thing I did
[44:59] with Bitcoin when it's been trading down
[45:01] here. I think it's going to go higher. I
[45:03] think the things people need to watch
[45:05] and I've started to post more on X on
[45:06] this and the subscribers who are
[45:09] interested, I will start a YouTube
[45:10] channel on it. I've been posting more
[45:13] and more the ecosystem index that I
[45:15] created of crypto,
[45:18] it took out the, you know, the mid-June
[45:20] highs. Bitcoin did not take out the mid
[45:22] mid-June highs. Ethereum's been
[45:24] outperforming Bitcoin. All these things
[45:26] are suggesting that the market is
[45:27] starting to look more towards the
[45:29] revenue side of the equation, which
[45:31] would be more towards Ethereum and less
[45:33] towards Bitcoin. And that's a positive
[45:36] thing in the long run I think we're much
[45:37] closer to this.
[45:39] Look again,
[45:41] Doge,
[45:42] which was something I never thought I'd
[45:44] talk about, but I'm very interested in
[45:47] getting retail back into crypto and
[45:48] until I see Doge break above, it's been
[45:50] below the 20-day moving average for the
[45:52] longest time in its history. Until we
[45:54] break above it and we've run into it
[45:56] multiple times over the course of the
[45:58] last 3 months. Until we break above it,
[46:00] I don't think retail is participating in
[46:02] this and I think this is people just
[46:04] trying to stick their nose in a bear
[46:05] market trend hoping that it'll go
[46:07] higher. I think it's coming soon, but
[46:08] clarity act needs to happen. If we don't
[46:10] get past on that, then we're going to go
[46:12] right back down towards the lows and
[46:14] we'll have to go through this bottoming
[46:15] process again.
[46:17] >> To me the the clarity act is um
[46:20] so performative. And I know that some
[46:23] people, you know, will will not like
[46:25] hearing that, but um I said to a friend
[46:27] earlier uh
[46:29] Bitcoin doesn't need the clarity act cuz
[46:31] Bitcoin already has clarity.
[46:33] Right? It is kind of everything else and
[46:35] it's it really is like the industry and
[46:38] the incumbents fighting it out trying to
[46:40] figure out, you know, who gets to do
[46:42] what, who has a regulatory advantage,
[46:44] who has to get certain licensing. Like
[46:45] like it's all the the edge case stuff,
[46:48] but the assets themselves are going to
[46:50] continue to do what they uh what what
[46:52] they do.
[46:53] The reason I bring it up is I think that
[46:55] there was a lot of excitement around
[46:57] tokenization. If you look at, you know,
[46:59] Securitize went public. Uh obviously
[47:01] stable coins have been very popular. Um
[47:04] but you don't really see the retail
[47:06] demand for a lot of that stuff. It's not
[47:09] like, you know, the retail investors on
[47:12] X maybe just like a sentiment check. I
[47:14] don't see them talking about buying
[47:15] Circle stock or Securitize stock or some
[47:17] of these things that are more
[47:19] traditional finance infrastructure that
[47:22] is coming from the crypto industry. And
[47:24] so just like you maybe look at Dogecoin
[47:27] as an interest, I do wonder at some
[47:29] point will there be that like froth of
[47:31] retail interest in all of these what is
[47:34] really kind of more boring aspects of
[47:36] crypto. But if you go and you talk to
[47:38] the traditional folks, they're like
[47:40] tokenization's going to be a huge thing.
[47:42] Stable coins are going to be a huge
[47:43] thing, right? It's just like a mismatch
[47:45] between what Wall Street is excited
[47:47] about and what like the crypto retail
[47:49] folks are excited about.
[47:51] >> Well, let
[47:52] let's make this um
[47:54] all AI related now in terms of crypto. I
[47:58] honestly, you know, whenever you listen
[47:59] to someone say, "How are we going to
[48:01] regulate AI?"
[48:02] You can't regulate intelligence. Like,
[48:04] this is ridiculous. So, the problem is
[48:08] a year ago, it was very hard for me to
[48:10] say the network effects of crypto and
[48:13] not have someone roll their eyes and be
[48:14] like, "I don't even know what that
[48:15] means."
[48:16] I will tell you right now, every single
[48:18] person on the hedge fund world
[48:20] understands what it means now. No
[48:22] question about it.
[48:23] It's not working yet, and I've seen this
[48:25] too many times. Investing and trading is
[48:28] about making money. Nobody wants to buy
[48:30] an asset that has no enthusiasm, no
[48:33] interest. Trust me, I remember Mike Cron
[48:34] the first quarter of last year. So,
[48:38] Japan
[48:39] put through major, major regulation last
[48:42] week on an acceptance of of crypto and
[48:44] specifically of Bitcoin.
[48:47] South Korea, same exact thing.
[48:49] Russia this week. The reason I I I think
[48:52] this is important is
[48:54] there is a race now to the acceptance
[48:57] that, okay, Bitcoin's not a global
[48:59] currency. It's not replacing the dollar.
[49:01] What is replacing Swift
[49:05] is stable coins. Like, you can't get
[49:06] away from it now. Everyone's adopting
[49:08] it. So, this is where I think your
[49:09] comment about this is performative. I
[49:12] actually completely agree. I just want
[49:14] the Clarity Act out of the way. Um if it
[49:17] gets done, I think it's a big positive.
[49:19] I mean,
[49:20] when you have the CEO of Goldman Sachs
[49:22] publicly come out and say, "Let's just
[49:24] get the Clarity Act through now."
[49:27] You're starting to reach a completely
[49:28] different time where they get
[49:30] tokenization is happening. There's no
[49:32] way to get away from it. I presented to
[49:33] the New York Stock Exchange on it. It's
[49:35] happening. There's no way to get around
[49:36] stable coin adoption. There's no way to
[49:38] get around the fact that the move that
[49:40] Stripe tried to do with PayPal. It's a
[49:43] big merging of the two worlds. PayPal's
[49:46] an old, boring company the way you
[49:48] describe kind of, you know, the way
[49:50] maybe people perceive at this point some
[49:52] of the the public companies that are out
[49:54] there. But why did they do that? Because
[49:56] Stripe has the merchants and PayPal has
[49:59] the consumers. Like you have to
[50:00] understand we're getting to a world
[50:01] where we are merging these two worlds of
[50:03] the past and the future and it's
[50:06] happening at a very very fast pace. So I
[50:08] think crypto will be looked at as
[50:10] connected to AI every single day and the
[50:13] US has had the ability to sanction
[50:15] people. It's had Swift. Everyone's going
[50:17] to want their hands on the new financial
[50:19] guardrails. This is a very very
[50:20] important thing from a government
[50:22] perspective and also from an investor
[50:24] perspective. People have to be over
[50:26] crypto. It's way past the time of you
[50:28] not doing your homework and worrying
[50:29] about whether clarity gets through. I
[50:31] agree with you. I think this is
[50:32] something very important for people to
[50:34] be spending their time on because of AI.
[50:36] >> Where where where do you want to send
[50:38] people? Send people to to the YouTube
[50:40] tomorrow?
[50:41] >> Yeah, they should go to the YouTube for
[50:43] Sunday morning at 8:30
[50:45] and they should go see what's going on
[50:47] and if they want to go visit the website
[50:49] and read some of the papers. Oh, I have
[50:50] a new prompt. You'll like this one too.
[50:53] I'll give you a little description. So I
[50:56] I I spend a lot of time
[50:58] thinking about how to use AI in this
[51:01] mixture of experts type approach. So
[51:03] I've done it with my investing side
[51:05] meaning I'll use all five of the models
[51:07] and each five of the models I'll give
[51:09] them a different expertise and have them
[51:12] do deep research on the exact same
[51:13] topic. So if I say I want you to do deep
[51:16] research on on the companies that will
[51:17] benefit from Vera Rubin. I want you
[51:19] you're an expert in financial
[51:21] accounting.
[51:22] I could do another one that's an expert
[51:24] in hardware and semiconductors. I could
[51:26] do another one that's an expert in
[51:28] thinking about the way things look in
[51:29] the future on the power side and you
[51:32] have them all come out and then they
[51:33] come back with their answers. Well, I I
[51:35] started doing the same thing. I used to
[51:37] go to conferences
[51:39] and at a conference I might see five
[51:41] people speak on the same topic. But
[51:43] because they were spread out over you
[51:46] know
[51:46] uh maybe a 3-day period, it's very hard
[51:50] to get the nuances and remember that to
[51:52] connect dots on all of these. The brain
[51:54] does not process information like that.
[51:56] It processes things one at a time, and
[51:58] that's why like you can't listen to
[52:00] music and read a book and be able to sit
[52:03] there and consume both of them at the
[52:04] same time. Your brain's picking one. So,
[52:07] what I did is I created a prompt called
[52:09] the Mosaic prompt, which I upload three
[52:11] separate transcripts, which I did this
[52:13] week with Moonshots. I did it with the
[52:15] All-In Pod, and I did it with the Steve
[52:17] Hoe interview on Forward Guidance, and I
[52:19] created a mosaic of the answers. It came
[52:22] out really cool. I'm putting that up on
[52:24] the subscriber site, and people can see
[52:25] it at AI uh that 22V research.com. But,
[52:29] it's a it's a new way to go through it,
[52:31] and for you fantasy football fans that
[52:33] are ready get to do your uh your draft,
[52:35] this is actually a great way to take
[52:37] five separate experts, upload their
[52:39] podcast, and find the places where they
[52:42] connect on the same thing. It's kind of
[52:44] a crowdsourcing feature, so you're
[52:45] getting the mosaic in a bigger picture.
[52:47] But, what it really allows the AI to do
[52:49] is to be the the conductor of the
[52:52] experts and do more future forecasting
[52:55] like a super forecaster would.
[52:57] >> Well, you know, if you start solving
[52:58] people's uh fantasy football problems,
[53:01] you'll be very popular, Journey.
[53:03] >> I As someone who doesn't have the time
[53:05] to do fantasy football um cuz I don't
[53:07] half-ass anything,
[53:09] uh I decided a lot about 5 years ago I
[53:11] couldn't keep up with the people that
[53:12] were doing it cuz it was way too
[53:14] time-consuming. Uh so, now my job is to
[53:16] help uh my son and other people who are
[53:18] still obsessed with this stuff.
[53:20] >> I love it. I love it. Um all right.
[53:23] Well, I appreciate the uh the time
[53:24] today. As always, I had learned
[53:25] something, and I hope other people did
[53:27] as well. We'll do it again next week.
[53:28] >> See you, bud.