Enjambres de agentes, el virus de OpenAI y la era de la volatilidad parabólica — Jordi Visser (1 febrero 2026)

[0:00] All right, still on the road now in Los
[0:03] Angeles.
[0:04] Uh, lot going on again. Let's try to get
[0:07] through it. Uh, software collapse,
[0:11] weekly recap. Um, silver and gold
[0:14] collapse on Friday. Uh, Fed holds, job
[0:17] picture, PMI set to rise.
[0:21] We'll go through the industrial metals,
[0:23] Brazil. uh this open a virus continues
[0:27] to spread and I think it remains a story
[0:30] uh for the rest of the year. Swarm risk
[0:33] from AI agents, the corny meta corning
[0:35] metadal and Bitcoin remains asleep. So
[0:39] I'm just going to keep running with the
[0:40] theme. Uh you want to be long scarcity,
[0:42] short abundance for the entire year. Get
[0:44] used to these parabolic moves higher and
[0:47] parabolic moves lower. Uh daily
[0:50] volatility and things that are scarce is
[0:52] going to remain on the high side. You've
[0:53] seen it with a lot of assets during the
[0:55] month and also on Friday. Uh in the case
[0:59] of software, Gavin Baker put out another
[1:02] thing and again I I posted too. The
[1:05] beatings continue. I use Claude in
[1:08] Excel. Blew me away. I'm going to show
[1:11] some of that later on in this, but um
[1:13] it's really hard to not get negative on
[1:16] the software side. And then Microsoft
[1:18] comes out and the chart looks horrible.
[1:21] Um, you know, at least you still have an
[1:23] upward sloping 200 day moving average at
[1:25] this point, but you've broken below, by
[1:28] the way, on the charts for the week. I'm
[1:29] doing this on Friday,
[1:32] uh, before the market is closed. Uh, so
[1:35] there's a couple things. The prices
[1:36] aren't exactly going to be in where they
[1:38] are, but give or take 15 minutes, unless
[1:40] something big happen uh, in that 15
[1:43] minutes. Just know that these are not
[1:44] the actual closes for the month, but you
[1:47] can see the charts negative on
[1:48] Microsoft. Um Dan Ies who's been talking
[1:52] up Microsoft even kind of has his new
[1:54] price target a little bit lower but I
[1:56] think the story and this is really
[1:57] important um is really here they had
[2:01] massive RPOS
[2:03] so this is kind of like the story for
[2:06] Oracle but here's the thing that I said
[2:08] my issue for these companies the
[2:11] hyperscalers that are spending the money
[2:14] is that if they're depending on the
[2:16] cloud revenue they're depending on the
[2:18] data centers is being built and the
[2:20] bottleneck's not happening. We have
[2:22] massive bottlenecks that are going to
[2:24] continue and this story is going to make
[2:26] it difficult for them to be able to get
[2:27] the capacity at the same pace that
[2:29] they're spending the money. So that has
[2:32] implications for buybacks. It has
[2:33] implications for a lot of them and I
[2:35] just think at this point the uncertainty
[2:37] over when these companies will actually
[2:39] get the revenues just continues to grow.
[2:40] So you have to be wary about the
[2:42] spending. And here's the chart of
[2:44] Oracle, which is the orange line and led
[2:46] the way lower. But Microsoft is now
[2:48] following it. It may not be following it
[2:50] for the debt reasons, but it's
[2:51] definitely following it for the revenues
[2:53] down the line reasons. Uh this is SAP
[2:56] out of Germany. This problem in software
[2:58] is not just a US thing. This is a global
[3:01] thing. software to me, as I'll keep
[3:03] saying, if you're spending your time
[3:05] trying to pick the bottom in these, you
[3:07] don't make money picking the bottom in
[3:09] these charts unless you have a long
[3:11] enough time to wait for this to bottom.
[3:13] Uh, and I'm not really sure that the
[3:14] multiple compression story is the only
[3:16] thing. As someone again who uses this
[3:18] seven days a week, all day long, there
[3:21] is not a single hour of the day where
[3:22] I'm not using AI unless I'm in a meeting
[3:24] or in some kind of exercise class.
[3:28] The ability of building things and
[3:30] competing hurts at least the growth
[3:31] rate. You can sit there and debate me on
[3:33] all the things that I've highlighted in
[3:35] the last few weeks, but the charts are
[3:37] speaking as SAP plunges on cloud backlog
[3:40] myths. Again, this whole backlog of
[3:43] cloud, I just think it's an issue. Now,
[3:45] Palanteer has been brought into it.
[3:46] Palanteer definitely benefits from the
[3:48] data analytics, but when you have
[3:50] multiple compression happening across
[3:51] the software space,
[3:53] they're not immune to this. even if the
[3:55] stock ends up going higher in in in the
[3:57] future, you could go back, you know,
[4:00] significantly and still be up a lot
[4:02] since uh uh since the beginning of 25.
[4:05] So, just be very wary in these names.
[4:07] And when you get this kind of story, the
[4:09] only reason I'm putting this up there is
[4:12] when there's good news, and these stocks
[4:15] fall a lot, not a little, they fall a
[4:17] lot in charts like that. Uh that's a
[4:20] problem. So, good news being bad
[4:22] reaction, not worth looking. I I think a
[4:24] lot of people this week uh tried to fade
[4:26] Chimath who's been continuing to pound
[4:28] the table on on the SAS side and again
[4:31] legacy SAS has a short window to convert
[4:34] RPO to revenue. Completely agree. I
[4:36] don't understand why this is such a
[4:38] complicated thing other than people's
[4:40] egos thinking that they can pick the
[4:42] bottom of something. There are better
[4:44] games to play here. But I think the
[4:45] issue is going to be can you be safely
[4:48] being long Western Digital at this point
[4:49] when it falls violently on on earnings?
[4:52] Can you be long silver? When it falls
[4:54] violently, unfortunately for these
[4:56] parabolic charts on the upside,
[4:59] they're just going to have extreme
[5:00] volatility. They're not going to be able
[5:01] to be a software name that you can sit
[5:03] in without having to worry about it. I
[5:05] wanted to bring this up because just
[5:06] remember this was February 2nd of last
[5:09] year. There's a new kind of coding I
[5:11] call vibe coding. Elon Musk giving his,
[5:14] you know, I think software is dead.
[5:16] Remember, he is really uh been vocal
[5:19] about this. So again, everyone wants to
[5:20] fade. Elon Musk as well. Here's the
[5:23] facts. Cursor, most people had not heard
[5:27] of it. February 7th of last year, just
[5:29] five days after
[5:32] Andre Carpathy put out the vibe coding
[5:34] thing. So five days, Curser went from
[5:36] one to 100 million ARR in 12 months.
[5:39] Fastest SAS to achieve this. This week,
[5:42] just canceled cursor. End of an error.
[5:45] kind of crazy, but I would have paid
[5:46] like $2,000 a month for it only a year
[5:48] ago.
[5:50] That's what I showed you in terms of the
[5:53] time frame. So, I just asked Rock, "What
[5:56] does this mean? Does this highlight the
[5:57] speed of competition for things built on
[5:59] code where AI can replicate and prove
[6:01] the wave of cursor cancellations you're
[6:03] seeing, especially the pattern around
[6:04] January 26th? very clearly highlights
[6:06] just how brutally fast competition moves
[6:08] into AI powered coding tools and how
[6:11] little mo rapper products have once the
[6:14] underlying models become commoditized
[6:16] and powerful enough again not all
[6:18] software is going to fit into the same
[6:20] component
[6:21] but at this point everyone who has
[6:23] software attached to them is going
[6:24] through a repricing and I think it is
[6:26] not only justified I'm going to keep
[6:28] saying this you want to be long scarcity
[6:30] you want to be short abundance anything
[6:32] built on code you have to prove that you
[6:34] can adapt your business. You have to
[6:36] prove that you're not going to get eaten
[6:37] by these little companies, not just in
[6:39] the US, globally. And I think that's
[6:41] going to happen. Um, another Claude bot
[6:44] was out. This got it. Um, feels like the
[6:46] chat GPT moment. I can't tell you how
[6:48] many Chat GPT moments there have been
[6:51] with Claude over the past three months.
[6:53] They just keep coming out time and time.
[6:55] And this is why you're seeing this.
[6:56] Anthropic was founded in 2021.
[6:59] Estimated revenues 1 billion, 10
[7:01] billion, 18 billion, 55 billion, 148.
[7:04] Now again they are dominating. This is
[7:07] absolutely competition for software
[7:09] companies. So if you don't believe it
[7:12] is, it is. This is all enterprise coding
[7:15] going on. They're seeing the revenue. So
[7:17] they're seeing companies use it. And
[7:19] maybe they're all startups. Maybe
[7:20] they're not all big enterprise. Maybe
[7:22] they're just getting on token usage. But
[7:25] I'm very wary. All right, we're
[7:26] finishing off uh the month. Everything's
[7:30] pretty much up across the board. So
[7:31] despite all the volatility that's been
[7:33] going on, um,
[7:37] still finishing up with the first five
[7:39] days of the year up and the month up. I
[7:40] think it's like 92% of the years when
[7:43] that happens, the market's up. Even
[7:44] though that's numerology, I'm just
[7:46] bringing it up because it went on for
[7:47] the week. S&P barely budged. So again,
[7:50] despite all of the news, we've had three
[7:52] weeks in a row of not moving. Barely
[7:54] down, barely down, barely up. Um, in
[7:57] terms of the S&P 1500 level two, GIX,
[8:02] best performing group year to date,
[8:04] energy. I'm going to keep pounding on
[8:06] energy. Number two, capital goods.
[8:08] Number three, materials. If this doesn't
[8:09] scream the physical upgrade that we've
[8:12] been talking about, nothing will. On the
[8:15] bottom, software, by far the worst
[8:18] performing sector, down 12% year to
[8:20] date. IWM down 2.3%. and it outperformed
[8:24] for the month but gave a little bit back
[8:26] in the last week. Q's barely bunched for
[8:28] the week. Mag 7 up slightly after
[8:31] earnings. Uh here's gold, here's silver.
[8:35] Look at these days. Nothing over the
[8:37] past five years that's even comparable.
[8:40] Multiple standard deviation fall
[8:42] collapsing in these. And then you look
[8:43] monthto date. Uh another great month for
[8:47] silver up here. and gold had its best
[8:50] month in the last 15 years, even with
[8:53] the collapses on Friday. Don't get
[8:55] caught into the narratives of all these
[8:57] things. Uh tenure rates again as people
[8:59] trying to chase all over the place and
[9:01] tenure rates. Again, it's going to be a
[9:02] frustrating trade. The dollar big week
[9:05] down, a little bit of a bounce when
[9:06] Kevin Worsh was announced as the Fed,
[9:08] but regardless, the dollar
[9:12] big move.
[9:16] Trump exacerbated the situation by
[9:18] saying he's not concerned with the
[9:20] decline of the dollar. Uh the Fed held
[9:23] rates.
[9:24] And again, as you go through this and
[9:26] you think about why, I just want to keep
[9:28] reinforcing if we don't get bad news
[9:30] news leading into a payroll number on
[9:33] the on the labor side, then they won't
[9:35] do anything.
[9:36] This is literally becoming ridiculous
[9:39] that we are talking about humanoids in
[9:42] the same week and autonomous vehicles
[9:44] and autonomous trucks and we're
[9:46] literally talking about whether the
[9:49] possible that the economy is strong
[9:52] while we're laying off people or at
[9:54] least not hiring people. But Amazon
[9:56] announces they're laying off 16,000
[9:57] people due to AI. UPS to cut up to
[10:00] 30,000 jobs and closed facilities.
[10:03] Oracle may lay off 20 to 30,000. And
[10:05] these were all this week.
[10:07] Home Depot, explain it to me like I'm a
[10:10] 5-year-old. What signs of stabilization
[10:11] do you see in the jobs market? She's a
[10:13] perma bear, so I'm not really caring
[10:14] about that. I'm just showing the fact
[10:15] that the labor market is not well. And
[10:18] here's the Atlanta Fed wage growth
[10:20] tracker. Big leap down that is overlaid
[10:23] with the ECI year-over-year.
[10:27] So again, wages are coming down,
[10:29] productivity is going higher, profit
[10:31] margins are moving higher, and the fact
[10:34] that we have this happening and the Fed
[10:36] is not sure what to do makes sense.
[10:38] Consumer confidence hit a 12-year low
[10:40] this week. Stocks hit an all-time high
[10:43] on the same day.
[10:46] All right, PMIs going higher. Andreas
[10:49] Stenol Lararsson with all the regional
[10:52] PMIs out, I'd highlight manufactures
[10:54] doing exactly what we expect. you're
[10:55] getting the bounce if this comes in
[10:57] where he has it. It would be the highest
[10:58] level since 22. We just had a big
[11:00] Chicago number today. I think it's
[11:02] happening. So again, abundance and
[11:04] scarcity. You want to focus this entire
[11:06] year in both in in investing in whatever
[11:08] is scarce. You can go look at the
[11:10] charts. You can go try to pick them, but
[11:12] if it's scarce, go software will remain
[11:14] the abundance. If you want to go try and
[11:16] pick the abundance side, have fun. Uh I
[11:19] go through in this piece why AI is
[11:22] running even into silver. So for
[11:24] everyone thinking silver is purely
[11:25] precious metals, the critical mineral
[11:27] side, the raw industrial uh metal side,
[11:30] I want to make sure that people realize
[11:32] that the amount of metals needed for AI
[11:34] is insane. In particular, in the silver
[11:37] side, it's in every single component.
[11:39] And I wanted to make sure that this was
[11:40] not a precious metal story. Again, I
[11:43] showed this chart last week. It doesn't
[11:45] include the close from today, but you've
[11:46] got the same parabolic nature in memory
[11:49] as you do silver. So when people want to
[11:51] say that this is trading like a
[11:52] commodity, this is trading like scarcity
[11:54] parabas. If you guys want to hear some
[11:57] podcasts to listen to over the weekend
[11:59] on critical minerals, macro voices with
[12:01] Craig Tinddale, good one. Um, Smarter
[12:05] Markets with Robert Freedelland and Jeff
[12:07] Curry. Great one. I'd go listen to that.
[12:10] Um, in terms of the critical minerals,
[12:12] again, Trump can't speak about them more
[12:14] enough. He's doing deals with them.
[12:16] Everything is focused on it. he had an
[12:19] announcement on this. I think this in
[12:22] particular in terms of backtracking on
[12:24] price floors is going to start to be
[12:26] more of a story around the globe. Now, I
[12:28] wanted to highlight
[12:30] this chart in terms of the raw
[12:33] industrial metals. So, the CRB ren the
[12:35] CRB raw industrials metals which has a
[12:38] high correlation to countries uh that
[12:41] are in the metals game and the minerals
[12:43] game over time. Uh, I created a copper
[12:46] and silver because they're one of or two
[12:48] of the components. This one ends up
[12:50] being threequarters copper and a quarter
[12:53] silver to normalize it to some degree
[12:55] with the volatility of silver. So, this
[12:57] is the chart over the course of the last
[12:59] 24 years into July. And I picked July
[13:02] because July is when I wrote the end of
[13:04] July is when I wrote the piece on PMIs
[13:07] going higher and believing that this
[13:09] would kick in in the coming uh in the
[13:11] coming months. And now that we're
[13:13] starting to see it as we get into the
[13:15] new year, uh you've had a huge move in
[13:18] DRAM prices and now you've taken this
[13:20] chart and this is the copper silver.
[13:22] You're going to have a pullback uh in
[13:24] this back towards 0.5. But the reality
[13:26] is for anyone looking unless you don't
[13:29] believe the copper silver story which is
[13:31] needed for electricity and the fact that
[13:33] we have supply shocks. Are we going to
[13:35] go straight up in a line? Absolutely
[13:36] not. We did as I showed before. We had a
[13:39] huge month in silver again for the fifth
[13:41] month in a row. But you got to pay
[13:43] attention to the story that the raw
[13:45] industrial metals, aluminum, all these,
[13:46] there's going to be switching that
[13:48] happens because when those metals go up
[13:50] up, the ones that can be switched are
[13:52] going to go to other places. This is the
[13:54] overlay of that chart with EWZ. Brazil.
[13:57] Brazil. Brazil. I can't emphasize it
[14:00] enough. I've talked about it. I think it
[14:02] fits in so many things for this year. So
[14:04] much so that this week for 22V I will
[14:07] have a paper come out on Brazil and the
[14:09] green computer arbitrage of the AI
[14:11] decade decade and it goes through a
[14:13] variety of different things. Brazil
[14:15] finishes up the month again this isn't
[14:17] completely updated. It was down about
[14:19] one a little over 1% but you're still
[14:21] talking about the biggest rise since
[14:22] 2020.
[14:25] Um silver with the price going higher.
[14:28] This is how it's going to trigger into
[14:29] the energy market. So silver is a major
[14:33] part of solar and just to give you an
[14:35] idea this chart here shows the black uh
[14:39] component here is silver's percent of
[14:42] solar panel cost. So it went from 3% in
[14:45] 2019. This was it into this year now
[14:49] with prices up and this was using a
[14:51] price of around 90 something dollars. So
[14:53] again you're talking about a rise in
[14:56] this where it's going to make it harder
[14:57] for solar to be a buildout. Silver
[14:59] squeeze leaves solar panel makers
[15:01] feeling the heat. Solar is for power. We
[15:03] need power. Go listen to this podcast.
[15:06] Rick Rule went through and he talked
[15:08] about the fact that oil it's time for
[15:10] oil and gas stocks um which we're
[15:13] seeing. But I think the rationale was
[15:15] we're in the parts of a boom now of
[15:19] of exactly what we said. So again just
[15:22] highlighting this is before the close on
[15:25] Friday. This is actually through
[15:26] Thursday. These numbers have come down.
[15:28] But the main point is again you are
[15:30] getting massive moves in materials and
[15:34] energy for investors for mutual funds
[15:36] for everyone who's looking at it. You
[15:38] have to be involved. Chevron verse
[15:41] salesforce.com I've highlighted this
[15:44] trade since in this area in here we've
[15:47] seen the breakout. This is not updated
[15:49] through uh all of Friday. I will give
[15:52] you the
[15:54] monthly performance. If you were smart
[15:57] enough to put this trade on and follow
[15:58] my lead, you're up 45% this month. Uh I
[16:02] did have some people that thanked me at
[16:04] the event in Miami who had put this on
[16:06] by buying calls on Chevron and puts on
[16:09] Salesforce.com. Again, I've shown XLE
[16:12] over IGV. It doesn't really matter to
[16:14] me. And this is the chart. We're at the
[16:16] very beginning stages of this, guys. Um
[16:18] this is going to continue to go higher
[16:20] in my opinion. Software is in the early
[16:22] stages of a rerating. Power is in an
[16:25] early stage and I do think you're going
[16:26] to see energy prices move higher at a
[16:28] time when everyone is bearish on them.
[16:31] SMH versus IGV. Again, this has been my
[16:33] favorite trade. Semis, I've now moved
[16:35] into energy as more as one that is less
[16:38] invested in. I'd rather be in the energy
[16:40] trade for this year uh than SMH.
[16:44] Although X XSD or any of the package
[16:47] things that I did which are more the
[16:49] analog names, I do think those names can
[16:51] double this year. But when you're
[16:53] dealing with Nvidia and Broadcom in
[16:55] particular, I think those have already
[16:56] had a good run as has Intel. I think
[16:58] you're still going to have rises in this
[16:59] because we don't have enough chips. Uh
[17:02] but again, I think you look to energy at
[17:04] this point. OpenAI on Friday, fourth
[17:08] quarter IPO plan to race to beat
[17:11] anthropic to the market. And I've said
[17:13] this before. I think Open AI is just a
[17:15] cancer right now for uh any of the
[17:18] hyperscalers.
[17:20] They are forcing so many other issues,
[17:23] doing deals. Everyone's trying to frontr
[17:25] run each other. They're doing deals with
[17:27] any place where scarcity is going to
[17:28] kick in and I think that ends up to a
[17:30] race to the bottom. I think it hurts in
[17:32] terms of the cloud ability because of
[17:34] the data center side. The models are
[17:36] moving extremely fast regardless because
[17:38] of the efficiency side. But I think open
[17:40] AAI, I think XAI, I think Enthropic, all
[17:43] of these fourth quarter or all of these
[17:44] IPOs that are coming to the market along
[17:46] with SpaceX, you're talking about
[17:48] trillions of dollars
[17:50] for growth companies out there or the
[17:52] growth bucket, you're going to be
[17:53] saturated with deals. And I think OpenAI
[17:55] is just going to continue to hurt
[17:57] things. And I want to remind people back
[17:59] in June of 2007, Blackstone, heavily
[18:02] into credit and heavily into the real
[18:03] estate market, basically came to the
[18:06] market just before the great financial
[18:07] crisis. This tends to happen over time
[18:09] in terms of IPOs or deals coming out
[18:11] around the time of peak. I don't think
[18:13] this is a peak in the market. But I do
[18:16] think for growth, I do think for
[18:19] anything related to AI and software and
[18:22] code, I'd be very very wary that OpenAI,
[18:25] that XAI, that anthropic that all of
[18:27] these deals coming out have an eerie
[18:30] similar feeling to what happened when
[18:31] Blackstone went in. especially when you
[18:34] combine it with the fact that they got
[18:35] the curse of being this at the end of
[18:39] last year on December 29th. So, I'm
[18:41] going to stick with my side that you
[18:43] want to be long commodities, you want to
[18:44] be long industrials, you want to be long
[18:47] this stuff. Um, I just want to highlight
[18:49] this came out this week. I saw this in
[18:52] um in X and I just wanted to take you
[18:56] through basically taking these dots and
[18:58] kind of bringing it together and getting
[19:01] ideas. Again, I'll be doing these demos
[19:04] for people. Uh, but when you take and
[19:06] you see something like that and you put
[19:07] it into chatbt or Gemini and you ask to
[19:10] explain it and what it means and you
[19:12] look for ideas and it shows you high
[19:14] bandwidth memory, Seridus, Pam, optical.
[19:19] You get these stories and you start
[19:21] realizing, okay, we've got memory
[19:23] pressure. How how how big is the
[19:25] scarcity issue? One major manufacturer
[19:27] saying all of its nan flash production
[19:29] for 2026 is already sold out. how the AI
[19:32] buildout is pushing up electronics
[19:33] prices. This will be a story as we go.
[19:35] Samsung raises NAM prices by 100%.
[19:39] Supply can't keep up with demand. And
[19:41] again, I remind you, nobody wanted to be
[19:44] long memory in the summer of last year.
[19:46] Now, Elon Musk is saying memory the
[19:49] memory needs on the earnings call. Frame
[19:51] memory as a long-term key bottleneck for
[19:53] Tesla's AI ambitions and said Tesla may
[19:56] need to build its own large
[19:57] semiconductor fab that includes advanced
[19:59] memory. If you're not going to trust me,
[20:01] at least trust Elon. Um, he said when he
[20:04] looks three to four years out, the main
[20:05] limiting factor in Tesla's growth is
[20:07] chip production. Do you really want to
[20:10] belong software when everyone is talking
[20:12] about the fact that we don't have enough
[20:14] chips? That AI demand is insatiable.
[20:18] Jensen Yu Wong is talking about $85
[20:20] trillion of a buildout. And to have that
[20:23] buildout work, it's all going to be in
[20:24] AI factories and on the edge, which
[20:26] means we need as much chips as we can.
[20:29] This story is not over yet. Memories had
[20:32] a big run. I'm sure best case scenario
[20:34] over the next year. Uh sorry, two years.
[20:37] Best case scenario, I can't imagine
[20:39] Micron more than doubling. I can imagine
[20:42] battery names more than doubling. I can
[20:44] imagine many of the smaller semi-names
[20:47] going up, but we've taken these things
[20:49] up to big levels. But until we start
[20:50] seeing something where we catch up on
[20:52] the earnings front, you're going to keep
[20:54] getting the movements that you got. So
[20:56] SKH Highix reported 66% year-over-year
[20:59] increase and 137% increase in operating
[21:03] profit. HBM revenue more than doubled
[21:06] year-on-year. Um we're still in the
[21:08] early stages, believe it or not, of
[21:10] memory in terms of the story. Um,
[21:12] SanDisk
[21:14] didn't close up as much as it was here
[21:16] pre-market on Friday, but I I think it's
[21:19] just more the earnings per share beat by
[21:22] almost 100%. Uh, on this the guidance 13
[21:27] verse 421. When in history have you seen
[21:30] those types of numbers? So, I took that
[21:32] chart, I said, I want to expand uh past
[21:35] that. What are the other verticals that
[21:37] will benefit from this chart? I put it
[21:39] into nano banana and this is what you
[21:41] get. Advanced compute, power
[21:43] infrastructure, metals and minerals to
[21:45] be able to build this which is what is
[21:47] happening. We actually need those
[21:50] things. You'll come back in three years
[21:52] and you'll be there. ASML, I've talked
[21:54] about this one. Uh again, it's been a
[21:56] good time to be long these names. Apple
[21:59] sellers surge on 16% on staggering
[22:02] iPhone demand. I mentioned how much
[22:04] better the iPhone is. I am grateful for
[22:06] having got this when I did because the
[22:09] cost of everything is about to go higher
[22:11] because of memory. Significant increase
[22:12] in memory prices make smartphones more
[22:14] expensive. If you haven't purchased your
[22:17] high-end computer, this Mac I bought in
[22:19] the last two months, the phone I bought
[22:21] in the last two months, go get them
[22:23] before the prices go up significantly
[22:25] because they're going to go up
[22:26] significantly. Get ready. Computers in
[22:28] 2027 will be triple the price they are
[22:30] now. As Clawbot gets more popular,
[22:33] tokens get more expensive, and open
[22:35] source models become more powerful,
[22:36] there will be an obscene consumer rush
[22:38] to buy high hardware. I won't read you
[22:41] all of it, but let's just say you want
[22:44] to be long the things associated with
[22:45] it. This week's video for the week
[22:49] is Deis Sabas and Dario Modi. This was
[22:51] actually last week, but I didn't see it
[22:53] till this week. I thought it was
[22:55] worthwhile to show this week because
[22:57] they got into a lot of different topics,
[22:58] but also these are the two nicest people
[23:01] in the AI world. These are the ones that
[23:02] I think are the most honest. They're the
[23:04] ones that will tell you the negatives
[23:07] along with the positives. Uh but this
[23:09] interview, they went back and they
[23:10] talked about a lot of different things
[23:12] that are important. Um Habas remains
[23:15] cautious on, let's say, the creativity
[23:17] for science side, uh which is his
[23:20] definition of AGI. Uh so he's still
[23:23] talking you know five years for true ASI
[23:27] or where you're going to actually have
[23:28] science breakthroughs meaning more than
[23:30] five years while Dario Modi is a little
[23:33] bit more. They both talked about jobs
[23:35] and they talked about the fact that
[23:37] there's going to be a medium-term
[23:40] issue. Um it just is. And I think their
[23:44] main point was governments are
[23:45] unprepared. You're going to start to see
[23:47] this. It's not just governments,
[23:48] organizations are unprepared. Uh I'll
[23:50] talk about that more as we kind of move
[23:52] forward. Demos on the AI shift 10 times
[23:55] bigger and faster than the industrial
[23:56] revolution. I mean he's talking about
[23:59] massive GDP expansion on this. Uh in a
[24:02] world where AI moves exponentially, what
[24:04] happens to the value of longduration
[24:06] assets? I wanted to bring this up
[24:09] because I I just can't get around the
[24:11] same way I talked about tenure rates and
[24:13] the Fed thinking about things like
[24:15] today. If you're involved in long
[24:18] duration assets, the amount of
[24:19] assumptions you're taking without the
[24:21] liquidity associated with it, I don't
[24:23] see how they have any value. Personally,
[24:26] you can't predict six months out. That's
[24:27] how fast things are moving. I think
[24:29] everyone in the market needs to adjust
[24:31] to time. We've seen things with memory.
[24:33] We've seen things with silver. We've
[24:35] seen things with gold. I think
[24:37] longduration assets, I really don't know
[24:40] how you can think more than three years
[24:42] out on anything. So, I think these
[24:43] things are going to remain an issue. Uh,
[24:45] I think the money is going to look for
[24:47] liquid things. I think the volumes on
[24:48] the liquidity side are going to be there
[24:50] because I think people still have to
[24:51] manage money. I only bring that up
[24:53] because I see higher volatility uh, in
[24:56] terms of the markets and you're going to
[24:57] have to get comfortable because the
[24:58] things to invest in now are about
[25:00] scarcity. They can run up dramatically.
[25:02] They can fall back 50% in a short amount
[25:04] of time as we saw with silver and gold.
[25:06] You're going to see this over and over
[25:08] again. You're not going to be able to be
[25:09] involved in these on any kind of a basis
[25:11] and I don't think you're going to have
[25:12] time to get in. You're going to have to
[25:13] be buying on days if you're not in when
[25:16] things are crazy. You're going to have
[25:17] to be getting out during the par
[25:18] parabolic side, but you run the risk of
[25:20] these things never coming back to you. I
[25:22] know multiple people in Micron Sandisk
[25:24] that now are on board, but they can't
[25:27] buy them up here. Um, the arrival of
[25:29] AGI, just to show how close it is,
[25:31] Google DeepMind announced
[25:34] that Shane Le is the chief AGI
[25:38] scientist, the co-founder of DeepMine.
[25:41] Humanity
[25:44] needs to wake up. Dario Modi put out a
[25:46] new
[25:48] warning in an AI constitution.
[25:52] Uh it's worth going through and it's
[25:54] just highlighting the risks that are
[25:56] coming. And this gets back to Demisab
[25:58] said the same thing. Sam Alman also said
[26:00] the same thing which I'll read in a
[26:01] second. Governments are not prepared.
[26:04] But this goes further because I have to
[26:06] answer constantly quantum risk for
[26:08] Bitcoin. And I sit there and say, "Are
[26:10] you kidding me? With everything going
[26:11] on, you're worried about quantum risk?"
[26:14] And yet still, intelligent people keep
[26:16] coming my way. Hey people, I want you to
[26:18] read this and just realize we are at the
[26:20] level now where AI agent swarms are
[26:23] going to be a story.
[26:25] Bitcoin is far far far more prepared for
[26:29] agent AI agent swarms than JP Morgan and
[26:32] Goldman Sachs and any of the banks you
[26:34] want. We saw a run on a bank recently
[26:36] with SF SVB. what would happen if this
[26:39] happened with multiple agent swarms or
[26:41] the money was taken out and I'm sure it
[26:43] would end up in Bitcoin um at some point
[26:46] over the course of the next year. The
[26:47] more I listen to this, the more that
[26:49] they are screaming, we've already seen
[26:51] the danger that's happening. When these
[26:53] guys say we are going to reach the cyber
[26:55] security high level on our preparedness
[26:56] framework soon for for open AI,
[27:00] we're getting there. It is very
[27:01] important the world adopts these tools
[27:03] quickly to make software more secure.
[27:04] There will be many very capable models
[27:07] in the world soon. This is starting to
[27:09] get out of control. These guys have to
[27:10] release this. They're spending the
[27:12] money. They're going to release things
[27:13] before the security under the other side
[27:16] is ready. So, the system is going to be
[27:17] ahead of it. I would just be very wary
[27:21] and thinking about the fact that this
[27:23] quantum risk for Bitcoin, which is
[27:26] multiple years down the line. In fact,
[27:30] timelines also remain highly contested
[27:32] with projections ranging from a few
[27:34] years to several decades.
[27:37] They're going to have preparedness
[27:38] because they're going to watch what
[27:40] happens with agent swarms on these old
[27:42] archaic fiat systems. Mark my words,
[27:45] there will be an event with agent swarms
[27:46] that will happen in the banks. That will
[27:49] be something that goes on whether it's
[27:51] the stock exchange, all of these things.
[27:53] We are getting to the point where
[27:55] billions of AI agents will be able to
[27:57] target things and the defense is just
[28:00] not there. AI token consumption
[28:02] increased exponentially last year due to
[28:04] reasoning models. Now it's going to
[28:06] increase exponentially again with
[28:08] agentic workloads.
[28:10] The agentic world is about to explode
[28:14] guys.
[28:15] Here's more of the agentic world. The
[28:17] future of personal AI assistance. So
[28:20] this is all Claude Claudebot which has
[28:22] gained most of the stories. That was an
[28:24] Apple Mac story place. Uh the most
[28:27] important video you'll watch. Claudebot
[28:28] has taken X by storm and for good
[28:30] reason. It's the greatest application of
[28:31] AI ever. Every week there's a new thing
[28:33] by Claude guys. Every single week Gavin
[28:35] Baker put this out just to show that the
[28:37] productivity boom is here already and
[28:39] it's going to explode as AI agents come
[28:41] in. Revenue per employee up 75% for the
[28:44] top decile of AI software companies in
[28:46] 2025. Remember this is with software
[28:49] going down
[28:51] probably doesn't slow down in 2026 given
[28:53] the December revolution in AI coding
[28:55] agents.
[28:59] Prism accelerating science writing and
[29:01] collaboration with AI. This is just
[29:03] speeding up the process of the medical
[29:05] side allowing scientists to go on just
[29:08] continues to make it easier and easier
[29:10] for everyone to make breakthroughs. And
[29:13] here's one of them again David Sinclair.
[29:16] Uh I follow him there as part of my HRV
[29:20] substack. Uh I do a lot of things from
[29:23] his book FDA greenlights life biosens
[29:26] human clinical trial on a near total
[29:28] reset. A near total reset in longevity.
[29:34] First of its kind freightonly driverless
[29:36] truck network hits commercial scale.
[29:39] This was on Wednesday. Get ready to see
[29:41] trucks driving down the road with no
[29:43] people in them at some point. Next six
[29:46] months are going to be really weird.
[29:49] I follow AI adoption pretty closely and
[29:51] I've never seen such a yawning inside
[29:53] outside gap. People in San Francisco are
[29:55] putting multi- aent claude swarms in
[29:58] charge of their lives consulting chat
[30:00] bots before every decision wireheading.
[30:02] If you read through the rest of it,
[30:05] this is all and this is Kevin Ruse who I
[30:07] believe is a New York Times technology
[30:09] writer. He's saying at the same time
[30:12] you're finding people that don't use it.
[30:15] I completely agree as I do this for a
[30:17] living and I talk to people about it.
[30:19] Bill Gurley, the number one way to
[30:21] protect your career from AI is to be the
[30:22] most aggressive user of AI in your
[30:24] field. I'll say it again and again on
[30:26] here. Uh mainly because it's resonating.
[30:29] Uh I will be my I did my first video for
[30:32] the webinar series. Um we will have the
[30:34] payw wall going up in the next two
[30:36] weeks. They're just going through all
[30:37] the bugs at this point. I know that I
[30:38] say that every week. It is coming, guys.
[30:40] I know you're waiting. I know you've
[30:41] been patient. Your kids have to be ready
[30:44] for this. They have to be ready. They're
[30:46] the ones that have to be AI native. They
[30:48] have no choice. They're not being taught
[30:49] it in the school before they get there.
[30:51] They have to be using it all week long.
[30:53] And I'm going to show you. I thought the
[30:55] first video went well. Uh this is kind
[30:58] of a summary on what it is.
[31:02] Repetition builds comfort. Comfort
[31:04] builds fluency. Fluency turns AI from a
[31:07] tool into a partner. AI is not an
[31:09] answerbot. Its real value comes from
[31:11] exploration. If you guys have gotten to
[31:14] like the way that I think about markets,
[31:16] the way that I think about going from an
[31:17] idea to something, that all has been
[31:20] exacerbated by AI. I always had that.
[31:22] I've been a curious person. I collect
[31:24] dots, as I call them, as I go along. For
[31:27] those of you who listen to my podcast,
[31:28] Green Marbles, I collect them all the
[31:30] time. I had one uh moment of integration
[31:33] that happened where all of a sudden the
[31:36] systems thinking side of my brain popped
[31:37] out and answered a question for me. I'll
[31:40] get to that at the end. Human creativity
[31:42] versus AI optimization. The only edge we
[31:44] have as people is right now on the
[31:46] context side and finding the ideas and
[31:48] being able to filter. Our creativity is
[31:51] what allows us with the ability of
[31:54] having a polymath with us find things.
[31:58] That is the game plan for everyone. And
[32:00] here's an example of it. There was a
[32:02] deal this week. Meta and Corning. I've
[32:06] recommended Corning before.
[32:09] a $6 billion multi-year deal between
[32:11] Meta and Corning. Optical fiber is the
[32:14] emerging physical bottleneck. So
[32:16] remember, memory is a bottleneck. I've
[32:19] talked about Corning. I've talked about
[32:20] it and referenced it. So it's up a good
[32:23] amount since then. We're not even in the
[32:25] big phase of Corning. This is before the
[32:27] glass buildout. This is for the need
[32:29] before the needs of the autos, the needs
[32:31] of the phones, the needs of the
[32:32] computers, which is coming. They're
[32:34] talking about a shortage in optical
[32:35] fiber. Why the scaling east west traffic
[32:37] that goes back to the thing I showed you
[32:39] earlier with the Steve Jobs side in it.
[32:41] So let's go back to my how do you take
[32:45] something like that that happens and
[32:46] immediately go hey there's a news
[32:48] article give me and I run my skill on
[32:51] this and then I go in and do my deep
[32:54] research. So I did that. I ran this
[32:57] while I was on a plane flying from Miami
[32:59] to LA. It runs this whole thing. It ends
[33:01] up giving me a fivepage thing. Viser
[33:04] labs the fiber ingredient it goes
[33:07] through everything in there creates all
[33:09] the details it gives me a fivepage thing
[33:13] with a bottom line this this deal is
[33:15] less about Corning specific revenue
[33:16] uplift and more about institutional
[33:18] validation the optical bottleneck thesis
[33:20] the constraint migration from GPU power
[33:22] optical is now being priced into
[33:23] hyperscaler procurement behavior this
[33:26] all gets into data movement data traffic
[33:28] optical is a better way to go through it
[33:30] so I take all of that stuff it goes
[33:32] through my investment process and what
[33:33] do come out with
[33:36] a list of about 15 to 20 names in this
[33:39] framework that I've shown over and over
[33:40] again. So for any of you who want it,
[33:42] the optical fiber bottleneck, there's a
[33:44] bunch of names in here that will benefit
[33:47] all with the same thing. They've got a
[33:48] technical score from Johnny Ro. They've
[33:50] got a commentary score in terms of are
[33:52] they already seeing the optical fiber
[33:55] side? Is it already showing up and are
[33:57] is it already progressing over the last
[33:58] three earnings reports? And again, are
[34:00] revisions heading higher in a strong
[34:03] way? Are they going? All of that's going
[34:05] here is
[34:07] clawed. So, I basically took I don't
[34:11] know two minutes. I put the ad in. I
[34:13] launched it and then I went in here and
[34:16] I asked it to do something on Microsoft
[34:19] for me. So, can you go through the
[34:21] recent Microsoft earnings report and put
[34:22] together a model on how their financial
[34:24] model has changed due to rising RPOS? If
[34:26] data centers are bottlenecked and they
[34:28] keep spending at current rate, when will
[34:30] this become an issue? So, it goes
[34:32] through, finds all the estimates, puts
[34:34] this in based on everything it gets from
[34:36] both the earnings commentary,
[34:38] and then it comes through and it says
[34:40] this. Now, only thing I did was this.
[34:42] That's all I did. I hit it. It spent
[34:44] about 3 to five minutes going through.
[34:47] It gave me a Excel file with all of this
[34:50] stuff in it with numbers that I can go
[34:52] check with the street. I can do whatever
[34:53] I want. But if you wanted to ask a
[34:55] question, have it modeled out, and then
[34:57] go bring it back and look at your own
[34:59] analysts and have them go through it.
[35:01] It's saying if things don't change and
[35:03] we go at this run rate of RPOS and this
[35:06] capex, and that's the thing I want. How
[35:09] do you get this number to go faster if
[35:11] the RPOS are going and the reality is
[35:14] you want to see what's going to happen?
[35:16] So, I did that. You can do it on your
[35:18] own. HRV, thanks to everyone who's been
[35:21] going through it. I made the best new
[35:24] Substack list number five uh on the HRV
[35:27] thing. Uh I I've really enjoyed talking
[35:30] back and forth with people who have got
[35:32] it. All right, Bitcoin still asleep.
[35:36] Doesn't want to move given what went on
[35:38] this week. It was actually a pretty
[35:39] quiet week. Um given how much silver and
[35:41] gold were down, didn't follow silver and
[35:43] gold on the way up. Kind of kind of went
[35:45] through its little girrations. The chart
[35:47] is still horrible. uh until we get a
[35:49] change in it, it is in the list of hey,
[35:52] we need this thing to go for the short
[35:54] term. So, from a trading perspective,
[35:56] it's very hard to buy something like
[35:57] this other than stick your toes in and
[35:59] not and stop yourself out. It has not
[36:02] acted well at all. I wrote this paper,
[36:05] the Substack,
[36:07] uh I highly recommend everyone read it.
[36:10] Um it goes for the silver side, it goes
[36:13] for the micron side, it goes for
[36:14] everything.
[36:16] It was clear that there was parabolic
[36:18] envy. This is all at the event I was at.
[36:20] I just want to remind people silver
[36:22] didn't move for 14 years and then when
[36:25] it finally did, it went parabolic.
[36:27] Micron didn't move for a long time. It
[36:29] went parabolic. Corning didn't move for
[36:31] a long time, 15 years, and then it went
[36:33] parabolic. I'm just telling you these,
[36:35] this is the world we're in. When things
[36:38] turn, they're going to go quickly. The
[36:41] sausage squeeze of how many names on the
[36:43] software side are not going to
[36:45] participate. I do not believe that the
[36:46] Mag Seven can get you the alpha that is
[36:49] necessary. I think there's a lot of
[36:51] headwinds for them. Part of them is the
[36:52] RPO, part of them is the IPO that are
[36:54] coming. Part of them is just the amount
[36:56] of debt these guys are are now taking. I
[36:58] want to be long energy. I want to be
[37:00] long materials. But when you go through
[37:01] things, I still believe Bitcoin will be
[37:05] part of this equation once we get
[37:07] through it. One more thing uh to finish
[37:10] up. Uh I have never mentioned NFTTS. I
[37:13] have never brought it in. So when I
[37:15] mentioned a moment of integration, I do
[37:17] want to highlight that between
[37:19] conversations I've had with very
[37:20] respected people in the NFT side, but
[37:22] also something Demis has said last week
[37:27] on AI purpose. I finally have a place to
[37:30] where I understand where NFTs will be an
[37:32] important thing. This is not a digital
[37:34] art thing. This is not a people thing.
[37:35] This is not a cyberpunk thing. This is
[37:37] not Pudgy Penguins. none of the things
[37:40] that you guys probably remember from
[37:42] 2122.
[37:44] But I will say as part of the utility
[37:46] function for what is happening in
[37:47] crypto, stable coins, realworld asset
[37:50] tokenization along with NFTTS is a big
[37:53] deal. I think you should read this
[37:55] report. I had a fun time writing it. I
[37:57] had a fun time coordinating with some
[37:58] very very smart people in the space. But
[38:01] a lot of these issues have come up and I
[38:03] look forward to spending more time
[38:04] talking about in the future. That's it
[38:06] for this week. I look forward to being
[38:08] back and doing it from Brooklyn next
[38:10] week. Thanks, guys.