All Time Highs Built On A Compute Shortage — Jordi Visser (23 abril 2026)

[0:00] All right, new all-time highs. I did not
[0:03] think we would be seeing those until
[0:06] possibly later this year if at all. Uh
[0:10] multiple compression is still my overall
[0:12] story, but clearly uh
[0:15] the rally has been unprecedented. I'll
[0:18] go through a lot of that. Uh but you got
[0:20] new all-time highs. Uh a lot of new
[0:22] things I've added to the website that
[0:26] I'll show you some of this week. These
[0:28] will keep coming, guys. For those of you
[0:30] who've not only been subscribing, but
[0:32] been reaching out and have been doing
[0:33] well with the model portfolio, which has
[0:36] ripped the highs. Uh
[0:37] but I'm getting a little more granular,
[0:40] adding some technical pieces. I did a uh
[0:43] a spotlight update this week around
[0:45] Oracle, uh which
[0:48] proved to be timely. There was some
[0:49] names in there that are in the model
[0:51] portfolio list. Uh those I will continue
[0:54] to talk about cuz they were mentioned
[0:55] again this week on a podcast. Uh
[0:58] regime change. Inflation is still
[1:01] spreading, guys, and don't forget it. I
[1:03] still think it's going to be a big part
[1:05] of this year. Not in terms of runaway
[1:06] inflation, but just the supply
[1:08] disruptions at this point in the AI
[1:11] macro regime, which is all about
[1:13] scarcity. Uh
[1:15] there's a compute shortage. I'm going to
[1:16] take you through that. Very, very
[1:18] important because it is something which
[1:21] will have an impact. Uh
[1:23] a lot of important podcasts this week,
[1:26] some new ideas for me out of them. Uh I
[1:28] wrote some papers that I really on
[1:30] Friday uh related to uh not just the
[1:33] terrafab, but just kind of this compute
[1:36] shortage and what it actually means. And
[1:38] new edge device report.
[1:40] Bitcoin breaks out. So, let's get right
[1:42] to it. Uh I'm going to go through these
[1:44] points quick. The S&P's up for the third
[1:45] week in a row. And again, we were down 5
[1:47] weeks in a row. Now, we're up 3 weeks in
[1:50] a row. I think you're going to see a lot
[1:52] of these pendulum swings this year. I do
[1:53] not think this is the the thing as last
[1:56] year. I do not think this is the same.
[1:58] In fact, I'll use these um
[2:00] down down volume to up volume, nothing.
[2:04] Uh
[2:04] I I again, we never saw a panic.
[2:07] But then we also not never saw panic buy
[2:09] side. This has been more about rotation.
[2:11] This has been been about destroying
[2:13] gross exposure uh rather than uh the net
[2:17] side as nets got cleaned up when people
[2:19] were panicking on Iran.
[2:22] We get a rally back the other direction.
[2:24] I'm going to show you why some of the
[2:25] CTA positioning uh had a huge impact on
[2:28] this.
[2:29] Uh we never saw the kind of capitulation
[2:32] or the kind of turbulence that shows up
[2:34] near bottom. So, again, last year in
[2:37] liberation day, we kind of went down uh
[2:39] without having a turbulent day and
[2:42] across assets and we've had them all
[2:44] year, but none of them have gone in. So,
[2:46] now I'll start looking again for some
[2:48] kind of a quiet time. We're back above
[2:51] obviously the moving averages. The vol's
[2:52] come back down in terms of the Fed. So,
[2:55] we'll look for turbulent signs, but
[2:58] rather than focus on the direction of
[3:00] the market, I'm going to say this all
[3:01] year because I think it is messing with
[3:03] people. Uh too many people are worried
[3:05] about whether the S&P is trading higher
[3:06] or lower. There are names that are up
[3:09] well over 100%. There's a lot of names
[3:11] up over 50, and these are all the names
[3:13] that are related to this.
[3:15] And then the names uh that again have
[3:17] been hit hard, the software names and
[3:20] things like that, they've bounced, but
[3:22] don't don't get fooled. You want to keep
[3:24] looking into downtrends, like I said,
[3:26] for names like Micron, which went from
[3:28] 330 back up to the highs. Uh the names
[3:31] that are in the model portfolio, you
[3:33] should have technical levels on them and
[3:34] be ready to trade them because this is a
[3:36] structural advantage. Despite everything
[3:39] that's happened uh in Iran, we still
[3:41] have oil so far this year up 24% for the
[3:45] 12th contract out. This is why this is
[3:48] important. Um last year we spent most of
[3:50] the time with oil below 65 and below 60
[3:53] for a good portion of it. Regardless of
[3:56] whether
[3:57] oil stays up at 6 80, it stays up at 70
[4:01] for near term, the longer term oil
[4:03] prices have gone higher and I do not
[4:05] think that that's going to shift as PMIs
[4:08] rise, as the CapEx spend continues and
[4:12] the data center shortage becomes real.
[4:13] So, think about energy independence and
[4:16] I don't I think you're getting a gift
[4:18] here to be looking at the power trade
[4:20] and the energy trade and I'll go through
[4:21] that. But, here are the names again
[4:23] year-to-date. So, forget everything. The
[4:25] S&P's up 4%. It is not being driven by
[4:28] this stuff down here. It is being driven
[4:30] still even in the rally by this stuff up
[4:32] here. Energy, semis, capital goods,
[4:34] materials.
[4:36] Needless to say, when you get in the
[4:37] tech hardware, transportation, these are
[4:39] all PMI sensitive positive names and if
[4:41] you don't believe me, look at the
[4:43] year-to-date performance for the
[4:44] industry groups. What type of macro
[4:46] regime does this best represent?
[4:48] Inflationary supply constrained CapEx
[4:50] regime rather than a clean recession or
[4:52] a broad risk. That's the way you should
[4:54] be thinking is a reflationary
[4:56] stagflation-leaning industrial scarcity
[4:58] regime. That's what you want to make
[5:00] money on.
[5:01] That's what the model portfolio is built
[5:03] on because it's all built on the fact
[5:05] that we are running out of things that
[5:08] we need for artificial intelligence,
[5:10] which as Jensen Huang said this week
[5:12] with Dwarkesh, which I'll go through
[5:13] later,
[5:15] you need
[5:16] power, you need chips, and you need
[5:19] engineers.
[5:21] Uh we've got a shortage right now of a
[5:22] lot of chips. We've got a shortage of
[5:24] memory. We've got a shortage of CPUs
[5:25] now.
[5:26] And we are about to have a shortage in
[5:28] power that will become more of an issue.
[5:30] So far, we've made all of this without
[5:32] the data centers being built because of
[5:33] efficiency gains, but now you're
[5:35] starting to see signs that this is not
[5:37] the case. Warren Pies good thing here
[5:39] just showing
[5:41] uh kind of the unprecedented move, but
[5:43] how few 50 new 52-week highs occurred
[5:46] with new all-time highs.
[5:48] Uh Uh and the most extreme momentum
[5:51] event 40 years, the Nasdaq 100's RSI
[5:54] went from 28 oversold to 70.5 by April
[5:58] 15th in just 11 trading sessions. That's
[6:01] the fastest oversold to overbought
[6:03] transition. The previous fat- fastest
[6:06] was 25 sessions after liberation day.
[6:08] So, it took 11 this year, 25 last year.
[6:11] Both of those were records. Whenever I
[6:13] see those types of it never happened
[6:15] since in 40 years. It and the the the
[6:18] one that was the next one was last year.
[6:21] That just shows the violence that's
[6:23] going on. I think this is going to
[6:25] become part of the nature and part of
[6:28] what we end up living with. But you can
[6:30] see here that in terms of the CTA
[6:33] buying, we went from
[6:34] massive net selling to massive net
[6:36] buying in a very short amount of time.
[6:38] Now, last year, we never saw the buying.
[6:42] We did see this at these other two
[6:44] points here, which I'm going to show you
[6:46] the reason why. And that's because the
[6:48] S&P vol. These were the three times
[6:50] before where you saw
[6:52] uh
[6:53] large buying.
[6:55] To get that kind of buying, normally
[6:57] what happens, there's a correlation
[6:59] between the market moving lower by 15,
[7:01] you know, 10 15% and vol expanding in
[7:04] terms of realized vol. Now, the VIX went
[7:06] higher, but realized vol in the S&P, the
[7:09] actual moves never went higher because
[7:11] this was more about factor vol. This was
[7:13] rotational. It's why I've talked about
[7:15] this being a just a horrible year for
[7:18] hedge funds on a gross basis needing to
[7:20] trade this because the factor volatility
[7:22] remains extreme.
[7:24] The S&P vol's not moving. So, index
[7:26] doesn't really help you that much in
[7:27] terms of the movements. Now, if you
[7:29] bought a lot of call options near the
[7:31] lows,
[7:33] you made money because even though the
[7:35] VIX was high, it still went higher. But
[7:37] I think index vol's going to be an issue
[7:39] all year for people that trade it. Uh I
[7:41] do want to show you some of the
[7:43] parabolic look at these charts.
[7:46] Again, I think it's unprecedented some
[7:48] this is the transports.
[7:50] Again, follow through. I told you last
[7:52] week. I'm going to keep saying it. I am
[7:54] shocked at how many people are doubting
[7:56] the strength of the
[7:58] the nominal GDP strength of the economy.
[8:01] It's just evident that we have shortages
[8:04] and I think too many people economists
[8:06] that are perma bears in
[8:10] on X are too focused on the consumer.
[8:13] The consumer is fine. The balance sheets
[8:15] are fine. The unemployment rate is low.
[8:19] It's it's not a problem. There's no
[8:21] acceleration and people are feeling the
[8:23] pain of inflation, but you're getting
[8:25] parabolic charts higher in anything
[8:28] related to industrial. You've got the
[8:30] Cosby which came down and then moved
[8:32] right back up. You have Taiwan Stock
[8:35] Exchange. I mean these moves are just
[8:37] unprecedented in terms of how fast they
[8:39] are and how violent.
[8:43] You also have
[8:45] SMH over IGV. So when this started to
[8:48] break down people got comfortable. I
[8:50] mean the amount of people trying to pick
[8:51] the top in semis which is great for you
[8:53] guys watching these because you're not
[8:55] going to hear me talk about it other
[8:56] than a
[8:58] you know a temporary pendulum shift that
[9:00] goes and that's why during the Wednesday
[9:02] or Tuesday video I did this week of
[9:04] which for the subscribers I will be
[9:07] doing those more regularly. It could be
[9:09] once a week. It could be every two
[9:11] weeks. It could be three in a week. It
[9:13] just depends on what happens. Last week
[9:15] it was based on Oracle. I thought it was
[9:17] really important to highlight because
[9:19] we're entering a compute thing. We're
[9:21] kind of out of the software is ending
[9:23] part. We'll get another phase of that,
[9:25] but we are definitely in the compute
[9:26] shortage chart. So these are the model
[9:28] portfolio things. Every single one of
[9:30] them made new highs for the year aside
[9:34] from the
[9:36] the chemical ones or which oh no, the
[9:38] chemical ones did. It's the power one
[9:40] that didn't yet.
[9:41] Briefly did on Friday, but you have
[9:44] exponential moves in the packaging and
[9:46] semi size. The whole rack side is moving
[9:48] higher. I'm going to go through that.
[9:50] That's been a big one
[9:52] for me as well.
[9:54] Seeing where these names have gone and
[9:55] just how fast things like Marvell, which
[9:57] I've mentioned.
[9:58] And then you've got the optical side,
[10:00] which is just continued to go higher. We
[10:02] are in the first inning of these.
[10:05] So, when I say to you, like look at this
[10:06] green chart here of the packaging of
[10:08] semis,
[10:09] we are still just above. We only broke
[10:12] out of the 2024 level. We We're just
[10:14] above two-year highs.
[10:16] We have a long way to go on these
[10:18] things. So,
[10:19] this is for you subscribers, you can go
[10:22] get this this week. This is not
[10:24] completely finished, but there's a ton
[10:26] of good information in this already. So,
[10:28] what I'm going to start to do is give
[10:29] you guys each Friday uploaded a
[10:32] technical recap
[10:34] of all the names within there. It's
[10:36] going to highlight the top long setups,
[10:38] the names that are at risk, and the
[10:40] trigger points. This way for you guys
[10:41] that are trading,
[10:43] I'll give you some more information if
[10:45] you like trading on technicals,
[10:46] increasing your exposure, decreasing
[10:48] them. It'll give you the breakdown of
[10:50] all of the different components within
[10:52] the each individual basket, the 100
[10:54] there.
[10:55] For those of you seen the list, you know
[10:57] what all the names are in these. It is
[10:59] giving you a score in the overall index
[11:01] in terms of the breakdown. What
[11:03] percentage of them are above the 50-day?
[11:04] What percentage of them are above the
[11:05] 200-day? The average RSI within side the
[11:08] space. You can see that because of the
[11:10] correction we had, nothing is really
[11:12] that overbought in terms of the average
[11:14] RSI. So, you can go find the names that
[11:16] you like, and I give you the individual
[11:18] ones. So, you can go down and go look
[11:19] for every name, find the theme that it's
[11:22] in. You can go through it that way.
[11:25] And then you can actually go see what
[11:27] the score is. And if you're interested
[11:29] in the way I set up the technical
[11:30] scoring on this, Basically, you get
[11:33] points for each of these uh
[11:35] uh
[11:36] and most of them are momentum-based in
[11:38] terms of we want things that are
[11:39] trending well. We want the positive
[11:41] slope on the moving averages, 1-month
[11:43] positive. We want an RSI between 50 and
[11:46] 70 to avoid the overbought list. So, if
[11:48] you only want to buy hundreds, there you
[11:50] go. I'm also doing relative. I'm doing
[11:52] the volume 20-day average. Just things
[11:54] to help you from a technical basis. And
[11:56] so, now that the website for the
[11:58] subscribers is up and people are making
[12:00] money and they're sending things and
[12:01] they've had a good time with kind of
[12:03] having the ideas and also getting the
[12:05] updates,
[12:06] this is the way it's going to write. I'm
[12:07] going to write thematic papers. Um like
[12:09] I said, there's a new edge device
[12:11] report that I put there on the back of
[12:14] some things from Jensen and Elon. Uh a
[12:16] lot of those names are already included,
[12:18] but it's another way of looking at it.
[12:20] There's a few new ones for sure. Then,
[12:22] once you get into there, you're going to
[12:24] have the thematic lists. You'll get the
[12:26] spotlight pieces during the week if
[12:28] anything triggers, whether it's a
[12:29] technical thing, whether it's something
[12:31] that's correlated. Uh the acceleration
[12:33] commentary, this is going to be when
[12:35] earnings come out and I see something
[12:37] change for one of the names in the list,
[12:38] I'll put something on the subscriber
[12:40] list. You guys will get it. Uh the
[12:42] technical scoring, which will be weekly.
[12:44] This way you can look at if you guys
[12:45] want to, you know, move things up and
[12:47] down based on the way you like them, uh
[12:49] that's fine. You can do it from any of
[12:51] these three pieces as I add things. It's
[12:53] all to make you help you make better
[12:55] decisions uh and use the list
[12:57] accordingly, which I think are going to
[12:58] be secular trends. Uh
[13:00] in terms of the PMIs, the one data point
[13:03] we got was the Philly uh Fed. Uh this is
[13:06] the new orders for Philly Fed, which
[13:08] jumped up to the highest level since
[13:09] 2021 during the heat of PMIs being above
[13:12] 60. And this is new orders, which uh
[13:15] we'll see how I goes, but this would
[13:17] take the new orders category for the
[13:19] PMIs above 60.
[13:22] Uh I did, like I said, do a video this
[13:24] week on combing through the software
[13:26] rubble to give you uh I believe six
[13:29] names including Oracle as places where
[13:31] you could look. Those names worked well.
[13:34] Uh
[13:35] this was the video. Now, for those of
[13:37] you around the globe, I am in the
[13:39] process right now of working with some
[13:41] groups in terms of getting these videos.
[13:43] This one I'm not going to play it for
[13:45] you now, but if you're interested, you
[13:46] can let me know. I'm converting these
[13:48] into other languages so platforms that
[13:50] are out there around the globe that
[13:52] watch this that want to put this on
[13:53] their own in a language that they think
[13:56] people will be
[13:57] interested in seeing. The right now it
[14:00] was just a spotlight piece, but I'm
[14:01] happy to do the video I'm doing today
[14:02] and convert it. This one is in Chinese.
[14:05] Uh I did one in Japanese this week as
[14:07] well. This is the Japanese one and
[14:09] that's been sent out to a couple groups.
[14:11] So, for those of you are interested, I
[14:13] think this is an important thing to kind
[14:15] of get you guys the opportunity to have
[14:18] people around the globe that are
[14:19] interested in looking at data-driven
[14:20] stuff and model portfolio list. I'm
[14:23] happy to work with you guys on kind of
[14:24] going through how to how to get it in
[14:27] your own language. One negative that's
[14:29] starting to pop up I mentioned it last
[14:30] week which I think you I think we have
[14:32] to keep our eyes on just as a potential
[14:34] next phase of what happens and this is
[14:36] the earnings revisions. So, we are
[14:37] seeing revisions negative for the second
[14:39] week in a row. Now, as we saw we can get
[14:42] these from time to time particularly as
[14:43] we enter into an earning season,
[14:45] but we want to see where this goes to
[14:47] see if we end up in a situation. Just
[14:48] remember that during the rally last year
[14:50] that continued throughout the year
[14:51] coming out of liberation day,
[14:54] there was an overestimation of the
[14:56] impact of tariffs. I think this time
[14:58] there's been an underestimation of the
[15:00] impact. We're seeing this in the US, but
[15:02] we're also seeing it globally and
[15:04] remember as I said the US is a global
[15:06] index. Any disappointments in earnings
[15:09] and I think you'll see a leg lower in
[15:10] the S&P and then we'll have a leg higher
[15:12] and then we'll have a egg leg lower. Ed
[15:15] Yardeni just put something out just
[15:16] highlighting that as of now the
[15:18] bottom-up analysts are still moving
[15:19] higher in their earnings projections.
[15:22] This is my big fear. If we do see
[15:24] earnings uh revisions change and you
[15:27] start to get the AI disruption again
[15:29] because that is secular, we are going to
[15:31] have software names that are seeing the
[15:33] pain that we've worried about. Remember,
[15:35] it's only been 3 months
[15:37] of Anthropic seeing their revenue run
[15:39] rate go higher. This will clearly be an
[15:42] indication of some cancellations of
[15:44] orders. If you start seeing a change in
[15:48] the earnings for any of the seat-based
[15:50] software names, I I think they're going
[15:51] to get hit hit hard because once it stop
[15:54] starts, I don't see a chance that it
[15:56] stops. We've had only good news in
[15:58] software from an earnings basis. If we
[16:00] start to see bad news, which would make
[16:02] sense since Anthropic only saw the
[16:04] revenue run rate accelerate dramatically
[16:07] in January, February, and March. So, if
[16:08] that's what we're going to see, I think
[16:11] there's going to be an issue. And just
[16:12] remember, the returns for the S&P are
[16:14] great when the CPI is below four.
[16:17] They're bad when they're above four.
[16:19] Doesn't mean the history is going to
[16:20] play out again, but I think that
[16:22] differential and that history, when you
[16:24] don't know what the Fed is going to do,
[16:26] you have to think that negative news and
[16:28] all the issues that come with the debt
[16:29] is going to play an impact if inflation
[16:31] does in fact trade above four. Now,
[16:35] we got the U Mich number and Anthony
[16:38] Pompliano and I talked about this and
[16:40] debated about this this week. The main
[16:42] reason I want to show this is forget all
[16:44] the political bias in these things.
[16:46] There's no way to refute that inflation
[16:48] is a problem for people. Um people are
[16:50] living paycheck to paycheck and even
[16:52] though they have jobs, they're worried
[16:53] about their jobs because of AI. And now
[16:55] they've got gas at the pump moving
[16:57] higher. The question is, will inflation
[16:59] come right back down? I don't think so.
[17:01] I think we are going to see the plastics
[17:03] impact, the fertilizer impact, a lot of
[17:05] things where input costs are going to go
[17:07] higher where they're going to have to
[17:08] raise them or profit margins are coming
[17:09] down.
[17:11] Uh and I just think that at this point,
[17:12] unlike the tariff situation of last
[17:14] year,
[17:15] this one is a little bit more difficult
[17:17] because it is the commodity input price
[17:19] side.
[17:20] And especially for things like phones
[17:22] and computers where we have severe
[17:24] shortages at this point because we don't
[17:26] have enough memory.
[17:27] In the Beige Book this week, it was very
[17:30] clear that input cost pressures beyond
[17:32] energy related increases were also
[17:33] widespread. Technology costs rose for
[17:36] both hardware and software. Again,
[17:38] semiconductors are in a every appliance,
[17:40] everything that you purchase almost
[17:43] you know at this stage that is electric
[17:46] in it and semiconductor prices are all
[17:48] rising at this point. Here's the tail
[17:51] that we have to watch out on a global
[17:52] basis. Europe has maybe 6 weeks of jet
[17:54] fuel left. Despite all the happiness and
[17:57] kind of people getting excited about the
[17:59] street reopening, remember we already
[18:02] have we've used up the inventories.
[18:04] There can't be any any
[18:06] you know movement backwards and at this
[18:08] point we haven't seen any factual
[18:10] information. Everything is on
[18:11] speculation. So I think now that the
[18:13] force positioning which is completely
[18:15] driven by the trend in the market
[18:17] through the CTAs. We've seen their
[18:19] buying and we're going into earnings
[18:20] period where the blackout is in,
[18:22] I think we're probably going to reach a
[18:23] top here at some point and then we'll
[18:25] start to go through another
[18:27] at least consolidation.
[18:29] Qatar warns of huge economic fallout
[18:31] from the war in coming months. I'm only
[18:33] showing these just to highlight that
[18:35] everyone who is a quote-unquote energy
[18:37] expert,
[18:39] they've obviously been way too bearish.
[18:42] They've been wrong about stock prices
[18:44] going higher, but it doesn't mean
[18:45] they're going to be wrong in terms of
[18:47] the impact that things are going to have
[18:48] going forward
[18:50] on at least getting inflation above 4%
[18:52] and we'll see what happens when we get
[18:53] there. I think people are extrapolating
[18:55] now that the same thing as last year is
[18:57] going to happen. Like I said, not just
[18:58] on the earnings front, but probably on
[19:00] the inflation not coming through. Last
[19:02] year we had inflation going down the
[19:03] entire year. We had the Fed cutting
[19:05] rates. Right now we don't have either of
[19:06] those and oil has reset as I've shown.
[19:09] Even longer dated oils reset 25% higher.
[19:13] Um,
[19:14] there is no old normal to go back to.
[19:15] We've lost over 1
[19:19] billion barrels of production.
[19:22] Stocks are ignoring it with JP Morgan.
[19:24] I'm just showing you these to show you
[19:26] the breath of it. Um, on the import
[19:28] price side, again, I'm going to
[19:30] highlight that the semiconductor side is
[19:32] a major, major story that I don't think
[19:35] people realize.
[19:37] We didn't have this in the 1970s in
[19:39] terms of the importance of
[19:40] semiconductors, but semiconductors are
[19:42] very important in this rise has only
[19:44] recently started. I'll go through and
[19:46] show you. We've got other places around,
[19:48] you know, the world where material
[19:50] shortages are starting to squeeze the
[19:52] global supply chain.
[19:54] Petrochemicals in China.
[19:57] Um, here are the import and export
[19:59] inflation numbers for Korea this week.
[20:01] And again, Korea is a major provider of
[20:04] memory. And what you have here is just
[20:07] shocking numbers. This is not just in
[20:10] imports. So, this is not just oil. This
[20:12] is imports and exports. Um, I mean,
[20:14] these are explosive numbers that are not
[20:16] 100% driven by oil, uh, but it is there
[20:20] and they warned about chipflation. This
[20:22] is from Korea.
[20:24] AI-driven black holes absorbing memory
[20:26] semiconductors causing prices for chips
[20:28] and electronics like PCs, smartphones,
[20:30] and home appliances to surge, a
[20:32] phenomena dubbed chipflation.
[20:35] AI is toppling everything. This is the
[20:37] supersonic tsunami reference. And
[20:39] remember, we are only 3 months into
[20:42] this.
[20:43] 3 months. We will not have CPUs for
[20:46] quite some time. And so, you're going to
[20:48] have to deal with this. Some mobile
[20:49] phone brands are coping with cost
[20:51] pressures by adjusting product prices
[20:53] and configurations.
[20:54] The uptrend in semiconductor prices may
[20:56] continually potentially suppressing
[20:58] demand for consumer electronics. As I go
[21:00] through the rest of this, I'm going to
[21:03] get into, uh,
[21:04] just what Elon has talked about
[21:07] regarding this. Uh this is translated
[21:09] from a China paper, uh semiconductor
[21:11] price increases affect multiple
[21:13] industries.
[21:15] Uh Gerard at 22V, uh the problem with
[21:19] inflation is not that it should be
[21:21] should be accelerating. The problem is
[21:22] that it has been. We already had prices
[21:25] going higher before Iran. And that's one
[21:28] of the fears that people should have in
[21:30] just dealing with the 1970s where when
[21:33] inflation is high and we don't know how
[21:34] high it's going to go. I happen to think
[21:37] it'll be above 4% relatively soon. If we
[21:40] do get above 4%, the question will be
[21:42] will people start to worry that there's
[21:44] no stopping this. If oil prices go back
[21:47] up towards 100, then you're going to see
[21:48] the inflation pressures remain and if
[21:51] it's because of inventory and supply
[21:54] problems which are continuing to push
[21:56] plastic prices, this is all going to
[21:58] flow through over the course of the
[21:59] year. So, you just have to be ready for
[22:01] it uh at this point. Uh Goldman took up
[22:04] their inflation forecast.
[22:07] And here's the thing and I think Luke
[22:09] Gromen, uh you know, did a good job of
[22:12] just kind of bringing this point up. So,
[22:15] John Roukis highlighted this in his
[22:16] recent technical work. The trend in in
[22:19] yields is higher right now. And we've
[22:21] been consolidating during a period where
[22:23] inflation went down. The debt situation
[22:26] in the US has not gotten better or
[22:27] around the globe. So, really we're left
[22:30] with the exact same situation of the
[22:32] fears that came back here. The only
[22:33] thing that's changed is short-term rates
[22:35] have been taken down by most central
[22:37] banks.
[22:38] But the issue is are we starting to see
[22:41] inflation and will
[22:44] countries have to react or will they not
[22:47] be able to react because of the debt
[22:49] situation? And I think that's the
[22:51] problem that comes in is if you start to
[22:53] see bond volatility move up and Henry
[22:56] Paulson this week talked about that we
[22:57] should prepare for a vicious bond crash.
[22:59] I I the bond market is the place now to
[23:01] watch and I've said for a while that
[23:03] it's a distraction, but that was during
[23:05] a time where yields were going lower, AI
[23:07] was accelerating,
[23:09] and it was accelerating a
[23:10] non-inflationary basis.
[23:13] Everything changed this year and that's
[23:15] the thing I want to make sure you guys
[23:16] know. Um the reason for the tone change
[23:19] and the reason that this year is
[23:20] different is there is tremendously new
[23:22] information we have that have hit. We
[23:24] ran out of memory this year. We've run
[23:26] out of CPUs this year. We already saw
[23:28] the inflation process moving higher
[23:31] before Ron. Now we have oil set at
[23:33] higher levels. We saw silver prices
[23:35] rally dramatically from the summer till
[23:37] now, all when DRAM did. We already have
[23:40] an inflationary problem and as I go
[23:42] through this, I'm going to say it again
[23:44] and again. The physical constraints of
[23:46] the world have been met and compute
[23:48] shortages are showing up everywhere.
[23:50] This is not a minor story and something
[23:52] that needs to be paid attention to. At
[23:54] the same time, mythos is going to be
[23:57] released, which means hackings are going
[23:59] to happen. We're leaving the sweet spot
[24:02] sweet spot of just getting IQ higher,
[24:06] and now we're getting into the real
[24:07] dangers that come from AI and the world
[24:09] is not ready for them both from a power
[24:11] basis, a semiconductor basis, a memory
[24:14] basis, a gas turbine basis, a
[24:16] transformer basis. I can go on and on.
[24:18] The world AI is moving faster
[24:21] than the physical world can keep up with
[24:23] it, which is still linear and is under
[24:25] invested. That's why you want to be
[24:26] continuing to invest in those areas
[24:29] and focus on the places where you can
[24:31] make some money. At the same time, when
[24:34] we have the debt problem and we have the
[24:35] governments and they got to figure a way
[24:37] to pay for it and you're getting more
[24:38] and more stories of wealth taxes.
[24:42] Um obviously California's been talking
[24:44] about it and going through this already,
[24:46] but then this week you had New York
[24:48] going through luxury second homes.
[24:50] You've got a story growing in a property
[24:53] tax hike 50% in South Hadley,
[24:56] Massachusetts. Canary in the coal mine.
[24:59] This is not a US problem. Swiss back a
[25:01] new wealth tax on on the rich in
[25:04] Switzerland.
[25:05] So,
[25:07] I think with all of those problems, this
[25:09] is the one that uh again, if I'm right,
[25:12] we're going to have inflation. And
[25:13] again, I'm using Thomas Malthus for for
[25:16] the paper I put out on on Thursday. Uh
[25:18] again, subscribers saw this. Um
[25:22] We have a token shortage, and that was
[25:24] the point of it. And what I really did
[25:25] was write a paper based on what I heard
[25:28] from Dwarakesh Patel with Jensen Huang.
[25:30] If you haven't listened to it, it's
[25:32] phenomenal. I will help you guys with
[25:34] that in a little bit, but Dylan Patel
[25:37] gave an interview on here about CPUs,
[25:39] about reinforcement uh learning
[25:41] environments, and AI-driven workloads,
[25:43] and just how this explosion came because
[25:45] AI agents came faster than what everyone
[25:47] explained. Um it has led to, you know,
[25:50] why neo clouds emerge as credible
[25:52] competitors. We basically have such a
[25:53] big shortage. Uh CPUs have become a
[25:56] major bottleneck.
[25:58] His bottom line is that in a true AI
[26:01] gold rush, almost any decent chip can
[26:03] find demand. We are out of them, guys,
[26:06] and we just started the agentic world.
[26:08] How do you catch up?
[26:10] Not easy. And again, Jensen Huang will
[26:12] teach you.
[26:13] Chips are one thing, but we need power
[26:15] as well. Uh his Blackwell and Vera Rubin
[26:18] are become are making it more efficient
[26:20] where he needs less power. That's what
[26:21] the tokens per watt is.
[26:24] But Anthropic doesn't have anything
[26:26] really in any big way with uh
[26:29] Nvidia, so
[26:31] we're using so much AI that computing
[26:32] firepower is running out. I've run into
[26:35] problems continuously with Anthro with
[26:37] Claude this week. Uh before I did this
[26:40] on some of the things I was running, I
[26:42] literally ran out of it. Uh the irony is
[26:45] we don't hear much about the AI
[26:47] overbuild anymore. You've got CoreWeave
[26:49] CDS coming down. I did the piece on
[26:51] Oracle, which basically said that we've
[26:53] run out of compute. Oracle's is not a
[26:55] software company. It's been hammered
[26:57] because we were worried that they'd get
[26:58] the revenue in. But now we have an AI
[27:01] compute shortage and they're a compute
[27:02] company. So the question is
[27:05] I think they're going to be fine because
[27:07] the demand side is just going through
[27:09] the roof. Um Brad Gerstner talked about
[27:12] it.
[27:13] Dylan Patel talked about it. Here's
[27:15] what's happening to the data centers in
[27:17] terms of the expectations. I'll go
[27:18] through that. I've done a ton of work on
[27:20] that, uh which again I will upload to
[27:23] the uh subscriber site if people are
[27:25] interested in doing their own data
[27:27] center analysis. I'm happy to share the
[27:29] work that I do with uh
[27:31] uh with some of the institutional
[27:32] clients. Uh it's it's worth looking at.
[27:36] At least you don't have to read the news
[27:37] and you can go do your stuff on your
[27:38] own. And for those uh institutions that
[27:40] follow the data center, uh
[27:43] this will hopefully be useful for you as
[27:45] well. The AI compute crunch is becoming
[27:47] a real constraint. Claude has had 98.95%
[27:51] uptime over the last 90 days. Company
[27:54] began rationing tokens during peak hours
[27:56] in late March. OpenAI had to get rid of
[27:58] Sora. CoreWeave raised their prices. The
[28:01] prices for Nvidia's advanced chips were
[28:04] up 48% in 2 months. You can see the
[28:06] whole thing and
[28:08] their revenue has gone from 9 billion
[28:10] annualized to 30 billion annualized. The
[28:12] question is, will that continue to grow?
[28:14] I heard Brad Gerstner say it'll be at
[28:15] 100 billion. Well, they better get
[28:17] enough compute and they better hope that
[28:19] the companies don't push back on the
[28:21] price rise. Uh
[28:23] Claude uptime has been terrible. Doesn't
[28:25] work at all right now.
[28:27] I can agree that it doesn't work
[28:29] anything like it used to. So think about
[28:31] all the companies that are now paying
[28:34] Anthropic.
[28:36] Uh
[28:36] and it's getting harder and harder for
[28:38] them to provide good service. That is
[28:40] just not a good situation.
[28:42] Taiwan
[28:44] still out of NAND.
[28:47] Memory contract negotiations suggest
[28:49] DRAM contract contract prices could rise
[28:52] another 60% quarter over quarter while
[28:54] NAND could increase again, memory is
[28:57] just nowhere.
[28:58] The whole rack, remember I wrote this
[29:00] paper, um these names have been flying.
[29:03] I think this is the key paper to go back
[29:05] to and to go read for those of
[29:06] subscribers.
[29:08] Um
[29:09] again, I highlighted back six weeks ago
[29:12] that CPU vendors were number one on the
[29:14] list. I mean, at the time, Intel was
[29:16] $42, I believe. It's now almost 70.
[29:20] Uh the CPU shortage is severe.
[29:23] That comes from Amazon.
[29:25] You would need at least twice as many
[29:27] CPUs for what's happening. So, this is
[29:29] the problem. The agentic world came out
[29:32] and the and it and accelerated the usage
[29:34] and the amount of CPUs that are needed
[29:37] was far greater than what's available.
[29:38] Unlike memory, CPUs are currently
[29:40] unavailable, even if you're willing to
[29:42] pay a premium.
[29:45] PC makers serving enterprise clients are
[29:47] particularly affected. Intel CPUs are
[29:49] reported genuinely unavailable. With
[29:51] Intel estimated to hold roughly 90%
[29:53] market share in industrial segments, the
[29:55] impact on IPC makers is pronounced.
[29:59] Intel's PC business survival hinges
[30:03] on [laughter] the shortages.
[30:05] You can't buy a Mac mini. You can't buy
[30:07] a Mac Studio. Remember when I told you
[30:08] guys to go out and get them before they
[30:10] they ran out? Well, they're gone. Now,
[30:12] remember, when you're looking for
[30:14] signal,
[30:16] the main point of this is a year ago,
[30:19] okay? This is as of
[30:22] one year less than one year ago. This is
[30:24] September of last year. Intel was
[30:26] trading $25.
[30:28] So, one year ago, $25.
[30:32] There were stories in September, five
[30:34] years after Apple broke up with Intel,
[30:36] Intel is begging for money.
[30:39] The government gave money. Here's what's
[30:41] happened since then.
[30:42] So again, you've never seen anything
[30:45] like this. Here are Intel's returns from
[30:46] 2001.
[30:48] Look at this parabolic move for
[30:50] something that is an old industrial
[30:52] boring company that was thought to be
[30:55] out of work out of business. That's how
[30:57] quickly AI goes. That is the enormous
[31:00] size of the CapEx
[31:02] and the agentic flows have just started,
[31:04] guys. Again, just started. We're talking
[31:07] about somehow or another
[31:09] the run rate for Anthropic to get up to
[31:12] a hundred billion in by December,
[31:15] according to Brad Gerstner.
[31:18] You're going to have to have a lot more
[31:19] compute for that.
[31:22] Anthropic is changing the way they're
[31:24] they build their enterprise clients.
[31:25] They will begin billing them on token
[31:27] usage. This is another thing that has
[31:29] shown up. So here's what the CTO of Uber
[31:33] said.
[31:35] Coding tools, particularly Anthropic
[31:37] Claude code, has already maxed out its
[31:38] 2026 AI budget.
[31:41] This is Uber.
[31:42] I'm back to the drawing board because
[31:44] the budget I thought I would need is
[31:45] blown away already.
[31:49] Companies are overrunning their initial
[31:51] budgets for inference. This is from
[31:52] Goldman Sachs. By orders of magnitude,
[31:54] we heard one industry data point on
[31:56] interest cost and engineering now
[31:57] approaching about 10% head count head
[32:00] count cost, but could be on track to be
[32:01] on par with head count cost in the next
[32:04] several quarters.
[32:06] So think about what that's saying. It's
[32:08] based on the token usage, getting rid of
[32:10] people may not be an option. That is not
[32:13] good for the model companies if that's
[32:14] the case. If they are pricing people
[32:16] out, the only way to pay for this is
[32:19] somehow or another to pay it through
[32:20] labor.
[32:22] That we're we're reaching some some
[32:24] issues here.
[32:25] The electrical component shortage is
[32:27] delaying data center construction. The
[32:29] Iran war push pushing up energy cost
[32:31] prices. The hyperscalers pivoting to gas
[32:33] plants because the grid connections
[32:34] aren't available. All of that is now
[32:35] showing up as a compute crunch affecting
[32:38] actual products that paying customers
[32:40] depend on. Demand is tripling faster
[32:42] than infrastructure can be built and the
[32:44] only short-term lever companies have is
[32:46] to raise prices. But raising prices in a
[32:48] market where you're still trying to gain
[32:50] users and already losing billions is a
[32:52] generally difficult trade. This is where
[32:54] the model companies and again, the
[32:55] hyperscalers, even in Anthropic, this is
[32:58] an issue. Demand is going too fast and
[33:00] we can't slow it down. 98.9% uptime is
[33:04] not a foundation you can build a
[33:05] business on. Anthropic's revenue
[33:07] tripling in 4 months while its
[33:08] reliability is declining is not a
[33:10] sustainable combination and that tension
[33:12] is going to get harder to manage the
[33:14] faster demand grows. This is from
[33:16] Hedgie.
[33:19] All right.
[33:20] Uh
[33:21] you have to read about Terafab. Um so,
[33:23] with all that going on, the question is,
[33:25] is it legitimate? Are we really out of
[33:27] chips? Is the demand so big that it
[33:29] won't be fixed or are we going to have a
[33:31] glut of chips, which is what most people
[33:33] tend to go. So, on this podcast, the
[33:36] Limitless podcast, they went through the
[33:37] details from Terafab. Um you can get
[33:40] them other places, but I just like to
[33:42] pick a podcast and go through.
[33:45] Elon Musk says, we won't be able to keep
[33:47] up with demand. More precisely, he says,
[33:49] current chip supply and current fab
[33:50] expansion rates are nowhere near enough
[33:53] for the compute scale he wants. He says,
[33:55] current AI global compute output is
[33:57] roughly 20 gigapots gigawatts per year
[34:00] and all fabs on Earth combined are only
[34:02] about 2% of what is needed.
[34:05] He says that they appreciate suppliers,
[34:08] the three or well, Taiwan Semi, Samsung,
[34:11] Micron would buy he would buy all the
[34:13] chips they can make, but their expansion
[34:15] pace is much slower than what he wants.
[34:17] Now, the reason this is important is
[34:19] Elon is talking about edge devices. He's
[34:21] talking about autonomous vehicle demand.
[34:23] He's talking about humanoid demand. He's
[34:25] talking about space
[34:27] demand.
[34:28] He has unlimited demand of chips. He
[34:30] would buy all of them. So, we already
[34:32] have a shortage and we have these three
[34:35] coming AI trends on the physical world.
[34:38] So, you literally have to go through and
[34:40] read what he's saying.
[34:42] He's building
[34:44] his own fab, a massive thing the size of
[34:47] San Francisco, as a response to a
[34:49] structural chip shortage relative to his
[34:51] future compute ambitions. He's using the
[34:55] money he's raising
[34:57] to do this.
[34:58] His Optimus needs.
[35:02] So, again, the chips are for cars and
[35:03] especially Optimus. Global vehicle
[35:05] production is about 100 million per
[35:07] year. Humanoid robot production could be
[35:09] 1 to 10 billion units per year.
[35:12] Regardless of whether you believe Elon's
[35:14] numbers,
[35:16] he is the master of knowing where things
[35:19] should be and he's basically starting
[35:22] this process and moving at hyper-speed.
[35:25] Musk asked suppliers to move at light
[35:27] speed in new chip making plan.
[35:30] The project will require something like
[35:32] 5 to 10 trillion, 5 to 13 trillion in
[35:35] capital spending.
[35:37] 5 to 13 trillion just for what he's
[35:39] talking about.
[35:41] Why compute scarcity is the new macro
[35:43] reality. So, I wrote this uh earlier
[35:45] this week. You guys saw it.
[35:49] And within there, just to give you an
[35:51] idea, I compared I looked at the edge
[35:53] systems and went to the scarcity and
[35:55] mapped out where everything will be,
[35:57] what the investment strategy should be.
[36:00] And all of this fits in my edge
[36:01] investment AI investment universe. So, I
[36:04] have a lot of details on this. Uh I put
[36:06] together a report on this. It is
[36:09] uploaded to the subscriber page already
[36:11] in the thematic ideas. It is 68
[36:14] companies. Many of them are there, but I
[36:16] also have a heat matrix, which I think
[36:18] is the most important thing, but also
[36:20] each of the verticals, so you can go off
[36:22] on your own. So, this is how many
[36:24] companies I have per vertical, all of
[36:27] the different component pieces that make
[36:29] this up. This was done from a Santa Fe
[36:31] Institute perspective, the way that I
[36:33] like to go through it to basically say,
[36:34] "We need to be here in 5 years according
[36:36] to Elon. He's going to build a fab. What
[36:39] are the places that he needs to fit fill
[36:41] this out so we can invest in these
[36:43] things?" The fact that most of these are
[36:46] already companies on the model portfolio
[36:48] list should give you comfort that this
[36:50] is not a short-term trend. This is a
[36:51] long-term trend for at least the next 2
[36:53] to 3 years. I expect the model portfolio
[36:55] unless there is a significant problem
[36:57] with rates or something that uh
[36:59] really kills the company's ability to do
[37:02] this, which would be bad for the market
[37:03] as a whole. The spending is going to
[37:05] happen. TSMC on their earnings report
[37:07] said their CapEx over the next 3 years
[37:09] will be significantly higher than the
[37:11] last 3 years when it's spent 101
[37:13] billion. Again, when we talk about the
[37:15] CapEx numbers, we're usually only
[37:16] talking about the seven hyperscalers.
[37:19] The fact is between OpenAI, SpaceX, what
[37:21] he's doing right now, I don't know how
[37:24] Terra fab will get included. XAI gets
[37:25] included. These numbers are getting
[37:27] bigger by the second because we're
[37:29] getting closer to the chip the the chip
[37:31] shortage, which will stop AI, and we
[37:33] need the power side, which would stop
[37:35] AI. Great interview. Um
[37:38] I have a tremendous amount of respect
[37:40] for Jensen Huang. I always did, but I
[37:43] think it's very hard to listen to this
[37:44] and not hear a guy who knows his
[37:45] business inside and out, knows the AI
[37:47] world inside and out, gives you alpha
[37:49] every single time he speaks. Uh the
[37:52] reason this one was so good, the reason
[37:53] it's gone viral, Dwarkesh is clearly on
[37:56] the uh
[37:58] Anthropic uh AGI will destroy the world
[38:01] side, and Jensen basically debated him
[38:04] on this thing. So, it was kind of like a
[38:06] political debate and whether Dwarkesh
[38:08] believes that or was just playing
[38:09] devil's advocate for the the beauty. It
[38:12] was a great interview, especially the
[38:13] last 45 minutes, which was a debate on
[38:16] China and whether they should have
[38:17] chips. I highly recommend it. It was
[38:20] such a good interview that what I'm
[38:21] doing this week. So, for the
[38:22] subscribers, you guys get this in in
[38:25] email every single Sunday at 9:00. The
[38:28] video podcast summary where I go through
[38:30] all of the videos, the Dylan Patel one,
[38:32] the Elon Musk one, and the Jensen one.
[38:34] You get this format. This week I'm
[38:36] including the Claude skill prompt. It
[38:38] shows you how to set it up. Go into your
[38:41] Claude,
[38:42] set up the skill using what I gave you,
[38:44] and then run it on the transcript which
[38:46] you can get at these sites. You can just
[38:48] do what I basically did, but I highly
[38:50] recommend going. It will give you some
[38:52] alpha creation. You can go through and
[38:54] get your own names and do your own
[38:57] uh rabbit hole learning.
[38:59] I'm trying to get you guys more of this.
[39:00] Uh particularly the highest end
[39:02] subscribers, I will be giving you more
[39:03] videos uh on learning. Uh the main goal
[39:07] is to get you to not only have a model
[39:09] portfolio with changes going on. This is
[39:11] why I'm giving you the technical
[39:12] analysis tool uh as well as an earnings
[39:15] commentary tool, but I think most
[39:17] importantly in this side is going to be
[39:19] the ability to learn AI and do some of
[39:21] this stuff on your own. Not just for
[39:23] what I'm doing, but also for your
[39:24] business. Particularly for the RIAs and
[39:27] the FAs where I've been developing a lot
[39:30] better connections, and I want to do
[39:32] presentations uh not just for me doing
[39:34] them, but Anthony Pompliano and I have
[39:36] committed to kind of serving you guys as
[39:38] well. And if you want us to come uh
[39:40] speak at some of your events to talk
[39:43] about getting people into Bitcoin, which
[39:45] I think is going to be a major story by
[39:46] the end of the year. When you guys
[39:48] decide that you're getting enough of
[39:50] that, you just let us know. Um here are
[39:53] the key themes from the Jensen side. And
[39:55] again,
[39:56] why it matters, what you can get from
[39:59] it. You guys can go through this. This
[40:00] is the type of stuff you can do with the
[40:02] Claude prompt skill. Uh it will give you
[40:04] a report, but you can kind of learn how
[40:06] you can drill down in your own favor. He
[40:08] really mentioned energy a lot.
[40:11] Uh
[40:11] this is where Jensen and Musk converge
[40:14] hardest.
[40:15] They agree that energy is the downstream
[40:18] bottleneck that matters most. This is
[40:20] why
[40:21] you need to be invested in the power
[40:23] theme. The energy names have come off
[40:25] because of oil. That is a mistake, guys.
[40:28] You want to be investing in power. All
[40:30] of the power theme names you should be
[40:32] looking at. Within the interview, Jensen
[40:35] mentioned all of these companies. Uh if
[40:38] you notice, Cadence Design Systems,
[40:40] Synopsys,
[40:42] I mentioned those uh this week on the
[40:44] call.
[40:45] Uh I highly recommend spending time
[40:47] looking at those names and getting my
[40:49] viewpoint on them.
[40:51] And then here are the big money lines
[40:52] from Jensen in there. The number of
[40:54] agents is going to grow exponentially.
[40:55] The software companies are going to
[40:57] skyrocket.
[40:58] We're pre-fetching the bottlenecks. He's
[41:00] getting ahead of everything and he's
[41:02] going to have the ability to continue to
[41:04] produce. I bought some more Nvidia this
[41:07] week. Uh I believe we are at a breakout
[41:09] level for Nvidia and I think the
[41:11] earnings are going to be the cause of
[41:12] it. Uh on the data center side, this has
[41:15] become more of a story and so I want to
[41:17] make sure you guys are getting more on
[41:19] top of it because the data centers are
[41:21] being delayed and it's being reported
[41:23] more and more.
[41:25] Uh you're getting more of this where
[41:27] because the demand is accelerating and
[41:29] that's what happens. When you have more
[41:30] tokens, you need more electricity. Don't
[41:33] be surprised if we start to hit kind of
[41:35] the physical constraint. So all of those
[41:37] power names, everything in there. You
[41:39] saw a deal with Oracle and and Bloomberg
[41:41] energy uh
[41:42] Bloom Energy. Um you're going to see
[41:44] more and more solutions. Uh the battery
[41:46] names were up big this week. Batteries
[41:48] are going to have to be a solution. So
[41:49] whether it was EOS or Fluence Fluence,
[41:52] they're inside the power basket. They've
[41:53] underperformed. You want to pay
[41:55] attention in these names because the
[41:56] battery situation is going to have to be
[41:58] a major investment. Tesla, same thing,
[42:00] has underperformed. I still love Tesla.
[42:02] It's not part of the model model
[42:04] portfolio right now. Uh but I will say
[42:06] that I'm looking for the next catalyst
[42:08] in terms of when to again from a trading
[42:10] perspective get more involved. So I'm
[42:13] sure you'll see something on that. Uh Um
[42:17] for 2027, aggregate hyperscaler CapEx is
[42:19] expected to materially exceed 26. Think
[42:23] about that. We just raised the numbers
[42:24] for 26, and we're talking about 27
[42:26] needing a raise. The data center delays
[42:28] threaten to choke AI expansion. Now, for
[42:31] those of you who are directly interested
[42:33] in this, I am going to put this up uh
[42:37] either on Sunday or Monday. Uh you'll be
[42:39] able to get this from the thematic ideas
[42:41] area. So, for the subscribers that want
[42:43] it, again, this has every single data
[42:46] center that I can find on there in terms
[42:49] of
[42:50] what they're supposed to do in
[42:52] megawatts, and then I go through the
[42:56] problems, what the percentage is. You
[42:58] can see the numbers at risk. Uh
[43:01] all of the things, if you want to go
[43:03] invest in in the areas like the
[43:05] transmission side. But, based on my
[43:08] model, the cleanest answer is about 80%
[43:10] of the future US data center pipe
[43:13] pipeline is delayed or at risk of delay
[43:15] by 2030. Now, this doesn't mean they
[43:17] won't get done. It just means
[43:19] there isn't almost a single data center
[43:21] that doesn't have delays at this point.
[43:23] So, you're left in that situation.
[43:25] Again, compute shortage, I want to
[43:27] convert this in the final four slides
[43:29] back to Bitcoin because we did see the
[43:31] miners move higher. This is an a chart
[43:34] of the Bitcoin miners, Bitcoin,
[43:38] which again did start to break out.
[43:40] That's the orange line.
[43:42] Circle, which has moved higher, and
[43:44] Coreweave. So, all of the compute names,
[43:47] the miners, Coreweave, you're starting
[43:49] to get anyone that has any kind of
[43:51] power, their value's going up. So,
[43:53] remember, on Bitcoin, we do have a
[43:55] weekly MACD signal. Um
[43:58] and got up Friday toward 78. I'm sure
[44:01] we're going to do some back and filling,
[44:03] but I fully expect as we get into
[44:04] negative real rates that Bitcoin will
[44:06] become a major story. Here is Ethereum,
[44:08] which not only broke a trend line, but
[44:10] also took out the highs from March. It
[44:12] also has a weekly buy signal. We'll see
[44:13] how that goes.
[44:15] And then finally, like I said, remember
[44:18] negative real rates are coming, in my
[44:20] opinion. Uh that'll be the stage
[44:24] where the Fed has to make a decision if
[44:25] inflation's up to 4% or 5%.
[44:29] And I think this is going to be where
[44:30] people realize that the government has
[44:32] an issue, uh and the issue is what do we
[44:35] do? And I think Bitcoin will be the
[44:36] place that people go. That's it for this
[44:39] week, guys. Again, thanks to the
[44:40] subscribers. Um thanks for watching
[44:43] this. Please subscribe. Please send it
[44:45] to friends. And if you are global and
[44:48] you are interested in having this
[44:50] converted into a language for you, uh
[44:52] reach out to me directly at jordy@visser
[44:56] da- hyphen labs.com.
[44:59] So, you can see it. Uh I'll respond to
[45:01] you and we can go through it that way. I
[45:03] have a bunch of people reaching out
[45:04] already. Love to hear from you. Thanks.
[45:06] Bye.