Listen on Apple Podcasts
Listen on Spotify
Economy & Markets
Eye On The Market podcast
Subscribe today and never miss an episode
Join Michael Cembalest, Chairman of Market and Investment Strategy, as he explores a wide variety of investment topics, including the economy, policy and markets. Your subscription, using your preferred podcast provider, will feature new episodes with the release of each Eye on the Market publication.
Summer mailbag
Welcome to the July on the Market podcast. This one’s called Summer Mailbag because I’m going to answer some of the client questions that we’ve been receiving recently. First, one quick comment. As many of you may know, we issued our 20th anniversary Eye on the Market a few weeks ago. And, we have some printed booklets. If you find those easier to read for a document that fast, it’s that long.
And so feel free to call your coverage team if you want to get a printed booklet of your own, dog not being included. Let’s see. So and quickly, a preview of the August Eye on in the Market, which is going to be a health care piece that I’ve been working on for the last few weeks. And, I want to show you a couple of charts here that are just kind of amazing.
For three decades, almost starting in 1989, a couple of years after I joined J.P. Morgan for three decades, until 2019, the health care sector was every bit the growth sector. The tech was and, tracked the returns remarkably almost identical over those three decades and with less volatility. Health care vol was about 15 versus technology closer to 25 over this 30 year period.
Ever since 2019, the health care sector has been really sick. And as you see here, from 2019 to 2025, technology has continued its march and health care is more or less flat, or almost flat over 6 or 7 years. So we’re going to take a look at all the things that are going on from proposals on capping drug prices to cuts to the CDC and the NIH, slower approval of new drugs, the patent cliff, and all the things that are associated with what’s dragging down health care.
We’re going to do that next, in the August Eye on the Market. So question number one that I've been getting, why is there such a long backlog for natural gas turbines? And do their higher prices mean that solar and storage is comparable economically as a form of baseload power? Well, you may have seen this. There was this, Pennsylvania Energy and Innovation Summit a few weeks ago and a bunch of companies, including Google and Core Wave, Blackstone, Westinghouse Energy Capital Partners announced 90 billion of intended data center and datacenter expansion.
And they also talked about the need for the power infrastructure to support all those data centers. And a lot of what they announced is intended to be powered by new gas turbines. The problem is supplies are pretty tight. Turbine delivery times, at least for combined cycle turbines rather than the combustion turbines. Can read 3 to 7 years.
And just over the last couple of years, their prices have more than doubled, from $1200 per kilowatt to around $2,500 a kilowatt. And some of the CEOs in the sector are also citing a lot of labor shortages and the skilled labor pools required to build these gas turbine project. And so, it's interesting to see just how much of a backlog there is now.
For many years, most parts of the United States and other places around the world have been giving grid preference to renewables. So there hasn’t been really that much gas natural gas infrastructure bill. And we wanted to take a look closer look at this, this chart shows the gas turbine capacity built over the last seven years or so, as a percentage of all the gas capacity operating in the state.
So, for example, in Texas, only 3% of all the gas capacity they’ve got have been built since 2019. And in California, that number is 9%. And in New York and North Carolina it’s around 10%. So only in Illinois have you seen kind of a meaningful expansion and maybe to a lesser extent in Ohio and Pennsylvania of natural gas capacity recently.
So there is a lot of demand. There hasn’t been that much built recently. And understandably, Siemens and GE and Mitsubishi, which are the three large turbine manufacturers are taking advantage of higher demand, higher prices, but are not meaningfully expanding their production capacity, at least not yet. So that these delays will be with us for a while.
And as I mentioned there already, the these delays have already pushed up prices. So I thought it was interesting last week, one of the big research institutes on energy wrote a paper basically saying we now are in an affordable era where solar plus storage can carry almost the entire load. And they cited 97% of the load in a place like Las Vegas.
Okay. So, you know, I remember that old Reagan saying trust but verify. And whenever anybody writes an energy paper, that’s exactly what I do. I will I will trust that they did it the right way. But then I’m going to verify every single calculation. And so we have an hourly grid model of our own. And we can easily set it to a place like Las Vegas for purposes of irradiation and hourly solar generation, things like that.
And if you look at this chart that we have in Eye on the Market and there which was the entity that made these forecast, that orange line at the bottom of the chart, that’s their estimate of different grid configurations for solar. Look at the end of the chart. So they’ve got something where 90 plus percent of the solar can come from photovoltaic solar, a large amount of backup battery storage, and then a rather small backup thermal system to carry the rest of the load.
And they get a number in the neighborhood of $105 a megawatt hour. Some of their assumptions we agreed with, a bunch we didn’t. And so when all the rest of the curves on this chart represent other iterations based on our cost, real world loads, different operating lives, different gas prices and things like that. And, you know, based on what you’re seeing here, the bottom line is I think, I think we're a long way away on an up on an unsubsidized basis, which has which is how you have to look at these kinds of things.
I think we’re a long way away from when solar and storage are comparable economically as a form baseload power to gas. And, anyway, we’ll talk about this a lot more in the energy piece, but these are the numbers. Okay. So I do think that the, power production is going to be something of a speed bump for a lot of the data center buildout.
All right. Next question. Is China constraining exports of rare earth materials to the US? And what’s the deal with the Department of Defense investing in EMP materials? The answer the first question is yes. At least for now. So this is a chart on Chinese rare earth exports to the US, and they fell by you know, 80, 90% in May.
Now, that’s when China was trying to apply maximum leverage to the United States. Once we get the final data for the rest of the year, particularly after there’s some kind of a deal, presumably they’ll go back up. But I just wanted you to see this chart because, China is already understandably exercising their leverage in these tariff discussions by really choking off rare earth exports to the U.S.
Now, a lot of U.S. entities, whether they’re the private sector or the military, have stockpiled some of these materials, but those stockpiles aren't going to last forever. Now the Pentagon talks about wanting a China free magnet supply chain by 2027. But I think this is a pretty tough goal to meet. China operates in an environment with scant environmental regulations overseeing their industry.
There are 39 Chinese universities with rare earth training programs. The US has none, which is why the China has a 90% share in the processing of these rare earths. There are some new U.S. processing capacity that will come online in 2026, 2028. But, it’s going to take a while. And for example, EMP materials, which I’ll talk about in a second, produces around a thousand metric tons of neodymium, compared to 300,000 metric tons in China.
And those are annual numbers. Pardon me. So that all may explain why the Defense Department became the largest shareholder, which is kind of unusual, but the U.S. Defense Department is now become the largest shareholder in EMP materials, which is a U.S.-based company. They bought about 400 million of its convertible preferred stock and is helping it build out a new processing facility.
And they’ve entered into a ten year agreement that puts into place a price floor for the companies Neodymium and Praseodymium Alloy, and promises for some period to buy 100% of the magnets that they produce. I think this is a good thing, for the United States to finally get serious about this, at the end of 2024, MP Materials, which is a U.S. company, their largest shareholder was a Chinese firm.
That’s who the Department of Defense is essentially replacing here. So it's if these if this build out in the U.S. can be done responsibly, it’s a good thing. I just wanted to also show you on this page a chart of what happened to rare earth oxide prices the last time China, put quotas on, before they took them off, prices rose by a factor of almost ten.
And so, that’s what happens when you got one country that’s got 90% market share on processing and has and has the inclination to apply export quotas. So it’ll be a long road, but at least what the U.S. Department of Defense is doing, it’s the first step. Next question. What are you guys seeing these days regarding the YUCs, the MEGAs and equity market valuations?
A lot of our clients are familiar with our YUC analysis. YUC stands for young unprofitable company. And when this YUC measure, which looks at the share of unprofitable companies as a percentage of market cap when this thing was spiking in 2021, in 2022, if that’s one of the reasons why we became very bearish at the time on anything having to do with SPACs or a lot of metaverse hydrogen, all of those kinds of things, there’s a lot of these young, unprofitable companies at the market all at once, and right now that number has come back down.
It’s not as low as it was from, let's say, around 2005 to 2012, when it was way less than 1% of overall market cap. But it’s down substantially from 2022. You know, the bazooka period for stimulus. And so that’s a good sign. It's one good sign for taking. It's not the only one. But in terms of risk taking, that's a positive to me.
The other positive is if we have to separate large cap and small cap here. The share of large cap companies that have negative earnings, that number is also coming sharply down. It was as high as 20% during 2021 and now is back down to about 7%. That’s close to the lower low levels that it’ been out over the last 30 or 40 years.
In contrast, small cap is an absolute desert of profitability, as it depends how you measure it. But, based on our measurements, more than 40% of all small cap companies have negative earnings, which is one reason why small cap performs has performed so poorly relative to the large or large cap stocks over the last decade. Then what about the MEGAs, which are the, you know, the super large companies?
We’re in uncharted territory, obviously on margin in terms of these companies now represent and have for the last couple of years, around 35% of market cap is just coming from the top ten companies. You know, we can look at this going back to 1990 and that’s a huge number. What makes it almost palatable is that, their earnings represent a similar share of overall market earnings.
Now, again, you have to measure this different ways. But if we look at consensus 12-month forward earnings estimates, these ten companies also represent 35% of the overall market share of the shareholders. So while it’s disturbing to see ten companies, at around a third of the entire market cap, at least their earnings share has grown with it on a trailing basis.
That earnings share is lower on a forward looking basis. That earnings estimate is right on 35%. So obviously a lot depends upon how on whether or not those 12 month earnings estimates come to fruition. But, you know, as a general measure, the top ten here don’t look quite as offside as you might think they are.
And then the last thing is on valuations, we, we like to separate tech and interactive media from the rest of the market. And as you can see here, the tech in interactive media valuations are hovering at that 30 times forward earnings number that they’ve been at really for the last eight years or so. And these are some pretty lofty valuations.
But they’ve been sustainable based on earnings growth. The rest of the market. Right. Which is you know, 450 plus stocks. Their multiples are more or less average when looking back at the last three decades or so, at about 20 times forward earnings, those valuations have been creeping up to be sure. But they don't look quite as offsides as the rest of the market does as much as the top ten do.
All right. Next question. There’s a lot of discussion about I mean, if you think about the broad Trump agenda and the issues around inflation and tariffs and deportations, one of the things that’s kept the business community and people, the CEO roundtable, business roundtable, one of the things that kept them engaged with the administration is the is the prospects for deregulation.
And, we’ve written many, many, many times about how the Biden administration set like a gold record in terms of the pace of government regulation compared to other administration. So, a lot of our CEO clients were happy to see that there would be a swing in the pendulum back in the opposite direction. These things take time, and it's a little bit early to measure the results.
Yet we think in a few months we will have a better sense of it. But one thing that is interesting is a way for the banks, which is what would happen if there were relaxation in bank capital standards. In general, you’d think that’s a good thing if you’re an investor, because banks would be able to buy back more stock and, banks would be able to do more risky lending.
So in principle, reducing bank responsibly, reducing bank capital requirements without going crazy, could be an economic boost. The problem is there are many bank capital standards, and what's being proposed is just relaxing one of them. And the only way this has a big impact is if the one they relax is the binding constraint. And if there isn't another capital constraint right behind it.
And unfortunately, neither of those things is true. And now here’s what I'm saying is the supplementary leverage ratio is the one that is being discussed to be relaxed. And as you can see in the chart that you'll see when you see in the Eye on the Market, it’s either the binding constraint or very close to it. For most of the big G sub AGC banks globally, you know, the systemically important bank.
And so if you if you relax the capital standard that's not binding or if it is binding and there's another one close behind it, it doesn’t really do anything. Here’s a way to visualize that. We have a before and after chart. And the only thing you get from revising these, supplementary leverage ratio rules is a little bit a couple of billion dollars for Goldman, Morgan Stanley, BNY Mellon, State Street.
And it doesn't do anything for J.P. Morgan city, Wells or B of A. It’s kind of a rounding error in terms of increased, capital, tier one capital for the banking industry. So that doesn’t really do much for you. One of the things that the Treasury, I think Bethany has mentioned, this is, well, at least that same relaxation would increase the Treasury buying capacity.
I’m not going to walk through all the mathematical reasons why. But, you know, we’ve done all of the work and we've modeled all of this. And in principle, the banks would have the ability, they currently have the ability to purchase another 3 trillion of Treasury bonds before some rules kick in. Now, if they make this rule change that number will go to 8 trillion.
So in principle, maybe the banks could act as a safety valve if there were an unexpected collapse in foreign demand for Treasury bonds by either foreign official buyers or private sector. Right. The problem is, if the banks are taking advantage of $3 trillion of Treasury buying capacity right now, I don’t know how expanding that number to 8 trillion, it’s going to make that difference.
So, I think this is kind of a nothing burger. It gets a lot of press, but I think it’s a nothing burger, this issue of relaxing the bank capital caps. Next question. What do you make of Apple’s paper citing the problems that reasoning models currently face. Well, you know, you have you have to start by saying who’s writing the paper?
Apple. A company that doesn’t have a reasoning model. Okay. So that’s step number one. Step number two is and we go into some of the debates around it. And there are a lot of you know, there are a lot of very, very technical debates about whether they did some of the questions right or wrong or whether some of the questions had no answers.
I’m not even sure I care that much about the Apple paper, or at least I don’t care nearly as much about the apple paper as I do about a whole bunch of other papers, which are about real world observations on AI adoption and efficiency. And, I’ll talk about a couple of those in a minute. But here’s our takeaway where we are. AI in the workplace is increasingly being used for summarizing materials and automating super routine tasks.
Now, if you think about it, there’s an enormous amount of very routine tasks that take place in society. So even if AI never does anything more than summarizing materials and automating routine text, that’s a lot. Now it still struggles with certain ancient user interactions. It definitely struggles maintaining confidentiality. At times. And almost everything they do requires double checking.
But, even a revision process that requires double checking is less time consuming and expensive than having to grade the work complete. You can crunch. And, you know, the bottom line is you don’t need AGI, to, you know, which is real, true, artificial intelligence to see some real productivity gains in the U.S. or say, you know, it may take years for those productivity gains to show up, but I don’t believe and my team doesn’t believe that we need AGI to see those productivity gains.
So the in the paper this week we cite I’m just going to talk about them quickly. There were three recent pieces that we’re kind of supportive of AI in the workplace and and one that was negative. So the three good ones was first there was, a warm foundation poll of teachers and, and an increasing number of teachers are using them to plan lessons, grade assignments.
They are saving six hours per week, which translate to six weeks per school year. And over half the teachers report that it’s improving the quality of their work. So within education it’s already having a big impact. Then there was another paper, and this one was really fascinating to me. There’s something called a Cochrane review, and some of you may know what that is.
And it’s when you synthesize all the available research on a given medical subject, and this is the kind of thing that that can take, several work years to do. And some researchers created a program to do this, and it did it in two days. And this tool that they create, it requires some human supervision. But compared to humans only, this program identified more relevant research and more accurately extracted data from the research that it found.
So this is kind of a remarkable achievement. And, you know, if you want to read more about this, there’s information on the market. And then there was a Stanford paper. They do a lot of interesting work on this, and they’re talking about an increasing share of adults using language models at work. And that for tasks where people use these language models at work, completion times fell from 90 minutes to 30 minutes.
And mostly in customer service, marketing and technology, which is where you’d expect that to take place. So those are all consistent with and rhyme with some of the other things we’ve been writing about. This year, the negative, the one with the negative outcome was Salesforce, which is an important participant in this discussion given what they do.
They tried to use an LLM, you know, a language model agent, and it didn’t do that well. It only got a 60% success rate in in simple Q&A scenarios. And it did even worse. And only 35 success rate in some what they call multitasking interactions. It also exhibited almost a terrible confidentiality awareness. And even when it was prompted, it sometimes, either ignores it or it compromises the performance.
And so this study also found that the language models struggle to request necessary additional information when it wasn't provided. So, that one was negative. On balance, this is what you’re going to get. You're going to get some success stories. You’re going to get some things that are not so successful, but, more good than bad. And so far what we’re seeing.
So bottom line is I don't think we should as investors, be paying too much attention to this ample paper on reasoning models. Okay. Let’s talk about tariffs. What'’ the latest on tariffs I know everybody’s bored of this topic. And so are we. But it’s important and everything keeps changing. So we have this giant model that has all the thousands of products and 200 countries and different rules for each thing.
And every time something gets announced, my team and I go in and play around with the model to reflect the latest boom and its developments. And so here, here’s the chart. So we started the year with the with an effective tariff rate on imported goods of, of about two, 2.5%. And it looks like it’s in the neighborhood of 17%.
And, there’s a lot of assumptions that go along with making these kind of charts that I don’t think people talk about enough because only around a third of that 17% is kind of legally in the clear already being collective. And everybody kind of understands the rules. The rest of it is being contested in court or subject to negotiation.
And there’s a lot of really important questions that we don't know the answers to and nobody knows the answers to, but that we have to guess at in order to, to create charts like this and make estimates like this. So a big one is how much trade with Mexico and Canada will ultimately be compliant with the MCA free trade zone.
Last year there was no incentive to do the paperwork because the most non-MCA trade was subject to almost our tariffs anyway. And so if you look at how much what the paperwork was done last year, you’d say, well only half of the trade with Canada and Mexico is MCA compliant. That's a dumb assumption because now that there’s an incentive to have that paperwork filled out, we think the numbers, the compliance levels could be as high as 90%.
So that’s one big question. The second one is how much substitution will take place from China. In other words, how much will the Chinese imports for China have already fallen by 40%. Now some mix of that is domestic substitution in the U.S., some of that is importing from other countries, and some of that is just production levels that are declining and they’re not doing anything.
So we don't really know what those substitution levels are going to be yet in the long run. All the Chinese imports are subject to fentanyl tariffs. And so that’s something that that still creates an economic incentive to try and avoid. And then what's going to happen with pending to section 232 negotiations on pharma and semiconductors?
We have no idea. We have not included any estimates of those yet because we don’t know what’ll happen. And then the big one is there are two big lawsuits taking place right now at the, appellate level in the courts, the way my constitutional law contacts explain it to me is Trump has to win both in order for those tariffs to remain.
If he loses one of them, or both, then they got to go back to the drawing board, presumably only after petition petitioning the Supreme Court for relief. And we really don’t know how the Supreme Court would respond to this kind of thing or the timetable for them responding to it and whether they would grant it stay before responding to it.
So we don't know any of that. And that’s going to take probably the next couple of months to figure out. But all of this tariff stuff is still up in the air, depending upon whether or not these two separate court cases. One of them is in the DC circuit and the other one is in the court of Appeals for the Federal Circuit.
So those are the two places that these two lawsuits are still being fought out. Okay. Now, assuming that the tariffs remain and without making any estimates of tariffs on semiconductors and pharmaceuticals, we have a table in here of where each where the big countries that export in the United States started the year and where they would be now.
So, for example, Canada only goes from 0.1 percent tariff to two and a half negligible. Mexico goes from 0.2 to 7 or 8. Okay. Imports on China go from 10 to 43. Germany from one and a half to 26. Japan one and a half to 16. Vietnam 4% to 18%. South Korea basically 0 to 20%. So there are some big jumps here.
And that's why I think we still kind of have to hang on if these are applied and see what the economic impact in terms of growth and inflation. Now it’s a one-time hit. Doesn’t happen forever. But I think it's large enough that that will that it makes it certainly I believe makes sense for the Fed to kind of be on hold and wait to see how some of this plays out before they take a firm stance on monetary policy.
So this tables in there and then what are the long term consequences? I don't know, like one question I have is that you put some very high tariffs on imported steel and aluminum. The problem is you got 50 times the number of workers in the downstream industries that use products based on steel and aluminum. So yeah, you can maybe protect some of the upstream industries, but there’s a huge multiplier effect in terms of both employment and output from all the downstream industries.
So I think there could be some unforeseen consequences of this because and then, you know, we’re also at a time where everyone's talking about AI infrastructure and data centers and more turbines and more solar and storage, but let’s focus for a minute on what’s happening to tariffs on imported capital, intermediate goods. We have a chart on the, on the market that looks at, you know, batteries and insulation and turbines and transformer equipment and tires, valves, machinery, air pumps, electric fittings, you know, plastic, plastic, components, liquid pumps, motors, generators, these are all of the kinds of things that are the input to the infrastructure build out that the administration is hoping to accomplish. And as you can see from the chart here, you’ve got, you know, 20 to 35% tariffs on these capital goods. And this is no longer just a tariff on imported, consumer goods, where I think the arguments are different. Okay. Next question. So I read lots of, you know, hundreds and hundreds of pages of research every week.
And one of my favorite reads is, a firm located in Hong Kong called Gavital. Some of you have probably seen Louie Gave or his father, Charles, over the years. They’re kind of relocated Frenchmen that have moved to Hong Kong. And, I have a lot of respect for their firm and for Charles in particular, who’s a great thinker.
But Charles recently wrote something that that jumped out at me and jumped out at some of the clients who kind of mentioned this to me, where he wrote that the U.S. dollar is already no longer the world’s reserve currency. Well, either, you know, that's where we would disagree, right? I have intellectual disagreement with that. It’s true.
So far this year, in trade weighted terms, different measures of the dollar have declined by 5 to 10%. Now they were at the most elevated levels since 2002. Right. So in almost 25 years, they were at the high level. And even looking back longer than that, the dollar was pretty expensive. So a 5 to 10% decline from elevated levels doesn't necessarily tell me that anything seismic and unstoppable is happening to the dollar.
And one of the ways that we measure that is there are several different things to look at about whether or not the dollar is holding up as the world’s currency. And there are six statistics in particular. We let and we looked at them in 2022. And then we've just refreshed them last month. And it’s things like what percentage of cross-border loans are issued in dollars.
Same for international debt securities. The dollar share of global transactions, the dollar share of foreign exchange reserves invested by foreign central banks, trade invoicing. How much of that is in dollars and then and then importantly, the dollar share of global Swift payments. And as you can see from the table, there haven’t been any large declines, except maybe an 8% decline in the trade invoicing measure.
And some of the measures have gone up since 2022. So I don’t really see the hard core evidence that the dollar is in the process of being displaced as the world’s reserve currency. I don’t see it. Now I share everyone’s concern. We’ve had these charts in here before, from the Yale Budget Lab about what the new budget bill is going to be doing to both the debt to GDP ratio and the deficit.
You would need a whole lot of tax. Sorry. You would need a whole lot of tariff revenue in order to offset the fiscal expansion associated with the bill itself. We’ll see. Yale has an estimate of about 200 billion a year from tariffs, but that’s before retaliation and all sorts of other economic impact.
So I don’t know. And I think it’s a fool’s game at this point to try and project it. What we do know is excluding tariff revenue, the bill would result in a fairly sharp increase in both deficits and the debt to GDP ratio. So I understand why people are concerned about, the, the dollar’s role. We’re just not seeing any impact of that quite yet.
And then I did want to show you this, but with this one chart, you remember the Sesame Street. One of these things is not like the other. This is a chart that has six lines on it, and five lines are from the Tax Foundation, the Tax Policy Center, the CBO, Wharton and Penn with projections of the growth impact of the budget bill.
And they all think it’s going to be positive. They just think it’s going to range from 0 to 1%. The White House projections are kind of off the charts and peak at 5%. In 2028, around the time of the next election. And so, but, you know, we'll see, there I wouldn’t write off completely the fact that the White House may be closer to the truth, although I would find it surprising if you had that kind of growth impact from the bill, given all of the other stuff that’s going on.
Okay. Almost done. I was asked this question in person at a dinner the other night, and, sometimes people don’t ask questions. They kind of say them and then say it was a question, but it was really more of a statement. I'm going to interpret it as if it was a question, which is, do J.P. Morgan’s blockchain project argue for building positions in Ethereum?
And, I, you know, obviously spend a lot of time talking to the people within J.P. Morgan who work on all these things. And the answer is not really, at least not yet. So I wrote about this in the stablecoins piece a few weeks ago. J.P. Morgan has created something like a Nexus product, which is a new intraday blockchain based repo settlement system.
And, you know, it’s a private permissioned blockchain. And importantly, it was built on a fork of Ethereum. And as a result, there’s no direct connection back to Ethereum in any way that would create tolling revenues for the network or gains for Ethereum holders. So should you buy Ethereum because J.P. Morgan is building this kind of blockchain project?
No, because there is no tolling revenue consequence based on the way that it was built. Just to be clear, the daily volumes for this thing are only $2 billion compared to the $10 trillion that J.P. Morgan handles in traditional payments every day. And it remains to be seen just how much demand there is out there for a blockchain based B2B payment system.
That’s it. You know, if J.P. Morgan or other firms create great products on the public blockchain and not on private permissioned blockchain, they might well be built on Ethereum. And related layer two chains connected to a dream or Solana or something like that. And then you could argue that there are some tolling revenue benefits for the crypto holders on that network.
J.P. Morgan has also just completed a proof of concept permissioned deposit token called JPMD. It's basically a stablecoin alternative. What's interesting is, what? You know, why should J.P. Morgan have to create a stablecoin alternative? What's wrong with stablecoins? Well, read our stablecoin base to find out. But anyway, J.P. Morgan created a deposit token called JPMD and, that will initially be for the firm's institutional clients to make on chain money transfers.
This one is a layer two product developed by Coinbase on top of Ethereum, and there may be some incidental benefits of tolling revenues to Eve holders over time, but this is a total proof of concept kind of thing. And so to me it feels early to argue that this kind of tokenization argues for building big positions in crypto.
But again, you know, I’m not a crypto guy. I don’t have, you know, crypto tattoos on my arms. So we'll see. All right. Next, I did want to I did want to mention this because I thought it was important. And I did get this question from some clients that I ended up seeing recently. Just last week.
What do you make of Sam Altman’s comments on fraud risk from voice authentication, cloning? And so, at a Federal Reserve conference last week, Altman referred to a significant impending fraud crisis with the words he use resulting from AI voice cloning technology that can fool traditional voice authentication methods, methods that banks typically use. And you’ve probably seen this some cloning tools only require three seconds of someone’s recorded audio to create a convincing impersonation.
That would fool. That could rule a person. But the thing is that most of these cloning tools, while they could fool a person, they can’t fool certain kinds of programs. And so I just going to walk through quickly some of the tools that banks can use to protect monitoring this kind of thing. If people should be aware, some of the options that I’m about to mention, you have to sign up for them.
The banks won’t always use them by default. Obviously you should contact whatever bank you deal with to find out. So there are, there are, you know, you fight fire with fire. There are deepfake detection tools that use AI to figure out, if the voice is generated by some kind of synthetic program. And one example is a deepfake detector by Reality Defender, which is an AI program, and it looks for unnatural pitch, rhythm, articulation, and other irregularities that humans wouldn’t spot.
But it can spot. And, are people tell me that that tool is works well and is a pretty interesting one. And there’s also in-app identification. Right. So if you call your banker or your customer service area and they don’t know who you are, they can require you to use push notification to authenticate you, on your on your authorized device.
So this is for establishing your identity, not payment authorization. But step one is you could use in-app authentication to authenticate who’s calling. You could also use digital payment verification and do the very same thing, where you have to sign into your account and click and approve something before any money can be moved. And I was talking to some of our bankers.
I’m still amazed that some of you are sending payment instructions by email. That just seems crazy to me in the day and era that we work with, and it just surprises me that everybody doesn’t require, certain kinds of phone calls to do that kind of thing. There’s also phone porting, SIM swap, you know, there's a vendor service available within the U.S. that can try to figure out if the caller is calling from a line that was SIM swapped or forwarded or phone ported.
And so that’s another protective thing. And then the last thing of course, is payment characteristic pattern recognition. That tries to figure out if what you're trying to do is out of character. And that would obviously trigger a whole other set of approvals. So I think Altman’s right that banks relying on traditional voice identification authentication methods only by people, may have certain risks, but I’m encouraged to see the number of tools that are being developed to potentially offset some of those things.
And then the last question is where, and when are your next fishing trips? So mine. And so let's see at the end of August, taking my Hopi Mirage outback kayak to British Columbia to fish for sturgeon. I have a picture here of last time I was there during the coho run. It was great. And then I plan to go back in November trying to convince George Cash from asset management to go with me, to fish for tarpon in Trinidad and podcast Still Dorado.
So anyway, that is the end of the summer mailbag Eye on the Market podcast. And, stay tuned. In early August, some point, maybe around the 10th of the 12th, we're going to have our health care piece come out. Thank you very much for listening. Hope you’re having a great summer.
(DESCRIPTION)
Logo: JP Morgan. Title: Eye on the Market
(SPEECH)
Welcome to the July Eye on the Market podcast.
(DESCRIPTION)
The speaker appears on the right. A painting on the left shows a man on horseback with a mail bag over his shoulder. Text July 2025. Summer Mailbag: Q&A from the Eye on the Market client mailbag
(SPEECH)
This one's called "Summer Mailbag" because I'm going to answer some of the client questions that we've been receiving recently. First,
(DESCRIPTION)
A photo shows a dog laying behind an open booklet. Title: Eye on the Market.
(SPEECH)
one quick comment. As many of you may know, we issued our 20th anniversary Eye on the Market a few weeks ago. And we have some printed booklets. If you find those easier to read for a document that, it's that long. And so feel free to call your coverage team if you want to get a printed booklet of your own. Dog not included.
And let's see. And
(DESCRIPTION)
Text: August Eye on the Market preview, health care. S&P 500 healthcare vs technology total return, 1969-2019. Index (100 = September 1989). A line graph shows similarities between lines for health care and technology.
(SPEECH)
quickly, a preview of the August Eye on the Market, which is going to be a health care piece that I've been working on for the last few weeks. And I want to show you a couple of charts here that are just amazing. For three decades, almost starting in 1989, a couple of years after I joined J.P.Morgan, for three decades until 2019, the health care sector was every bit the growth sector the tech was and tracked the returns remarkably, almost identically over those three decades and with a lot less volatility.
Healthcare vol was about 15 versus technology at closer to 25 over this 30 year period. Ever since 2019, the health sector has been really sick.
(DESCRIPTION)
Text: S&P 500 healthcare vs technology total return, 2019-2025. Index (100 - December 2019). A graph shows the line for health care lower than technology
(SPEECH)
And as you see here, from 2019 to 2025, technology has continued its march and health care is more or less flat or almost flat over 6, 7 years. So we're going to take a look at all the things that are going on from proposals on capping drug prices to cuts to the CDC and the NIH, slower approval of new drugs, the patent cliff, and all the things that are associated with what's dragging down healthcare. We're going to do that next in the August Eye on the Market.
So question number one that I've been getting, why is there such a long backlog for natural gas turbines? And do their higher prices mean that solar and storage is comparable economically as a form of baseload power? Well, you may have seen this. There was this Pennsylvania Energy and Innovation Summit a few weeks ago, and a bunch of companies, including Google and CoreWeave, Blackstone, Westinghouse, Energy Capital Partners announced $90 billion of intended data center and data center expansion. And they also talked about the need for the power infrastructure to support all those data centers.
And a lot of what they announced is intended to be powered by new gas turbines. The problem is supplies are pretty tight. Turbine delivery times, at least for combined cycle turbines rather than the combustion turbines, can read 3 to 7 years. And just over the last couple of years, their prices have more than doubled from $1,200 per kilowatt to around $2,500 a kilowatt. And some of the CEOs in the sector are also citing a lot of labor shortages and the skilled labor pool required to build these gas turbine projects. And so it's interesting to see just how much of a backlog there is.
(DESCRIPTION)
A bar graph shows gas plants added by state. Text: Gas plants added since Jan 2019 as a share of operating gas plants, Sorted in order of electricity consumption in MWh
(SPEECH)
Now, for many years, most parts of the United States and other places around the world have been giving grid preference to renewables. So there hasn't been really that much gas, natural gas infrastructure built. And we wanted to take a closer look at this. This chart shows the gas turbine capacity built over the last seven years or so, as a percentage of all the gas capacity operating in the state. So for example, in Texas, only 3% of all the gas capacity they've got have been built since 2019. And in California, that number is 9%. In New York and North Carolina it's around 10%.
So only in Illinois have you seen a meaningful expansion and maybe to a lesser extent, in Ohio and Pennsylvania of natural gas capacity recently. So there is a lot of demand. There hasn't been that much built recently. And understandably, Siemens and GE and Mitsubishi, which are the three large turbine manufacturers, are taking advantage of higher demand and higher prices, but are not meaningfully expanding their production capacity, at least not yet.
So these delays will be with us for a while. And as I mentioned, there are already these delays have already pushed up prices. So
(DESCRIPTION)
Text: Las Vegas utility scale grid configurations, $ 3 per MWh of system load. Source: NREL EMBER UPMAM, July 2025
(SPEECH)
I thought it was interesting last week, one of the big research institutes on energy wrote a paper basically saying we now are in an affordable era where solar + storage can carry almost the entire load. And they cited 97% of the load in a place like Las Vegas.
So I remember that old Reagan saying trust but verify. And whenever anybody writes an energy paper, that's exactly what I do. I will trust that they did it the right way. But then I'm going to verify every single calculation. And so we have an hourly grid model of our own. And we can easily set it to a place like Las Vegas for purposes of irradiation and hourly solar generation, things like that. And if you look at this chart that we have Eye on the Market and there, which was the entity that made these forecasts, that orange line at the bottom of the chart. That's their estimate of different grid configurations for solar.
Look at the end of the chart. So they've got something where 90 plus percent of the solar can come from photovoltaic solar, a large amount of back of battery storage, and then a rather small backup thermal system to carry the rest of the load. And they get a number in the neighborhood of $105 a megawatt hour. Some of their assumptions we agreed with a bunch when we didn't. And so when all the rest of the curves on this chart represent other iterations based on our cost, real world loads, different operating lives, different gas prices and things like that.
And based on what you're seeing here, the bottom line is I think we're a long way away on an unsubsidized basis, which is how you have to look at these kinds of things. I think we're a long way away from when solar and storage are comparable economically as a form of baseload power to gas. And anway, we'll talk about this a lot more in the energy piece. But these are the numbers. So I do think that the power production is going to be something of a speed bump for a lot of the data center build outs.
All right, next question. Is China constraining exports of rare earth materials to the US? And what's the deal with the Department of Defense investing in MP Materials? The answer to the first question is yes, at least for now.
(DESCRIPTION)
Shows a column graph for 2018 to 2025. Text: China rare earth exports to the us US$, millions. Rare earth materials, compounds, magnets. Source General Administration of Customs of China JPMAM, May 2025
(SPEECH)
So this is a chart on Chinese rare earth exports to the US. And they fell by 80%, 90% in May. Now that's when China was trying to apply maximum leverage to the United States. Once we get the final data for the rest of the year, particularly after there's some kind of a deal, presumably they'll go back up.
But I just wanted you to see this chart because China is already understandably exercising their leverage in these tariff discussions by really choking off rare earth exports to the US. Now, a lot of US entities, whether they're the private sector or the military, have stockpiled some of these materials, but those stockpiles aren't going to last forever. Now the Pentagon talks about wanting a magnet, the China free magnet supply chain by 2027, but I think this is a pretty tough goal to meet.
China operates in an environment with scant environment or environmental regulations overseeing their industry. There are 39 Chinese universities with rare earth training programs. The US has none, which is why China has a 90% share in the processing of these rare earths. There are some new US processing capacity that will come online in 2026, 2028, but it's going to take a while. And for example, MP Materials, which I'll talk about in a second, produces around 1,000 metric tons of neodymium, compared to 300,000 metric tons in China. And those are annual numbers. Pardon me.
So that all may explain why the Defense Department became the largest shareholder, which is unusual. But the US Defense Department has now become the largest shareholder in MP Materials, which is a US-based company. They bought about $400 million of its convertible preferred stock and is helping it build out a new processing facility. And they've entered into a 10-year agreement that puts in place a price floor for the company's neodymium and praseodymium alloy, and promises for some period to buy 100% of the magnets that they produce. I think this is a good thing for the United States to finally get serious about this.
At the end of 2024, MP Materials, which is a US company, their largest shareholder was a Chinese firm. That's who the Department of Defense is essentially replacing here. So if this build out in the US can be done responsibly, it's a good thing.
(DESCRIPTION)
Shows a line graph. Text: Rare earth oxide prices during Chinese export quotas, Index (100 = 2009). Source: JPMAM 2025
(SPEECH)
I just wanted to also show you on this page a chart of what happened to rare earth oxide prices the last time China put quotas on. Before they took them off, the prices rose by a factor of almost 10. And so that's what happens when you've got one country that's got 90% market share on processing and has the inclination to apply export quotas. So it'll be a long road, but at least what the US Department of Defense is doing is a first step.
Next question. What are you guys seeing these days regarding the YUCs, the MEGAs, and equity market valuations? A lot of our clients are familiar with our YUCs analysis. YUC stands for young, unprofitable company. And when this YUC measure, which looks at the share of unprofitable companies as a percentage of market cap. When this thing was spiking in 2021 and 2022, that's one of the reasons why we became very bearish at the time on anything having to do with SPACs or a lot of metaverse, hydrogen, all of those kinds of things.
There's a lot of these young, unprofitable companies hit the market all at once, and right now that number has come back down. It's not as low as it was from, let's say, around 2005 to 2012, when it was way less than 1% of overall market cap. But it's down substantially from 2022. The bazooka period for stimulus. And so that's a good sign. It's one good sign for risk taking. It's not the only one. But in terms of risk taking that's a positive to me.
(DESCRIPTION)
Text: Companies with negative earnings by size % of companies with the negative LTM earnings. A line graph shows small cap and large cap between 1990 and 2025. Source: Factset, JPMAM, June 30, 2025
(SPEECH)
The other positive is if we have two separate large cap and small cap here, the share of large cap companies that have negative earnings. That number is also coming sharply down. It was as high as 20% during 2021, and now is back down to about 7%. That's close to the low levels that it's been at over the last 30, 40 years. In contrast, small cap is an absolute desert of profitability as it depends how you measure it, but based on our measurements, more than 40% of all small cap companies have negative earnings, which is one reason why small cap performance has performed so poorly relative to the large cap stocks over the last decade.
(DESCRIPTION)
Top 10 companies share of total earnings and market cap, % of total. Source: Factset, JPMAM, June 30, 2025
(SPEECH)
Then what about the MEGAs, which are the super large companies? We're in uncharted territory, obviously, on market cap in terms of these companies now represent and have for the last couple of years, around 35% of market cap is just coming from the top 10 companies. And we can look at this going back to 1990. And that's a huge number. What makes it almost palatable is that their earnings represent a similar share of overall market earnings.
Now again, you have to measure this different ways. But if we look at consensus 12 month forward earnings estimates, these 10 companies also represent 35% of the overall market share of earnings. So while it's disturbing to see 10 companies at around a third of the entire market cap, at least their earnings share has gone with it on a trailing basis. That earnings share is lower on a forward looking basis. That earnings estimate is right on 35%. So obviously, a lot depends upon how on whether or not those 12 month earnings estimates come to fruition. But as a general measure, the top 10 here don't look quite as offsides as you might think they are.
(DESCRIPTION)
Tech & interactive media valuations vs the rest of the market. Forward P/E ratio. Source: Factset, JPMAM, June 30, 2025
(SPEECH)
And then the last thing is on valuations, we like to separate tech and interactive media from the rest of the market. And as you can see here, the tech and interactive media valuations are hovering at that 30 times forward earnings number that they've been at really for the last eight years or so. These are some pretty lofty valuations, but they've been sustainable based on earnings growth.
The rest of the market right, which is 450 plus stocks. Their multiples are more or less average when looking back at the last three decades or so, at about 20 times forward earnings. Those valuations have been creeping up to be sure. But they don't look quite as offsides as the rest of the market does as much as the top 10 do.
All right, next question.
(DESCRIPTION)
Text: What would be the real-world implications of a relaxation in certain US bank capital regulations?
(SPEECH)
There's been a lot of discussion about, I mean, if you think about the broad Trump agenda and the issues around inflation and tariffs and deportations, one of the things that's kept the business community and people, the CEO roundtable, business roundtable, one of the things that's kept them engaged with the administration is the prospects for deregulation. And we've written many, many, many times about how the Biden administration set a gold record in terms of the pace of government regulation compared to other administrations.
So a lot of our CEO clients were happy to see that there would be a swing in the pendulum back in the opposite direction. These things take time, and it's a little bit early to measure the results. Yet we think in a few months, we'll have a better sense of it.
(DESCRIPTION)
Shows a plot graph. Text: Excess tier 1 capital by ratio under current SLR rules. Source: Bloomberg, JPMAM, Q1, 2025
(SPEECH)
But one thing that is interesting as relates to the banks, which is what would happen if there were relaxation in bank capital standards? In general, you'd think that's a good thing if you're an investor because banks would be able to buy back more stock and banks would be able to do more risky lending.
And so in principle, reducing bank responsibly, reducing bank capital requirements without going crazy could be an economic boost. The problem is there are many bank capital standards and what's being proposed is just relaxing one of them. And the only way this has a big impact is if the one they relax is the binding constraint. And if there isn't another capital constraint right behind it. And unfortunately neither of those things is true. Another here's what I'm saying is the supplementary leverage ratio is the one that is being discussed to be relaxed.
And as you can see in the chart that you'll see when you see the Eye on the Market it's either the binding constraint or very close to it for most of the big G-SIB, 8 G-SIB banks globally, are the systemically important banks. And so if you relax a capital standard that's not binding or if it is binding and there's another one close behind it, it doesn't really do anything. Here's a way to visualize that.
(DESCRIPTION)
A bar graph appears. Text: Excess Tier 1 capital only increase by $6bn under new relaxed SLR rules, US$, billions
(SPEECH)
We have a before and after chart, and the only thing you get from revising these supplementary leverage ratio rules is little bit a couple of billion dollars for Goldman, Morgan Stanley, BNY Mellon, and State Street. And it doesn't do anything for J.P.Morgan, City, Wells, or B of A. It's a rounding error in terms of increased capital, tier 1 capital for the banking industry. So that doesn't really do much for you.
One of the things that the treasury, I think Bessent has mentioned this, is, well, at least that same relaxation would increase the treasury buying capacity. I'm not going to walk through all the mathematical reasons why. But we've done all of the work and we've modeled all of this. And in principle, the banks would have the ability-- they currently have the ability to purchase another $3 trillion of treasury bonds before some rules kick in. Now they'd be after-- if they'd make this rule change, that number would go to $8 trillion.
So in principle, maybe the banks could act as a safety valve if there were an unexpected collapse in foreign demand for treasury bonds by either foreign official buyers or private sector buyers. The problem is, if the banks aren't taking advantage of $3 trillion of treasury buying capacity right now, I don't how expanding that number to $8 trillion is going to make that big of a difference. So I think this is kind of a nothing burger. It gets a lot of press, but I think it's a nothing burger in this issue of relaxing the bank capital standards.
Next question. What do you make of Apple's paper citing the problems that reasoning models currently face? Well, you have to start by saying who's writing the paper? Apple, a company that doesn't have a reasoning model. OK, so that's step number one. Step number 2 is-- and then we go into some of the debates around it. And there are a lot of-- there are a lot of very, very technical debates about whether they did some of the questions right or wrong or whether some of the questions had no answers.
I'm not even sure I care that much about the Apple paper, or at least I don't care nearly as much about the Apple paper as I do about a whole bunch of other papers, which are about real world observations on AI adoption and efficiency. And I'll talk about a couple of those in a minute. But here's our takeaway where we are. AI in the workplace is increasingly being used for summarizing materials and automating super routine tasks.
Now, if you think about it, there's an enormous amount of very routine tasks that take place in society. So even if AI never does anything more than summarizing materials and automating routine tasks, that's a lot. Now it still struggles with certain aged user interactions. It definitely struggles maintaining confidentiality at times, and almost everything they do requires double checking.
But even a revision process that requires double checking is less time consuming and expensive than having to create the work the work from scratch. And the bottom line is you don't need AGI, which is real, true artificial intelligence, to see some real productivity gains in the US society. It may take years for those productivity gains to show up, but I don't believe and my team doesn't believe that we need AGI to see those productivity gains.
So in the paper this week we cite-- I'm just going to talk about them quickly. There were three recent pieces that were supportive of AI in the workplace and what it's doing and one that was negative. So the three good ones was first, there was a Walton Foundation poll of teachers, and an increasing number of teachers are using them to plan lessons, create assignments. They are saving six hours per week, which translates to six weeks per school year. And over half the teachers report that it's improving the quality of their work. So within education, it's already having a big impact.
Then there was another paper, and this one was really fascinating to me. There's something called a Cochrane review, and some of you may know what that is. And it's when you synthesize all the available research on a given medical subject. And this is the kind of thing that can take several work years to do. And some researchers created a program to do this, and it did it in two days. And this tool that they create, it requires some human supervision. But compared to humans only, this program identified more relevant research and more accurately extract the data from the research that it found.
So this is a remarkable achievement. And if you want to read more about this, there's information on the market. And then there was a Stanford paper. They do a lot of interesting work on this, and they're talking about an increasing share of adults using language models at work. And that for tasks where people use the language models at work, completion times fell from 90 minutes to 30 minutes, and mostly in customer service, marketing, and technology, which is where you would expect that to take place. So those are all consistent with and rhyme with some of the other things we've been writing about this year.
The one with the negative outcome was Salesforce, which is an important participant in this discussion given what they do. They tried to use an LLM, a language model agent. And it didn't do that well. It only got a 60% success rate in simple Q&A scenarios. And it did even worse at only 35% success rate in some what they call multi-turn interactions. It also exhibited almost a terrible confidentiality awareness. And even when it was prompted, it sometimes either ignores it or it compromises the performance.
And so this study also found that the language model struggled to request necessary additional information when it wasn't provided. So that one was negative. On balance, this is what you're going to get. You're going to get some success stories. You're going to get some things that are not so successful. But more good than bad is so far what we're seeing. So bottom line is I don't think we should, as investors, be paying too much attention to this Apple paper on reasoning models.
OK, let's talk about tariffs.
(DESCRIPTION)
A line graph appears. Text: Average tariff rate on all US imports - Assuming 20% US domestic and 20% foreign substitution from China. Source: Tax Foundation USITC, White House, JPMAM, 2025
(SPEECH)
What's the latest on tariffs? I know everybody's bored of this topic and so are we. But it's important and everything keeps changing. So we have this giant model that has all the thousands of products in 200 countries and different rules for each thing. And every time something gets announced, my team and I go in and play around with the model to reflect the latest developments. And so here's the chart. So we started the year with an effective tariff rate on imported goods of about 2%, 2.5%. And it looks like it's in the neighborhood of 17%.
But there's a lot of assumptions that go along with making these charts that I don't think people talk about enough because only around a third of that 17% is legally in the clear already being collected. And everybody understands the rules. The rest of it is being contested in court or subject to negotiation. And there's a lot of really important questions that we don't know the answers to. Nobody knows the answers to, but that we have to guess at in order to create charts like this and make estimates like this.
So a big one is how much trade with Mexico and Canada will ultimately be compliant with the MCA free trade zone? Last year, there was no incentive to do the paperwork because most non-MCA trade was subject to almost our tariffs anyway. And so if you look at what the paperwork was done last year, you'd say, well, only half of the trade with Canada and Mexico is MCA compliant. That's a dumb assumption because now that there's an incentive to have that paperwork filled out, we think the numbers, the compliance levels could be as high as 90%. So that's one big question.
The second one is how much substitution will take place from China? In other words, how much will-- the imports of China have already fallen by 40%. Now some mix of that is domestic substitution in the US. Some of that is imported from other countries, and some of that is just production levels that are declining and they're not doing anything. So we don't really know what those substitution levels are going to be yet in the long run. All the Chinese imports are subject to fentanyl tariffs. And so that's something that still creates an economic incentive to try and avoid.
And then what's going to happen with pending section 232 negotiations on pharma and semiconductors? We have no idea. We have not included any estimates of those yet because we don't know what will happen. And then the big one is there are two big lawsuits taking place right now at the appellate level in the courts. The way my constitutional law contacts explain it to me is Trump has to win both in order for those tariffs to remain. If he loses one of them or both, then they got to go back to the drawing board, presumably only after petitioning with the Supreme Court for relief.
And we really don't know how the Supreme Court would respond to this kind of thing, or the timetable for them responding to it, and whether they would grant a stay before responding to it. We don't any of that. And that's going to take probably the next couple of months to figure out. But all of this tariff stuff is still up in the air, depending upon whether or not these two separate court cases. One of them is in the DC circuit, and the other one is in the Court of Appeals for the Federal Circuit. So those are the two places that these two lawsuits are still being fought out.
(DESCRIPTION)
A table appears. Text: Estimated final destination tariff rates as of July 2025
(SPEECH)
Now, assuming that the tariffs remain and without making any estimates of tariffs on semiconductors and pharmaceuticals. We have a table in here of where the big countries that export to the United States started the year and where they would be now. So, for example, Canada only goes from 0.1% tariff to 2 and 1/2 negligible. Mexico goes from 0.2% to 7% or 8%. OK, manageable. Imports on China go from 10% to 43%.
Germany from 1 and 1/2 to 26%. Japan 1 and 1/2 to 16%. Vietnam 4% to 18%. South Korea basically 0% to 20%. So there are some big jumps here. And that's why I think we still have to hang on if these are applied and see what the economic impact in terms of growth and inflation. Now it's a one time hit. Doesn't happen forever. But I think it's large enough that it makes it certainly I believe makes sense for the Fed to be on hold and wait to see how some of this plays out before they take a firm stance on monetary policy.
So this table is in there.
(DESCRIPTION)
Text: Steel and aluminum manufacturing employees, 2022. Employees, millions. Source: 2022 Economic Census, JPMAM, 2026
(SPEECH)
And then what are the long-term consequences? I don't know. One question I have is that you put some very high tariffs on imported steel and aluminum. The problem is you got 50 times the number of workers in the downstream industries that use products based on steel and aluminum. So yeah, you can maybe protect some of the upstream industries, but there's a huge multiplier effect in terms of both employment and output from all the downstream industries. So I think there could be some unforeseen consequences of this.
(DESCRIPTION)
Text: Effective tariff rate on largest categories of US capital and intermediate goods imports, Percent. Source: USITC, White House, JPMAM, 2025
(SPEECH)
And then we're also at a time where everyone's talking about AI infrastructure and data centers and more turbines and more solar and storage. But let's focus for a minute on what's happening to tariffs on imported capital and intermediate goods. We have a chart in the Eye on the Market that looks at batteries and insulation and turbines and transformer equipment and tires, valves, machinery, air pumps, electric fittings, plastic components, liquid pumps, motors, generators.
These are all of the kinds of things that are the input to the infrastructure build out that the administration is hoping to accomplish. And as you can see from the chart here, you've got 20% to 35% tariffs on these capital goods. And this is no longer just a tariff on imported consumer goods where I think the arguments are different.
OK, next question.
(DESCRIPTION)
Text: Charles Gave wrote that the US dollar is no longer the world's reserve currency. What's your view?
(SPEECH)
I read hundrends and hundreds of pages of research every week. And one of my favorite reads is a firm located in Hong Kong called Gavekal. Some of you have probably seen Louis Gave or his father Charles over the years. They're relocated Frenchmen that have moved to Hong Kong. And I have a lot of respect for their firm and for Charles in particular, who's a great thinker.
But Charles recently wrote something that jumped out at me and jumped out at some of the clients who mentioned this to me, where he wrote that the US dollar is already no longer the world's reserve currency. That's where we would disagree. I have intellectual disagreement with that. It's true. So far this year, in trade weighted terms, different measures of the dollar have declined by 5% to 10%. Now they were at the most elevated levels since 2002. So in almost 25 years, they were at a high level. And even looking back longer than that, the dollar was pretty expensive.
So a 5% to 10% decline from elevated levels doesn't necessarily tell me that anything seismic and unstoppable is happening to the dollar.
(DESCRIPTION)
A table appears. Text: US$ shares of global markets. Source: BIS Quarterly Review, ECB, IMF, Gavekal Research, JPMAM
(SPEECH)
And one of the ways that we measure that is there are several different things to look at about whether or not the dollar is holding up as the world's reserve currency. And there are six statistics in particular. And we looked at them in 2022. And then we've just refreshed them last month. And it's things like what percentage of cross-border loans are issued in dollars? Same for international debt securities, the dollar share of global FX transactions, the dollar share of foreign exchange reserves invested by foreign central banks, trade invoicing, how much of that is in dollars, and then importantly, the dollar share of global SWIFT payments.
And as you can see from the table, there haven't been any large declines, except maybe an 8% decline in the trade invoicing measure. And some of the measures have gone up since 2022. So I don't really see the hardcore evidence that the dollar is in the process of being displaced as the world's reserve currency. I don't see it. Now, I share everyone's concern. We've had these charts in here before, from the Yale Budget Lab about what the new budget bill is going to be doing to both the debt to GDP ratio and the deficit, that you would need a whole lot of tax, sorry, you will need a whole lot of tariff revenue in order to offset the fiscal expansion associated with the bill itself. We'll see.
Yale has an estimate of about $200 billion a year from tariffs, but that's before retaliation and all sorts of other economic impacts. So I don't know. And I think it's a fool's game at this point to try and project that. What we do know is excluding tariff revenue, the bill would result in a fairly sharp increase in both deficits and the debt to GDP ratio. So I understand why people are concerned about the dollar's role. We're just not seeing any impact of that quite yet.
And then I did want to show you this with this one chart.
(DESCRIPTION)
Text: Projected GDP growth impact of O.B.B.B.A, Source: CRFB, July 2025
(SPEECH)
You remember the Sesame Street, one of these things is not like the other? This is a chart that has six lines on it, and 5 lines are from the Tax Foundation, the Tax Policy Center, the CBO, Wharton and Penn with projections of the growth impact of the budget bill. And they all think it's going to be positive. They just think it's going to range from 0% to 1%.
And the White House projections are off the charts and peaked at 5% in 2028 around the time of the next election. But we'll see. I wouldn't write off completely the fact that the White House may be closer to the truth, although I would find it surprising if you had that kind of growth impact from the bill, given all of the other stuff that's going on.
OK, almost done.
(DESCRIPTION)
Text: Do JP Morgan's blockchain projects argue for building positions in Ethereum?
(SPEECH)
And I was asked this question in-person at a dinner the other night. And sometimes people don't ask questions. They stay them and then say it was a question, but it was really more of a statement. I'm going to interpret it as if it was a question, which is, do J.P.Morgan blockchain project argue for building positions in Ethereum? And I, obviously, spend a lot of time talking to the people within J.P.Morgan who work on all these things.
And the answer is not really, at least not yet. So I wrote about this in the stablecoins piece a few weeks ago. J.P.Morgan had created something called Kinexys product, which is a new intraday blockchain-based repo settlement system. And it's a private permissioned blockchain. And importantly, it was built on a fork of Ethereum. And as a result, there's no direct connection back to Ethereum in any way that would create tolling revenues for the network or gains for Ethereum holders.
So should you buy Ethereum because J.P.Morgan is building this blockchain project? No, because there's no tolling revenue consequence based on the way that it was built. Just to be clear, the daily volumes for this thing are only $2 billion compared to the $10 trillion that J.P.Morgan handles in traditional payments every day. And it remains to be seen just how much demand there is out there for a blockchain-based B2B payment system.
That said, if J.P.Morgan or other firms create products on the public blockchain and not on private permissioned blockchain, they might well be built on Ethereum and related layer-2 chains connected to Ethereum or Solana or something like that. And then you could argue that there are some tolling revenue benefits for the crypto holders on that network. J.P.Morgan has also just completed a proof of concept permissioned deposit token called JPMD. It's basically a stablecoin alternative.
What's interesting is why should J.P.Morgan have to create a stablecoin alternative? What's wrong with stablecoins? Well, read our stablecoin piece to find out. But anyway, J.P.Morgan created a deposit token called JPMD. And that will initially be for the firm's institutional clients to make on chain money transfers. This one is a layer two product developed by Coinbase on top of Ethereum, and there may be some incidental benefits and tolling revenues to ETH holders over time. But this is a total proof of concept kind of thing.
And so, to me, it feels early to argue that this kind of tokenization argues for building big positions in crypto. But again, I'm not a crypto guy. I don't have crypto tattoos on my arms. So we'll see.
All right, next, I did want to mention this because I thought it was important. And I did get this question from some clients that I ended up seeing recently, just last week. What do you make of Sam Altman's comments on fraud risk from voice authentication cloning? And so at a Federal Reserve Conference last week, Altman referred to a significant impending fraud crisis with the words he used resulting from AI voice cloning technology that can fool traditional voice authentication methods that banks typically use.
And you've probably seen this. Some cloning tools only require three seconds of someone's recorded audio to create a convincing impersonation that could fool a person. The thing is that most of these cloning tools, while they could fool a person, they can't fool certain kinds of programs. And so I'm just going to walk through quickly some of the tools that banks can use to protect clients from this kind of thing. And people should be aware some of the options that I'm about to mention, you have to sign up for them. The banks won't always use them by default. Obviously should contact whatever bank you deal with to find out.
So you fight fire with fire. There are deepfake detection tools that use AI to figure out if the voice is generated by some synthetic program. And one example is a deepfake detector by Reality Defender, which is an AI program. And it looks for unnatural pitch, rhythm, articulation, and other regularities that humans wouldn't spot. But it can spot. And our people tell me that tool works well and is a pretty interesting one.
There's also in-app authentication. So if you call your banker or your customer service area, and they don't know who you are, they can require you to use push notifications to authenticate on your authorized device. So this is for establishing your identity, not payment authorization. But step one is you can use in-app authentication to authenticate who's calling. You could also use digital payment verification and do the very same thing, where you have to sign into your account and click and approve something before any money can be moved.
And I was talking to some of our bankers. I'm still amazed that some of you are sending payment instructions by email. That just seems crazy to me in this day and era that we work with. And it just surprises me that everybody doesn't require certain kinds of phone calls to do that kind of thing. There's also phone porting, sim swap. There's a vendor service available. It's in the US that can try to figure out if the caller is calling from a line that was sim swapped or forwarded or phone ported. And so that's another protective thing.
And then the last thing, of course, is payment characteristic pattern recognition that tries to figure out if what you're trying to do is out of character. And that would obviously trigger a whole other set of approvals. So I Altman's right that banks relying on traditional voice authentication methods only by people may have certain risks, but I'm encouraged to see the number of tools that are being developed to potentially offset some of those things.
And then the last question is where and when in your next fishing trips? So let's see, at the end of August, I'm taking my Hobie Mirage Outback kayak to British Columbia to fish for sturgeon. I have a picture here of last time I was there during the coho run. It was great. And then I planned to go back in November, trying to convince George Gatch from asset management to go with me to fish or tarpon in Trinidad and Bocas del Dragon.
So anyway, that is the end of the Summer Mailbag Eye on the Market podcast. And stay tuned in early August, at some point, maybe you're on the 10th or the 12th. We're going to have our healthcare piece come out. Thank you very much for listening. Hope you're having a great summer. Bye.
(DESCRIPTION)
Logo: JP Morgan
Previous Episodes and Research
Economy & Markets
The Maltese Falcoin
On cryptocurrencies and the blockchain
