AI Is Hiding A Weak Economy, Argues Lyn Alden

The issue is that many of the AI companies themselves are not profitable. That's obviously a concern because you can always have very good growth numbers when you're selling $20 bills for $10. We could see the rollover in kind of the for the price and then capex maybe follows by a year or two. If I had to give a base case, it might be something like that. We've already seen credit default swaps of some of these AI related bonds. That's an early signal the market saying, are you are you guys sure about this?

>> Whoever says Hosi is do you think they're still alive?

>> That's a good question, but my my guess is

>> today's episode is brought to you by Vanex Rare Earth and Strategic Metals ETF. ticker remx. Later in the show, you'll hear about the growing investment opportunity behind rare earths and why these materials are crucial to everything from smartphones to defense. Let's get into it. Hey everyone, to help you navigate this long conversation, Lynn talks about Fed liquidity at the 24-minute mark, about Bitcoin and who Satoshi is starting at the 33minut mark, and about whether AI is in a bubble at the 76 minute mark. Enjoy.

Very happy to welcome back to Monetary Matters, Lynn Alden of Lynn Alden Investment Strategy. Lynn, good to see you.

>> Always happy to be back.

>> Lynn, I'm going to get into many topics with you. AI, Bitcoin, Fed liquidity, everything. But just want to start off, Lynn, is is the economy bad? Is the non AI capital expenditure economy bad? You know, I just looked up what would the GDP growth be if there was no capital expenditure on data centers and AI? And according to Jason Ferman, the economist, who actually I've interviewed, it would be 0.1%. Whereas now the real GDP now cast from the Atlanta Fed for the third quarter is 4.1%. So there's just a huge divide and you know a majority of the growth in this country is from one industry data center capex. So is the economy weak other than other than data centers and are you concerned about the part of the economy that is not data centers?

>> The short answer is yes and yes. Most kind of indicators are pretty weak. I wouldn't bad I I would say they're weak more than they're they're bad. Bad is a more complicated term. They are generally speaking not catastrophic but certainly not good. Um so and I would even add to that things that are not on either the right side of AI capex or fiscal deficits. Those are really kind of the two big stimulus areas. Uh and then the weak areas is is most other things. So things that are on the wrong side of of tightish monetary policy. So commercial real estate, residential real estate, less liquid, you know, types of private equity, some some pockets of venture, a lot of that type of stuff is on the weaker side. We've had a long period of flattish PMIs, both manufacturing and service, and they're not like so low that they're you normally see in a recession, but they've had an unusually long period without kind of a reflation cycle in that regard. So there's lots of headlines around restoring manufacturing. When you actually look at the manufacturing PMI, it's it's just chopping along in like a mildly stagnating way for for years now and really no change here in 2025. So a lot of a lot of weakness. Now that's weakness built on a somewhat robust frames. I think banks are pretty well capitalized. They've gone into this with a lot of kind of risk mitigation in play. And so I I view it as kind of less likely to just completely fall out of bed uh in many of these areas. But there's this kind of like longer grind of weakness. And then on the other side is it's partially held up by structurally large fiscal deficits combined with the obviously the very important AI trend. Uh and so it's a I would describe it more as that twospeed economy rather than a weak economy because you have pockets that are weak but not necessarily catastrophic. You have a couple small pockets that are basically catastrophic. I mean commercial office real estate is a slowmoving train wreck. Um but for the most part you have stagnation and then you have boom and then of course it also triggers along class lines uh because a lot of the weakness we see is among lower income segments that are generally not on the right side of of of AI and then they're also often ironically not on the right side of fiscal deficits. You know there's obviously some portions of the fiscal deficits that go to that group but a large swath of the fiscal deficits go toward social security, Medicare, defense and interest expense. So that's not that that kind of big huge chunk of the pie chart is not really going to the the lower income. And so that that that's kind of the segment that's struggling the most.

>> Yeah. I mean, Lyn, I saw a chart. I'm sure you've seen it. You probably made this chart that but that the the net expenditure that you give or get from the government over as you age and go throughout your life. So from the ages of 20 to 55, you were just a massive taxpayer to the government. From zero to 20, you you get money from the government for childare and ch child tax credits and stuff. And then when you you are a senior 65 and over, social security, you are just a a massive recipient of government stimulus. And like that system, social bargain kind of worked maybe when the demographic pyramid looked like 1960, 1970. But I'm just looking at the sentiment out there and I think there are a lot of young people who are not happy with the current arrangement and they're like, whoa, there there's way too many of you to to be doing this.

>> Exactly. I think you touched on the demographics is is probably the biggest issue that it's a much more topheavy system than it was when it was designed. It was kind of designed with the idea that every generation would be bigger than the prior one, which of course has not not been the way that it's really gone. So, you have a more topheavy entitlement system and then you add on a couple extra things, which is generally speaking, you need more student loans to go into the workforce now than you did in in the decades prior. So, you kind of start from a deeper negative. And then two, it's it's demographics, but it's also kind of the skill with which we handle some of these systems. For example, in Japan, they obviously have have worse demographics. But for example, despite living longer, they spend less per capita on healthcare than we do for a variety of reasons. United States has the highest per capita healthcare costs, which is particularly relevant when you have an aging kind of worsening demographics population. So not only do we have this kind of demographic skew, we we then also kind of add fuel to the fire by having very high healthare costs and other frictions along the way. So you get this right now we've kind of the the the social fabric and the the social implications I'd say are weakening faster than the broad economic data because the broad economic data has that kind of higher income group and then those on the right side of fiscal deficits or AI capex kind of holding holding things up.

So if a majority of the economic growth in the US is from artificial intelligence and data center growth and a a majority of the corporate profits in the S&P not necessarily the corporate profits but the corporate profits growth is is also AI capex. Is the US economy and the S&P 500 just a bet on on AI now? And if so how do you feel about this bet? Lynn, as a someone who's trained as an engineer, what do you think about AI? And do you think that the massive sums being poured into this space, $400 billion this year, maybe a trillion dollars in 2028, something like that, I don't know, is going to be is going to be worth it from from a societal sense, from an investment sense as well?

>> Yeah, it's a great question. I mean, we could have a whole podcast on that topic. Uh, so it's hard to, of course, summarize that. The short answer I I I I think that it's real. Like I think that basically the economic the long-term economic growth in this will be substantial but like many things I think there will be bubbles along the way. Uh that was of course true for the internet where people said it's going to change everything and they did. It just it took took you know more time than some of those initial investments uh thought. For example, those laying the you know, the fiber optics cables. Uh when we look over say Bitcoin over the past 16 years it's both true that at times it's a bubble but that it's also been structural uh from bubble to bubble. Uh and so I I kind of view AI similarly that it it can get over its skis for periods of time, but that when we look back five 10 years from now, we will say yes, this was quite transformative and that it it it changed both the economy and kind of the nature of a lot of things. From a I guess kind of separate finance and engineering. So we look at it from a financial perspective. Obviously the issue is that many of the AI companies themselves are not profitable. uh it's very much funded by VC which is normal in an earlier stage of of a technology. But when you're talking about kind of some of the biggest private companies in the world uh still running unprofitably and importantly not really even having a forecast to get the profitability anytime soon that's obviously a concern because you can always have very good growth numbers when you're selling $20 bills for $10. But if your growth just comes to a halt when you when you aim toward being profitable and stop being kind of that externally funded system, it can it can reverse quicker than people think. So obviously Nvidia is making money and the uh you know the the big kind of hyperscalers uh are making money and it's really kind of the other areas that are unprofitable and therefore funded more by speculation uh about what they will do in in say over five years let alone not even really have tracking toward profitable in the next five years. saying that's a that's a reasonable concern right there that things can reverse quicker than people think even if even when you still have a very real trend. The engineering hat in me basically separates data center AI from portable AI because a lot of people's mind when they hear AI they also immediately lump in robots with them >> and uh one of the key things I think is is very relevant is there's a huge difference between like intelligence and portable intelligence uh which is true for a lot of things basically whenever you're taking energy and you're making it portable uh in this case when you're taking processing power and making it portable that's a radically different conversation. Uh and so I think that the a a the data center portion is very much real in the sense that there's a good chunk of white collar work that much like how there is a lot of blue collar work that we were able to displace uh with automation and basically make that each each remaining worker can therefore produce a lot more because they're basically overseeing and assisting machines in terms of making and doing things. I do think that in white collar work we already are seeing and will continue to see that type of thing where uh many types of white collar work not all of it but many types of it will be learning to work with and overseeing machines AI machines in this case that take things that used to be pretty labor intensive and make them much less labor intensive because we're outsourcing that to data centers now in the field auto like robots and things like that that's a much higher bar like when you look at data center when they when when you look at say like say Ray Kurs Kurswheel the whole you know the singular and all that he kind of maps out how much how much neuron activity is happening in the brain and then you can look at hardware over over years and decades and see you know what does it kind of cost uh you know when when are our computers reaching what the human brain can do uh and it's not it's not quite apples to apples because a 100 years ago a calculator could do more uh linear calculation than a typical human brain could do even as obviously it was way worse in virtually every other area. So it's not like apples to apples, but over time computers and especially supercomputers have kind of reached roughly speaking the the closest estimates we have for how much processing power happens in a human brain. The the key difference though is that the brain runs on something like 20 watts. It runs on less than a typical incandescent light bulb whereas that level of processing power in a data center is measured in megawws, millions of watts. And so that's not obviously not something you can in any sort of investable time horizon put into a portable system, right? So so in the field you're going to have a lot more shortcomings. And it's the edge cases where I think there's a huge difference. Basically that that overseeing role. Humans can automate things but they still are the solvers of edge cases because that's where those more nuances matter. So I do think that there will there will be disruption. I do think that there will be change, but then I do think that for at least for a long period of time, we're going to hit a plateau where we kind of absorb the changes that have happened and look out long enough, anything can happen, but in any sort of re like investable time horizon or career time horizon. I do think it's real, but not quite to the heights that maybe the the hyperbols would would say.

>> Yeah, thank you. Well, the hyperbulls think, you know, AGI AGI is coming any any day, any any week. So Lynn, explain that concept of difference between intelligence computing power and portable computing power because you know on my iPhone so many of the computations are done in a in a cloud in a data center and that's been the the huge transformation of the past 20 years is but yeah basically all the computation happens in in the data center not on the device and as such you you can do way more way more powerful things. I don't I don't I don't know but is that how a robot works or is all the action in the robot's chip uh you know in its in its brain?

>> Yeah, it depends on the type. Um there can be and there can be a mix of both. You can do local processing and off off local processing back to the data center. The challenge is this one is is obviously speed and bandwidth. anytime you're doing processing that is not local, you obviously, you know, you're going back to a data center where there's more resources, but then you're also introducing lag into the uh aspect, which if you're looking up something on your phone is fine. Uh if if you're a robot trying to deal in the field, that could be obviously more challenging that that speed can be an issue. Uh number two, your phone is often doing non-critical things. Meaning obviously phones are a huge part of our life, but it's you it's not going to be life or death if if your phone has a period of time where it it goes slow or that the the connection like the the data is is stuck or limited in some way. It's not as though that a car is going to crash or a robot that's doing something pretty critical is going to have like a a meaningful loss of property or life. And so we introduce those things. A lot of that more more more generally speaking has to be local because it has to be able to not get itself into a situation where losing data has very costly in many ways ramifications. So the kind of the ideal situation is you have something that's smart locally but that can call back to a data center for edge cases but that's to limit it to things that can only have major losses if its callbacks fail in one way or another. So I I you know I'm long-term bullish obviously on almost any type of technology you can imagine. Like I do think that robotics will be a bigger part of our life in in years and decades. Um I do think that a lot of white collar work will be you know automated and and our nature of of work will shift in that area. It'll become a a common part of our workflow. Uh and it already is uh for for people that have been on this pretty early. But that there's still I mean there's a fundamental difference between the types of thought that AI does and what the human brain does. I would say a AGI is pretty fundamentally different uh than the types of AI we've seen and that there's still kind of a lot of work to do in the robotics. I mean, one of my go-to things uh is for decades, they've been trying to make robot vacuum cleaners work, right? Uh and of course, they you know, they they quote work, but the the technology is not not even gotten to the point where it's so dominant that it just replaces, you know, manual vacuum cleaning. Th those companies have actually struggled if you look at some of their stocks. And that's a fairly simple thing overall because you have a for the most part a flat surface, occasional stairs and things like that. You have a pretty defined goal. Clean this flat surface with some stairs and stuff in a home where a human can occasionally help you with edge cases. And yet that's still been a struggle to make that economic reliable not encountering so many errors that it just the human just gives up on it and gets a normal vacuum cleaner. and and even that basically has been slower than expectations, let alone far more expensive, far more complex things. So I I do think that a lot of this will be solved. But a general rule in engineering is like hardware is always kind of way harder than you think, especially compared to like software. Obviously there are very challenging software problems all the time, but hardware like the the cycles with which you you try something and then and then correct that and try again and correct it again. Those cycles are much slower than they are in the world of pure software. Deploying things into the field, getting data, doing the next generation of things in the field. It it's often a much slower process than people think. And when you compare it, I mean, when you pair it with kind of Moore's law and just overall getting processing at the edges uh in a way that is both fast and reliable compared to the the the criticalness of what it's doing are some pretty serious limitations. Um, so yeah, I would describe myself as a moderate bull on AI and a bigger bull on things that are more readily put into data centers and more of a longer term bull, like basically less bullish on things when you're applying that in the field, especially in a non-controlled environment. So not in a manufacturing facility, but out in the real world.

>> And how is this making you think about investing in AI? And I'll just throw a few sectors out there. Obviously there's the semiconductor chips um most notably Nvidia there's the data center providers that that build them concrete makers all that sort of construction equipment there is the cloud computing companies Microsoft and Google Amazon as well which are investing a ton of money but they they have apparently demand for it that is kind kind of locked up right right now so it's slightly less speculative than a company like Meta which is investing just because they think it's going to transform their business which it very may will and then there's the the the very speculative which is open AI which is uh you know losing billions and billions of dollars and all that sector as well as China Alibaba is a is a leader there and the Chinese model has pretty impressive AI with way fewer investment way way cheaper chips and the like. So just that entire segment, how what's a bubble? What's not a bubble? Which do you like? Which do you think is over overhyped or just avoid it all? What do you think?

>> Good questions. I mean I I've been at times long and bullish on the chipmakers. Other times pull back when when they get I in my view over their skis. Uh I the short is I tend to like these profitable areas. So the chipmakers have been profitable for the most part. And I've also I mean I've been for example long Alphabet even though there were early concerns that they might actually get disrupted by AI. I've been on the camp that they're that I think they're pretty well positioned there. Obviously I monitor it closely. There are a couple reasons for that. One is that AI was mostly disrupting like nonprofitable types of searches, right? So if I want to if I want to look up what was Napoleon what was Napoleon's wife name, right? That's something that Google or AI could tell me the answer and Google's not going to make money selling me something. based on that me looking that up. It's just a fact I want to know. Whereas if I look up plumbers near me, Google very much wants to have that business. Um, and AI in many cases was disrupting kind of the former. Uh, it's a it's a super useful search engine. It makes search way better in many cases, sometimes worse, but often better, but it's not really disrupting Google's kind of core ad business in that sense. They've also been one of the leading players in AI kind of they don't get as much obviously focus as as Open AI does, but they've had they've had kind of a leading position there. They've obviously got a lot of profitable firepower to to back it up. And that also my view is that AI and and basically makes video generation uh less expensive. I think we're going to continue to see a growth of video content and Alphabet owns the biggest along with Tik Tok. YouTube is like the biggest video platform out there. They've actually been taking pretty decent market share in paid video, but even just like amateur creation, I mean, people use AI to do podcasts, like edit podcasts. They use AI to translate things. They use AI to help with graphics. Those higher bandwidth types of things are historically expensive. AI can compress the cost of doing so, which makes more of them. And if you run the biggest platform that that benefits from them as long as you don't get heavily disrupted elsewhere, it's interesting. So now Alphabet might be a little bit richly priced, but I I tend to fade the most bearish narratives on it. Uh so one of the ways I I take this is I I first I ask things like what benefits from AI, obviously chipmakers and things like that, but then I also say what is not disrupted by by AI but is cheap. So it's kind of like an flipping the question around. So one of my answers was was alphabet. Also for example MIT did some study. I believe it was MIT and they released it and they kind of showed how kind of small the ratio now of corporate trials for AI have kind of had firm results where they can say yes this saved us money or boosted our quality. It was it's kind of like a pretty low percentage of of AI trials did that. Even though obviously in the consumer level most of us use AI now and we say obviously yes it it it transformed a lot of the way we do things. when you try to apply it to a bigger setting, there's more frictions for doing it and they they have harder to say yes, this saved us money here and there and but one thing they did find is that AI was very helpful on the back end. And so one thing I I look at is like kind of regulated or monopoly type of businesses that can use AI to lower their cost on their back end but that are protected on it on their front end. So for example, banks or insurance companies. These are these are kind of in some sense capital intensive, not in the physical way, but in the in the way that obviously you need big balance sheets to do them. They often are operated sections of the economy. They've got back-end expenses that are some percentage of that is automatable and yet their core service is actually pretty challenging to disrupt. And to the extent that they are gradually disrupted, things like stable coins that that I'm I'm long-term bullish on, many of them can integrate those into them. So, I I I approach it both from what can benefit and what is likely not going to be either impaired or impaired as heavily as pricing would imply.

>> And what do you think about the so-called neoclouds that buy a ton of compute by by data centers and then rent out the compute? So the core weaves of the world, the nebuses of the world, and I don't believe that's Iron's business model, but and then there's the other thing of former Bitcoin miners that are now supplying all this electricity.

>> So the short answer is the part that interested me there is the ones that actually have the real resources, the ones that have the basically the property or the rights to to re like lowcost structurally lowcost energy for example that then incorporate that into their data center business because that's where I think the competitive mode is. I think energy is actually one of the the biggest moes here. And so kind of like how in in finance if you have access to the money transmitter licenses and all the regulatory stuff that's a very expensive and hard to a sale position compared to ones that are kind of renting that from others and using them in some way. I view similar with with data centers which is the ones that have kind of the root hard to replace thing I'm pretty constructive on. That would include a lot of those former Bitcoin miners or ones that are in many cases still mining but have shifted more and more of their business toward AI. Many of them at least the ones that I think are interesting are the ones that actually kind of have kind of structural access to that lower cost energy. Whereas those that are just kind of using that ability temporarily that can come and go. And I'm not probably the best person to ask on that but I can basically say where I'm interested which is those that have the kind of the the harder to replace resources.

>> So you like that space. You're also, I imagine, a bull on natural gas and power.

>> Yes. Yeah. Natural gas power. Uh the the businesses that even get that natural gas to where it has to go. Uh I think all all that infrastructure is is super interesting and and the resource itself. And we've had obviously we've had this long stretch where natural gas was kind of cheap relative to oil in terms of when you price it based on its energy content. Uh in large part because natural gas is less funible than oil. it's easier to to move oil around to to solve really big pricing gaps. Whereas natural gas, it's obviously much more infrastructure intensive to to move it. And so those gaps are are much slower and more costly to arbitrage. And so we've had this really long stretch. And I think over time that closes that natural gas kind of gets priced where it should be, which is generally higher. And natural gas kind of assets I think are are obviously very well positioned.

Hope you're enjoying today's interview. Rare earth metals are the hidden backbone of modern technology and defense. They power everything from smartphones and electric vehicles to wind turbines and fighter jets. Countries around the world are racing to build their own secure supply chains, driving investment focus into the companies best position to reshape this sector. Investors can gain exposure to these critical materials through the VANC rare earth and strategic metals ETF, ticker REMX. Go to van.com/remxjack to learn more. That's van.commxjack. Now back to the interview.

Thank you, Lyn. Now to to switch tax entirely. Uh, as I promised at the beginning, what's your current view of the Fed's balance sheet and its role in managing liquidity? In late October, we saw some pretty significant spreads of 10 20 basis points secured overnight financing rate above where it should be, indicating that the Federal Reserve was beginning to lose control of of interest rates. What's your view there and its consequences?

>> So far, this is going roughly what even the Fed themselves thought it was going to do. So, they they the New York Fed released reports over the past couple years like annually where they kind of predicted what their balance sheet's going to do and they kind of saw it gradually slow like the the decrease in the balance sheet slowing down by late 2025, maybe early 2026 and then from there growing in line with nominal GDP. that that's kind of how they've roughly seen it going. And we are running into liquidity frictions roughly where they already anticipated that their their pivot is likely to be. And now that's a pretty big window because they they weren't sure exactly where it would happen. They had kind of loose estimates for where it might happen. They have that framework of scarce reserves, ample reserves, and abundant reserves. Overall, this this hit a little bit, I think, earlier than expected in the sense that we're kind of comfortably in ample reserves. the way that they estimated and yet some of this activity points more towards scarce reserves. So perhaps a little bit sooner than they expected, but nothing particularly shocking and that's also why they have these standing facilities in place, the standing repo facility and reverse repo because they've learned from the September 20 2019 repo spike. Um, they've already got these facilities in place. They basically structurally like permanently increase the funggeability between cash, treasuries, agencies. These things are all pretty swappable without kind of major crises happening. And so, and I think part of the reason why this happened a little earlier than expected is because of the TGA. And I'm not the first person to say that, which is that partially because of the government shutdown, they overfilled the Treasury General account. So, they were targeting 850 billion. They temporarily overfilled it to over 950 billion. When when they do that, that basically sucks capital out of the financial system and into a void. Uh and so the more capital they have in that void, the less capital there is elsewhere, more less liquidity, I should say, that there is elsewhere. And so I think over time they're going to push some of that back in, but that we are we are roughly running into not only stopping the the QT, they've obviously reported that, but that I do think in the first half of next year, we're going to go back to structurally increasing the balance sheet. The a couple areas where this is noteworthy is that one, they're going to be likely increasing the balance sheet when we still have officially above target inflation, which which complicates certain types of discussions around it. And two, we're going to be going back likely to period of balance sheet increases from a higher base in the sense that prior to the global financial crisis, the the monetary base would gradually grow, but it was pretty small relative to broad money because you had a higher kind of fractional ratio. And now you're going to kind of probably go back to that environment of a gradually increasing monetary base rather than kind of the boom and bust you saw in the QE and QT era. But that that whole thing is a much higher percentage of nominal GDP or of broad money supply. And of course it's interest bearing in many cases because that that's part of their monetary policy toolkit to to maintain interests where they want. So that brings kind of political ramifications basically paying out money to to banks. And so it's kind of a more politically fraught environment, but that very little of this is actually surprising. It's mostly just managing the details like what quarter things happen in, what speed things happen in, and I think we're going to go back to a period of gradual balance sheet increases. I see people talking about their expectation that quote unquote something's going to break or we're going to get a really big like ramp up in liquidity of some sort. I I tend to fade that view. I think that that we're going to get these little kind of micro breakages that are already kind of have facilities to address and that those are kind of the warning signs for them to go back to a very very kind of mild longer term liquidity injection.

>> and what are the consequences of this environment where the Federal Reserve is in a small way expanding its balance sheet as compared to the past three years three plus three years and over where it's been shrinking its balance sheet through quantitative tightening or QT.

>> So short answer is not a ton because when they were decreasing their balance sheet they were doing it in a period where there there still was abundant reserves and abundant liquidity because you had a $2 trillion reverse repo facility. So the overall kind of net liquidity has basically been flat for 3 years like the base the base liquidity was flat for like three years and whenever it kind of got to the lower end of its range. So, one one of those periods was kind of March 2023, kind of going into the regional bank crisis, that was kind of a low point in its range. This recent period, this kind of TGA refill was another low point on its range. When they get to the low point of that range, that's when when they tend to run into these frictions and have to then kind of overshoot in the other direction a little bit, but that when you kind of zoom out, it's roughly been flat for three years. And so, I think that that what's noteworthy is I think we're going to get out of that flat absolute range. we're going to go to a period of kind of structurally increasing again. On average, generally speaking, that that's better liquidity, good for asset prices. Now, that's complicated by the fact that obviously AI and other things have already run in asset prices. So, it's not necessary expensive, crowded, unprofitable or even profitable like in the Mag7's case in in their view. It's more of like the things that have been stagnating for a while, I think, can start catching a bid once we once we enter that period of balance sheet increases, which we're not there yet. So, I think kind of the the strains and liquidity we've seen lately. It's not shocking that some of these like unprofitable kind of hyperrowth stocks have been off pretty substantially off their highs in many cases. It's not too shocking that that Bitcoin has had a little bit of a correction here. And I I think that that kind of turmoil probably continues for a period of time. But when we get to 2026 and specifically whenever we get to the phase where we're actually going back to kind of a structural increasing balance sheet, I think it's on average supportive of those types of assets. But with a caveat, of course, it's it's one variable among many. There's also what's happening with the economy, what what's the Fed doing with interest rates. But that's another thing is that people often thought that they couldn't go back to balance sheet increases till they get rates to zero. But that's not the case. the the balance sheet was increasing before the global financial crisis even when they had positive interest rates. It's a different environment now. Like I said, it's a much bigger share of the economy and and a bigger share of kind of the money system than it was before. But that basically the balance sheet and interest rates are two very different kind of levers going forward rather than balance sheet expansion just being something you do after you've already kind of exhausted the interest rate channel. So, I think we're going to enter a period of structural balance sheet increases while rates are not zero and don't really have a reasonable expectation of going to zero anytime soon.

>> And the way that the Fed would explain that is it's not quantitative easing of large-scale expansion in order to ease financial conditions. It's just increasing the amount of reserves so that the banks can function in the regulatory regime that the Congress and the Fed has has created for them. So, that's I think how the Fed would explain it.

>> Yeah, they would describe it as a technical issue less rather than a monetary issue. I would take and it's it's partially right. I would take some issue with it because to the extent that they're growing their balance sheet or going to grow their balance sheet forecast to grow their balance sheet with nominal GDP, a problem of course is that if the government is running structurally large fiscal deficits that directly affects nominal GDP and so you kind of have that recursive function where the Fed's balance sheet is the speed with which they eventually start going back to growing it. will partially be dependent on what Congress and the president are doing with spending, which is it's just a structural thing that's that's it's kind of an aspect of fiscal dominance. is kind of part of the nothing stops his train view that whenever the Fed runs into some sort of liquidity issue with these core systems. So in inter interbank interfinancial institutional lending or the treasury market they're going to step in and like the same day either with a standing facility or some sort of decision based on running into these liquidity kind of guard rails and they're going to accommodate and even if they it's the case that part of the reason they need to accommodate is because we're running pretty big structural fiscal deficits. So, we're still in a case where bank lending is not particularly aggressive right now. It's not like a rapid period of loan creation. And instead, it's still the case that that fiscal deficits are fueling a lot. And the only pace we really do see lending and and kind of stuff is is that AI area. So, it's really only the only two games in town, fiscal deficits and AI capex.

>> Right. Lyn, you mentioned Bitcoin. The the draw down is getting bigger from a few months ago to now down 25% below 100,000. what's your outlook here? What do you if anything is you think it's driving this this decline?

>> So, it's it's funny. I think the the bigger topic is the length of consolidation more than the draw down itself. Because, for example, even in like the explosive 2017 bull run, Bitcoin would have multiple 30% corrections along the way and they'd be the the kind of these brief violent dips and then a explosion upward and another really brief violent dip. So, Bitcoin merely having a 25% 30% correction is like a normal Tuesday historically for Bitcoin. The the more interesting fact I think is the fact that it's roughly flat from a year ago. It's been kind of a very noisy type of flat like often it it's intraday it touched something like 75K back in April. It it's peaked at over 125k. So, it's had a really big range, but that range is roughly centered around 100k for a year. And that's more I think what bears people up which is that when you look back two three years Bitcoin's had incredible runs but in this past year it's been a really choppy volatile wave going nowhere and that is of course very uncomfortable for a lot of bulls. I think a broader issue is when you look at the full crypto space, I I think it's completely almost completely exhausted of narratives and I I've definely been a bear on things other than Bitcoin and stable coins at least in any sort of like s like macro scale things and so I think that there was the there was the ICO period where it's like okay these coins are going to make ICOs better and then obviously that has all sorts of regulatory issues and in general is not there there then you know obviously NF s and kind of various things like that. There's DeFi for a while. These are all these kind of narratives that would fuel an entire cycle. Then it was literally just meme coins. It was kind of the most cynical narrative which is just it's basically crossber digital gambling. Then people even eventually kind of fully saturate that and say okay they're obviously most people who participate that in are going to lose money and there's really kind of no there there for a lot of this. Now there obviously something has to be on the back end of stable coins. something has to be on the back end of the handful of things that that do have substance, but those are pretty small tech rails. Those are not multiundred billion dollar kind of things. And so I think that it's partially weighed down by the fact that you have all this kind of baggage throughout the space that really needs to kind of gradually or in some cases not gradually deflate and become dimminimous while I think that that Bitcoin and stable coins are still structurally interesting. So my general view is that I think we'll see higher highs in 2026. I don't I don't have an opinion if it's early 2026. I I tend to air toward giving myself more time. So let's say late 2026. I I genally try to avoid giving price targets for Bitcoin in 2025 when quartered into one. I would say 150K as we only got to 12.

>> You said 2025. You meant 2026.

>> No, actually this one when I was quartered like I'm talking retroactively now. When I was cornered into saying what I thought Bitcoin would do in 2025, I'd say, well, you know, I think anything under under the 150 would be kind of disappointing, I wasn't giving those really bullish price targets, but we only got to like 125. So, we'll see what happens in 2026. I don't think the four-year having cycle is playing a role. So, there's some people selling just because they think that we're on the four-year having cycle with the part where you're kind of supposed to sell. I think that's a narrative that'll be gone next year. And it's mostly going to come in my opinion down to liquidity and then also how much of the AI thing is still kind of sucking capital in that general direction. So I think the over the over the structural overhang of the broader crypto space I think is going to be here for a while. It's kind of a downward variable kind of just pulling things. Uh but that I do think that the liquidity situation will improve next year and I also do think that the AI narrative will probably have lost some of its steam and therefore kind of some of that capital can shift back to to Bitcoin and elsewhere.

>> And two big buyers this year and over the past year as well ETFs as well as now this year Bitcoin treasury companies the the biggest one Micro Strategy. How significant have those two bids been for the market? And do you think that they will continue to be a force of strong buying in the future, both ETFs as well as those those Bitcoin treasury companies, or do you think that game is kind of over?

>> So those were basically the two biggest sources of demand this cycle, you know, specifically kind of last year and this year is the ETFs and the treasury companies. So on one side you had those accumulating pretty aggressively and the other on the other side this you know you had longerterm holders selling into that strength which actually happens kind of every cycle. So every time you have like way new highs tons of liquidity you do get that kind of it's kind of like how a startup company you know when early founders start to kind of uh as as the asset gets more liquid and tradable and and you know something that was 5% of their net worth is now 50% of their net worth because it has so much appreciation. they start to either rebalance or upgrade their consumption some matter some combination of those and you get that kind of distribution of coin. So you have buying of Bitcoin ETFs and treasury companies selling from OGs like the longerterm holders. There's very limited kind of raw retail demand. Although it's important to note that a sizable chunk of the ETFs and the treasury company demand is kind of retail by proxy. It's both institutional and retail demand for the asset but with an intermediary in this case. They either want levered Bitcoin or they want Bitcoin in their brokerage account and their 401k and their IRA and things like that. So I think that when we look back, we look over the next five plus years, I don't think that bid is totally done. Like I do think that there'll be there'll be more pools of capital uh including institutional pools of capital that will likely allocate to

Bitcoin, but I think the rate of change is probably done. In, or at least it's seen its high watermark. In the sense that, you know, I don't think three times MNAV for a treasury company usually makes sense. Or, you know, kind of the just the explosive growth. The big, the BlackRock's Bitcoin ETF was like their most successful launch ever. I think that rate of growth, that was kind of like a lot of pent-up demand happening at once. I think that is now behind us.

Now, obviously, we have a correction, and I think eventually we find a bottoming, go up. But that that goes up at a more gradual pace, most likely, because it doesn't have that new factor anymore. It's already been tried, tested, found its limitations. And I think we kind of, it just becomes a more organic structure rather than this kind of boom bust. And and so, so you're kind of describing Bitcoin as an asset class that is maturing. And as asset classes mature, they get less volatile. They get maybe a little less interesting. It's a little more boring. Boring can be good. But what do you think is going to be the the new factor driving Bitcoin in 2026 and the the years going forward? Obviously, you don't have a crystal ball, but what are kind of your your top few candidates for that?

Yeah, short answer, don't have a crystal ball. I think the the the advantages for Bitcoin next year, one is I think better macro liquidity. And two, I think that the ending of the four-year having cycle narrative will actually benefit Bitcoin this time. Because I think right now it's hurting it. Because like I mentioned before, people like it, it peaked in quarter four 2013. It peaked in quarter four 2017. It peaked in quarter four 2021. People are like, well, it's got to peak in quarter four 2025. And actually 2013, I'm not sure if it's quarter four, but the the last three were kind of quarter four on that four-year cycle. So, there's some selling pressure simply because of the narrative that the cycle's over. And it's it's kind of a self-fulfilling thing for a period of time.

I think that once Bitcoin finds a bottom, and we're in a fresh year, I think that that is when people kind of maybe realize that if it's the case that it's not over, and people in some case have to buy back in if they want to be on that train, uh, that I think is a a catalyst upside. Now, whether that happens in the first quarter of next year, or maybe it's already in, or maybe it happens second quarter, that's where I would not try to guess. I I do expect that we'll see like the lows in either either this quarter or probably the first half of next year. And then I I do think we see higher highs, 2026, maybe 2027.

I do think that this this the whole cycle I think is different now. And the main reason for that is that Bitcoin mining rewards, so new coins are a much smaller share than they were in those prior cycles. And the much bigger question is what what do you have to do to price to pry coins out of longer-term holders? When you drive it it up up enough, they do regularly sell into that liquidity, into that strength. They are at that point very overallocated. And people that bought 5, 10, 15 years ago, they have families now. They have very large percentage of the net worth in it. And they do distribute their coins. And that is by far the bigger factor than the having in any sort of like four-year cycle. So I think that that breaking that narrative is potentially one of the bigger bullish catalysts over say 2026 and 2027.

So the Bitcoin miners, every having cycle, the amount of rewards they get when they solve all these mathematical puzzles is is cut in half. And as such, for a while, what like seven, eight years, it's been a very tough business to mine Bitcoin. And I don't know the exact number, you can tell me, but you know, unless unless you have basically a supercomputer, it's unprofitable to mine Bitcoin at at current prices. And maybe even $125,000, I don't know. But it's a very, very tough business. Who is going to be verifying all the transactions if if no if if it's not profitable to mine?

So the short answer is you need ASICs. You need application-specific integrated circuits to solve it. So it used to be that you could do it with a CPU. Then you needed a GPU. And then you need literally a specific device that is manufactured only for this purpose. That's the most efficient way to do something with a very narrow scope, which is which is in this case Bitcoin mining. There's still different types of miners because there's like the, there's the really big ones that buy like the expensive new miners, like the new equipment. And they operate pretty big data centers. And they generally have power purchase agreements where they say to a power company, give us your absolute cheapest rate for electricity. And in exchange, we'll be the first to shut off anytime you have a a a shortage of any type. So we're we're your most flexible buyer. In exchange, we want your best rate. Whereas something like a hospital says, we don't care about your best rate. We want always electricity to be on, no matter what. We want we want we're the last customer you shut off. Right? So the Bitcoin miners on the opposite end of that spectrum. And that's that's their business model.

Then there's then there's kind of the scrappier Bitcoin miners that go out to like oil wells where they have stranded natural gas that they literally just light on fire. They flare it. And they say, "Well, instead of lighting it on fire, let's just put let's just have this little cart full of Bitcoin miners on it, attach it to that." When you add them all up around the world, that's actually a huge amount of spare energy. There's actually multiple times more energies worth of natural gas flared every year than the entire Bitcoin network currently consumes in terms of power. And so those generally buy older machines. They kind of buy secondhand machines that those other miners no longer find hyper competitive. So they have less capex cost. And it's more about location and things like that. Um, and the way Bitcoin works is that there's a difficulty adjustment. And so if it becomes too unprofitable for too many of these entities to mine, many of them stop mining. They say, "Well, literally we're we're losing money every time you mine. So we're going to either not do more capex and just slowly let what we have stop, or if it gets bad enough, we literally shut off our machines."

At that point, the Bitcoin network slows down to some extent. The most extreme one was when China banned Bitcoin mining. Something like half the network pretty quickly went offline. You start to get slightly longer block times. So instead of 10 minutes, it could end up to 11 or 12 minutes. In severe cases like that China ban, it could be 18 minutes. And then every two weeks or so, specifically every 2016 blocks, there's like a control loop that says, "Okay, because mining speed is slower now, we're going to make it easier to mine." And that kind of rewards those that are still kind of borderline profitable. And it actually makes them more profitable for them. So you always kind of find that equilibrium. And the short answer is who in the long term who mines Bitcoin is basically entities that can find virtually free energy, meaning that it's stranded in some way. It's either it's either stranded in terms of space, meaning it's it's out there in like North Dakota, like natural gas at an oil well that's not enough natural gas to build a pipeline. And they're literally going to light on fire. And you say, "Look, just give it give it to us for one penny and and we'll we'll do it." Or it's stranded in terms of time, meaning that it's it's like solar or something where it's it's it's on at noon. There's not enough demand to use that all up. So, you get like negative pricing. You have to curtail it. And the Bitcoin miner says, "Okay, we'll mine during the daytime and we'll shut off at night." And so I think that Bitcoin miners got to fill in over time all those little nooks and crannies and basically buy up the the nearly free stranded energy and use that for Bitcoin mining.

That's good to know about the easing adjustment. How long does the easing adjustment last?

It's permanent. So every two weeks roughly, specifically every 2016 blocks, it it changes. And then in two more weeks, it it keeps monitoring what's happening. And if there are still miners shutting off and and just not mining anymore, then it makes a deeper adjustment. On the other hand, if that prior adjustment worked, and now blocks are speeding up again, it might then increase the difficulty to some extent. Over over the long term, because Bitcoin has grown so much, is it more updates have been increasing the difficulty than decreasing the difficulty? This is actually kind of the key thing that Satoshi solved. Because before this, there already were kind of these like systems that have basically a decentralized database. And the biggest question was how do you decentralize who adds the next block of transactions to it? And how do you do it in a way that's doesn't get kind of inflated at once if we apply processing power? How do we deal with Moore's Law? And the answer was this difficulty adjustment. It says no matter how good computing gets, this the system's own control loops says, "Nope. Anytime you get faster than a new block every 10 minutes for like a consistent two-week period, you make it harder." Or in in rare cases, when you get that kind of pullback in price and an overall network activity, you have miners over their skis, blocks are slowing down, it kind of kicks in and makes it easier. So the the answer is it's permanent and at least for that next two-week period. And every two-week period is like a fresh start in terms of of monitoring it to make sure that new blocks happen every 10 minutes on average.

This is this is good to know. I'm sure you know a lot of coin fans watching this like I can't believe Jack didn't know this. Lynn has to educate him. But yeah, I I didn't know this. Obviously, I knew about the having. What about Okay, so I just looked up there's almost 20 billion, 20 million Bitcoin that have been mined. 19.95 million Bitcoin. And there's only ever going to be, what is it? 21 million or 21 million?

21 million.

Yeah, there's only ever going to be 21 million. What happen I think you can know where I'm going with this. What happens if the there needs to be net easing in order to attract miners to do the activity, but that net easing increases the amount of Bitcoin supply. And so what happens if you approach approaching that cap, that gap, as we are, but the then it would then that would slow down? Like, is it possible to get get over 21 million?

No, but but you're touching on a on a related question. So the way the difficulty adjustment works, it doesn't determine how many coins go out to miners for solving the the the puzzle. Uh, it goes it basically is their cost side. So no matter what happens, the same number of coins get created per uh having cycle. Uh, that doesn't change except for every four years when it decreases. Uh, and what changes is the difficulty of the puzzle to get those coins. So supply doesn't change, only the difficulty of getting that. And yeah, over the long term, I think the question you're actually getting to is is when all or nearly all the Bitcoin are mined, how do the miners get paid? Uh, and that is primarily transaction fees. So, to the extent that that people want to transfer Bitcoin, miners get paid both from new block rewards, new coins in other words, and they get paid from, let's say every block can have an average of 3,000 transactions in it. There's more nuance than that. It depends. But like let's say 3,000. And let's say 4,000 people want to be in the next block. Well, the 3,000 highest bidders, generally speaking, are going to get in that block. And the other thousand have to wait. And generally speaking, the lower fee uh in a crowded environment, your transaction is going to take longer if it gets in at all. And the more you pay, the the more quickly you can get out. Um, so if Bitcoin has robust demand uh at some point in the future, you know, 10, 20 years from now, when block rewards are very tiny, uh, it'll be primarily relying on that transaction fee. Whereas this this period of kind of new coins being created is kind of the bootstrapping phase. Uh, it's basically saying that we're even even during periods of low transaction fees, miners are still getting paid with the new coins, but that eventually becomes diminutive. Uh, and it becomes almost all transaction fees.

Okay. And what if we're in a world, which we kind of are now, where a lot of the Bitcoin is HODLed, just kind of hoarded in ETFs or in wallets that don't really move that much, and Bitcoin is less of a currency that kind of greases the the wheels of commerce around the world and more of a seen as a store of value. So, what if there's very few Bitcoin rewards because we're the having gap is so extreme and everyone's HODLing. So, it's it's it's not good to be a Bitcoin miner at all. Then what happens?

Yeah. So that that point it would kind of mean that just Bitcoin structurally doesn't have that much demand. Because the way that the the block size limit works, when you add up how many let's say, you know, 3,000 uh transactions per block, or in some cases 4,000 depending on how tightly they compress them. The the kind of the very like back of the envelope number is that with current block sizes, you can get something like 200 million transactions per year in Bitcoin, which for macro people might notice is actually about the same number as Fedwire. Uh, the number of Fedwire transactions that occur per year, it's about 200, 200 million. And so there's 8 billion people in the world, roughly speaking, 200 million like space for 200 million kind of base layer transactions per year. And so if if roughly speaking, if 20 million people want to have one monthly transaction, that's that's Bitcoin's block space. Uh, any more than that has to exist in higher layers, whether it's a financial layer like an ETF, or whether it's a software layer like the Lightning Network or Charming eCash or various kind of layer 2s that kind of lock uh into that network. So I do think that to the extent that either Bitcoin is successful or unsuccessful, most of it will be on those higher layers. Because that's kind of a important part of scaling. And Bitcoin's failure mode in that context is that in even in the distant future, let's say 10, 20 years from now, if not even a few tens of millions of people want to regularly transact with it and actually take even just even just simply taking custody of it, because that's a transaction. When you when you buy coins on an exchange and transfer them to your own wallet, that's a transaction. So merely if if tens of millions of people want to hold Bitcoin and very occasionally sell or or move it around in some capacity, that's enough to support the fee network. And then anything on top of that is additional fees. So you'd have to have structurally low demand for people that want to hold any coins at all in order to kind of achieve that kind of persistent low fee environment. At that point, because of the difficulty adjustment, the network can still function. And the key risk would be that it'd be pretty low cost to censor it. Government could come and say, "Oh, for a billion dollars, we could just, you know, kind of block all new transactions." Like, we'll go ahead and do that. At that point, so it kind of lowers the the cost to attack it. There's also kind of a difficulty loop there, which is that if it's attacked in some way in that sense, then the the fees to send can spike. And that can actually rekindle more demand for people that want to pay fees and use it because they have to because it's kind of like blocked and you need to encourage new hash power to come onto the network to help move your coins. So, there is somewhat of a feedback loop there. But the short answer is that the failure mode would be structurally low demand. And the numbers there are tiny. Like you don't need billions of people using Bitcoin for it to have a pretty robust fee network. You only need tens of millions that even just want to, and globally, not just in the US, globally, to want to occasionally take custody of it, occasionally move it around, occasionally open a Lightning channel, uh, and things like that.

And how many do we have now?

Roughly that number. I mean, right now there is a there is a fee network. The fees are are usually less than a dollar. Some cases comfortably less than a dollar to move it around. There's been occasional spikes where it can cost $100 in in kind of these really broad brief windows to move Bitcoin. Um, so it does generate pretty substantial fee network uh fee volume. I mean uh so the numbers are there. I think to make it comfortable, you'd probably want to at least triple it from here. Uh, so that you always have a fee network that's a little bit more robust. Ideally, you want to see at least the equivalent of like $5 fees for a given Bitcoin transaction. Because if you have 200 million transactions a year times $5 equivalence, uh, you start to get you get kind of a persistently robust fee network uh that's measured in in the billions rather than the the hundreds of millions.

Okay, thank you. So, you know, Visa and Mastercard, incredibly profitable organizations. They're global payment networks and they are just printing money every single second, taking taking a fee as well, but they're not very capital intensive at all. Their profit margins are very high. Whereas the Bitcoin miners, even now at, you know, it's a high price. Obviously, it's it's down, but it's it's a high price. Certainly when I was in college, it was at $2,000, $3,000. Even at these prices, it's still not super profitable or maybe unprofitable to mine Bitcoin. And all the Bitcoin miners are now going out to sell their compute and their power to data centers. So like is there a world where Bitcoin is $90,000 or even $60,000 or $40,000 where it's profitable to mine Bitcoin as the difficulty adjustment gets gets higher and higher and higher? Or does it require like Bitcoin to a million?

No, it can be profitable at the current prices. And it would just be a case where the difficulty adjustment has likely decreased somewhat uh to basically find that equilibrium. Is generally how it goes. Because it's historically been a growth industry, it's aired toward unprofitability. But if it kind of if it transitions from a growth industry to like a more steady state, that equilibrium would drift generally down toward profitability. So you'd have some on the higher cost curve that are kind of not quite profitable. you have others that are quite profitable. And the middle would probably be kind of like borderline profitable at that phase is generally how I'd view it.

Now it the types of mining that can shift to AI. It there's a thing there's basically there's a couple trade-offs there. One is obviously AI data centers, there's way more capex there. They care way less about electricity costs. They care more about uptime and they care more about low latency. So they want to be generally a little bit closer to population centers. They want to have pretty reliable electricity. And and they're willing to pay for higher electricity in order to make sure that all the all the equipment they spend this crazy capex on is is working as close to 24/7 as they can. Bitcoin miners have opposite economics where it's it's way less about capex. It's way more about electricity costs. They don't need to be near population centers. If you find uh you can be in the middle of North Dakota and you're you're just as good as as somewhere else. You generally want the lowest electricity cost possible, which generally means away from population centers. And so I think that over time these kind of more persistent ones end up doing AI data centers. And the the kind of that rougher sort is what mines Bitcoin. Either ones that are ready to shut off at a moment's notice, so they only mine during the day and then they shut off whenever you have kind of the the peak power loads, or the ones that go out to all these scrappy places in the world. And ironically, some of the scrappy ones are actually pretty big companies because they accumulate hundreds of sites, for example. So, it's sometimes it is small miners, but other times it just happens to be a large entity that's doing that kind of scrappy work. I think that's where Bitcoin mining increasingly ends up is around the edges. Ironically, that's I think a good thing because it actually helps decentralize mining. You don't want like a bunch of really big data centers mostly in one country doing all the Bitcoin mining. To the extent that there's like a a river in Kenya and they're mining Bitcoin there, and there's actually there's companies that do that. Then there's like natural gas oil well, like oil and natural gas in North Dakota, and there's some mining there. And then there's maybe some in Texas, there's some large data centers, but they're the weird type that shuts off a couple times a day. That's actually a pretty healthy state for Bitcoin mining because it means it's all it's all these little pieces in all the little nooks and crannies of the energy system around the world.

And are you confident with your high knowledge of finance and engineering and Bitcoin? Are you confident that in 5, 10, 20, 30 years there won't be a technical problem with Bitcoin? The likes of which I don't know enough to to kind of imagine, but I've I've tried to sort of approach that of just the the the incentives no longer working. Or you think that Bitcoin is so beautifully and crystally designed that it it's going to be working forever?

Well, the short answer is always risk. I I put in the same bucket as communication protocols. So, what is the chance that Ethernet is still highly relevant 20, 30 years from now? The chance that USB is still very relevant. The chance that simple mail transfer protocol is still very relevant. I think it kind of has achieved that communication like network effect communication protocol status. This one happens to be kind of the communication protocol for value. It's got that leading self-reinforcing security and liquidity network effect. And its simplicity is part of why it's so durable. So it's not this overtuned thing. It's this kind of really simple thing that other layers and stuff can build on top of. Now, what could potentially disrupt it? One is if there's just diminutive demand for a decentralized ledger. Like just structurally. Now it's already, you know, it's a $2 trillion network now, a little under today, a little bit over in recent months. But that just if if we're having this conversation 20 years from now, people like, no, I'd rather use a completely centralized ledger. There's not even in a world of 8 billion people, there's not even 20 million that that want to use that decentralized ledger. That's a failure mode for for Bitcoin in the broader space. The other one I think is at least worth watching is quantum. Now, at times I think that that narrative gets overdone for Bitcoin, but I do think it's it's worth watching. And so, for example, I've I've spoken to some of the people that that spend most of their waking moments trying to make sure that Bitcoin is prepared for the potential emergence of of quantum that is powerful enough to to crack some of the signature types. The short answer is that Bitcoin is upgradable to become quantum hard, but that there are trade-offs. So I mean, basically quantum-resistant signature types use more space. And therefore they they more readily run into some of those scaling things. You might have to potentially increase the block size and do a hard fork potentially in in that environment. So there are I can envision technical hurdles that the that could test the Bitcoin network. And anything that is non-invincible has a has a risk of of being derailed in some way. So yeah, I would put the risk at non-zero. The way I kind of look at it is I say, do I think there's structural demand for it? I think yes. It's basically decentralized portable capital, which I think especially in a world of fiscal dominance and sovereign debt issues that are only going to get worse over the next 10, 20, 30 years. I think that's valuable. And two, is it the best at what it does? And I think its simplicity and size and security and liquidity all point to yes. And then just then the remaining questions are what are the technical hurdles, watching quantum, just watching overall things like that to kind of look for those edge case risks.

So I I I think I understand what you're saying because uh, you know, I've heard this argument made by Martin Skrey, who's uh been a noted short seller of the the quantum, what he would call a bubble. Um, I mean, these companies have very anemic revenue growth. And uh, there's there's a lot of pie in the sky promises from CEOs as well as people who have been long with stock. And people who have been long with stock have made a lot more money than I have in quantum. And probably a lot more than Martin Skrey. But the argument is that for Martin Skrey is that quantum computing for literally anything that's commercial companies wise, it's not going to be 20 or 30 years or maybe even longer before quantum is up to the challenge. But that literally the only problem that is, you know, financially attractive for for that quantum computing could solve is is solving Bitcoin and maybe hacking Bitcoin. I barely even know what that means. So that's what you're referring to that quantum computing probably, if it's real and um scalable and and can do something commercial, one of the first few things it would do is be able to maybe like crack Bitcoin in a way that you know, Bitcoin's main value prop now is uh, it can't be hacked at all.

Yeah, that's one of the most potential profitable use cases for it. The actual applications for quantum broader than that have pretty severe limitations at least in that sort of investable time horizon. Now, just because we had the emergence of quantum wouldn't mean that for example, it can just press a button and like Bitcoin unravels. What it generally means is that certain certain transaction types become more vulnerable than others. So generally speaking, the really early coins like Satoshi's coins, for example, they are the most vulnerable. They're kind of sitting there as like a big quantum honeypot saying, "Hey, you want a several billion dollars?" Then that that's kind of like the canary in the coal mine for if anyone has a quantum computer of scale. Then there's steps people can do to kind of like put put the bulk of their coins in things that are much harder to types of like types of addresses that have not revealed the even the public key yet. For example, the more aggressive thing is if you if you had a quantum computer so quickly that it could could break a signature type in that kind of 10-minute mining window. That's that's the most severe kind of attack. Because the going after Satoshi coins, you have all the time in the world. They're just sitting there. And if you can get a powerful enough quantum computer, you could take them. The next step is can you take coins from anyone trying to transact? In which case your quantum computer would have to be super fast because you'd have to go in there and while it's trying to be transacted, you actually go and kind of snipe it and take it. And if it does start to reach these higher layers of actually disrupting, if there's a provable case of Satoshi or some other kind of coins that are sitting there being taken, that's like a shot across the bow that some of those potential upgrades for Bitcoin probably should be accelerated. Right now, it's still in that theory crafting phase where some people are like, "Hey, quantum could be a giant factor in three years." And then the other other people like Martin are are would say, "No, it's even in decades we it might still not be. I don't know." Right. There's there's this big debate now. So when you have a big debate like that, there's no action is what happens. If there were to be some sort of early kind of provable disruption, that's when it probably radically encouraged the incentives to say, "Okay, we have to make sure the worst-case scenarios don't happen. Some of these upgrades and the trade-offs that they come with are probably heavily worth pursuing at that point."

Do you think Satoshi was working with the team or was just one person?

That's a good question. I I mean, the person that posted on the forums, I think was most likely one person. There's pretty kind of normal wake sleep cycles and the way he used language. I I would assume he probably had one or two people that kind of knew who he was and maybe helped. But short answer is I I don't know. There's been a ton of work into obviously trying to to find out who he was or if it was a team. Generally speaking, the way that the code was released, it looks like a prototype. So generally speaking in software, you'll you'll kind of prove something works, but then before you actually release it, you'll completely redo it. You'll you'll clean it up. You'll make it modular. It was released in that kind of like rushed prototype non-modular way that's more indicative of a either a person or a very small team versus something that's this kind of polished product that you'd expect from a larger team.

So one person or a very small team. Let's this person, whoever Satoshi is, do you think they're still alive?

That's a good question. My my guess is probably, but I I don't know. There's obviously been some high-profile candidates over the years. Some of them have been disproven as being Satoshi. I I really could go either way.

So, you don't think it's healthy?

There. So, there was evidence that he was actively. And it was Jameson Lopp that that did this at a conference. He was a obviously a high contender for a while. I think it's possible he knew who Satoshi was or other he was like an early helper in some sense. He was involved in dialogue. Some of his ideas were kind of a precursor uh for what happened. Like he he did reusable proof-of-work tokens. So he was involved in some capacity. Whether he knew who Satoshi was or Satoshi, the evidence against it in addition to some other things like time zones and sleep wake cycles and stuff is that Jameson Lopp showed because Hal was a runner and some of his data is public. And he was running while Satoshi was posting.

So, at least it doesn't doesn't mean that he can't possibly have been in some way involved or known, but it means he probably was not the person doing the posting like the Satoshi as we as we know him.

Wow. Okay. I heard from someone in the know in crypto, a very smart person, you know, like yourself. Um, and uh, they told me that it was Hal Finney. And that a lot of people in also in the know think it's Hal Finney. Would you say that's fair to say that among among informed people like yourself, the top candidate is Hal Finney? Would you say that's fair?

I would I would say that there's like three or so candidates that for a long time have been in the top running. He he was one of the more credible. He was in that kind of top tier of people who credibly could be. It wouldn't be shocking to find evidence that kind of that that's kind of in the category he was. The evidence that I point to that Jameson Lopp pointed out about the running, that is fairly new. That came out something like two years ago.

So that was kind of like uh and the reason of course it's even worth having that at a conference is to your point, Hal was one of the leading speculative candidates for who it might be. Now I would I would point out that that's that's not been without cost. For example, his widow has been like threatened in many in some cases because they say, well, you have Satoshi's coins or something like that. So there's actually that's there there's a um a responsible way of kind of speculating about these things and other ways kind of potentially a human cost that it that speculation should kind of be done in a conservative and and responsible way. But yeah, he was a leading potential candidate that I think later evidence has kind of put some some back pressure on. There's also details like, you know, what programming language the different candidates were familiar with compared to what Bitcoin was written in. There's all sorts of um factors that go into this kind of speculation.

And the Satoshi wallet has never been touched. So those Bitcoin have is that accurate that they've they've remained in there forever?

Uh yes. Now Satoshi sent very early transactions, including one to Hal Finney, that were kind of test transactions. But apart from those kind of very tiny numbers, they've not moved. Now, it's technically the case that no one can 100% say these coins are Satoshi's because they can't even say who Satoshi is. But they can say that while Satoshi was active, there was this really big miner. Sometimes people call him Satoshi to say that it's probably Satoshi, but you can't prove it beyond a shadow of a doubt. It seemed to mine in a way that was supporting the network. So, whenever he became too big of a share of the mining, he would dial it back. And so he was kind of mining a lot, but not mining any more than he kind of had to to help bootstrap it. So for all intents and purposes, it was Satoshi. And yet those coins have basically been just completely dormant for the entire time. There's speculation maybe Satoshi died. Maybe he burned the private key. Maybe he just quietly still has the private key and had the discipline never to sell this entire time. There's there's a range of possibilities.

Yeah, I mean that that would to me support the view that this person is dead. Because if you had, I don't know the number, 200,000 Bitcoins that went from a value of zero dollars to tens of billions of dollars, maybe hundreds of billions of dollars, that there's tens of billions of dollars there. There's no way no human being would be not be able to sell unless they forgot their password or something, which is something I would do. But you know, even the biggest bulls in the world, Michael Saylor, yourself, you're you're a big bull, not the biggest bull in the world because you're you're very um uh rational. But but you know, the biggest foaming looking at the mouth Bitcoin bulls from zero to $90,000, I think they would would sell. So I I don't know.

I think yeah, the only scenarios where one is is could be someone whose ideology was stronger than their financial incentive. They would kind of know that if they sold they have a chance of pretty heavily damaging the perception of their network. And so they might say that the the mission's bigger than their gain. I think probably the most likely scenario is that they're either they either passed away or or that they're still alive, but they made the one-time decision to get rid of the keys. It's easier in the first five years when Bitcoin is not worth much to make one really hard decision to just get rid of your keys and you no longer have that choice versus having to make that choice every bull cycle. That's obviously, to your point, a much harder decision to do that as Bitcoin reaches $100 and then $1,000 and then $10,000 and $100,000 over 15 plus years to never sell. That's a much harder thing than that one-time decision to get rid of keys.

Wow. Well, I'm glad I'm glad we talked about Bitcoin because I know you're extremely interested in and you're very involved with the community. So, because my stuff is macro, I tend more actually about macro, but I'm I'm glad we we talked about this. Let's let's turn to the other the other hard asset, gold. It's unlike Bitcoin. Gold has is in a very strong bull market. It's over $4,000 and has been so for a month. What do you think's going on here? Why is why is Bitcoin weak and gold is strong?

Well, I've been a surprised bull on gold because I've been structurally bullish since 2018 and long. But if you would have asked me last year, would we see $4,000 gold in 2025? I'd say, well, I'd love to, but it would not be my base case. So, it certainly exceeded to the upside. You don't really expect to move that big in such an established asset. I mean, the short answer is I I do think that we are kind of entering a more multi-polar world where there's just more interest in neutral reserve assets like gold. So there's been some sovereign buying. There's been some kind of institutional realization that nothing stops this train. That these fiscal deficits are going to be with like at scale probably for 5, 10, 15 plus years. That this is like a just a new environment for developed market sovereigns. And that in kind of this increasingly kind of less globalizing world, there's there's just more risk of confiscation of each other's sovereign reserves. And so there's more interest in kind of having things local or having things that can't be debased and can't be just frozen with a stroke of a pen. So, I think there's kind of rational structural demand for gold. And there's kind of a release valve. Like it's just it's been kind of slow and steady for a while and then kind of came out, you know, kind of more than even some of us bulls would have guessed. And then silver and platinum played catch-up. My general view is that over the next 5, 10 years, I mean, I I still think gold has more to run. Like, I think it'll kind of re-enter the the system in some just larger capacity. But that when it's whenever it's overbought to this degree, I get nervous. So instead of instead of trying to trade around it and like selling all my position, I instead just kind of manage expectations and say if gold has a six or 12 month consolidation from here, nobody should be surprised by that because of how quickly it it it ran.

And and to your other point of question, why would bit why would gold run and not Bitcoin? I would say it's largely despite the the stock market reaching new all-time highs, it's largely been a risk-off environment in the sense that as we talked about earlier, everything outside of AI has been fairly weak. And gold is an asset that can do pretty well when liquidity is not bad. And and for this year, liquidity has not been bad, but economic activity outside of AI has been pretty weak. And so it's not shocking that that gold gets a bid in that environment. A whereas Bitcoin is more grouped in with other types of kind of risk-on assets. In addition, especially in a 12-month period, I mean, Bitcoin is a tenth of the size of of gold as like a market size. And now after the bull run, probably less than 10%. And one-time factors can move it. So, for example, when when Trump ran the election, the market is saying, well, okay, now I have to price in some non-zero chance of a sovereign Bitcoin reserve. Now, we've got to price in XYZ. And I in in interviews in early 2025, I I faded the idea of a sovereign Bitcoin reserve. My view was I'd rather estimate price as though it's not a thing and be surprised at the upside than incorporate that into my expectations and then be surprised when it doesn't come. That latter that that's what happened so far. There's been no movement there. So any sort of like these kind of like euphoric estimates that kind of people bought Bitcoin for a year ago, that has to kind of work itself out of the system as kind of the correct reasons to buy Bitcoin reassert themselves, right? There's there's kind of even a good asset can be bought for the wrong reasons or with wrong expectations. And so coins have to get to holders that have the what I would what I would argue more accurate expectations and away from those that bought for the wrong reasons for the wrong kind of timelines and things like that. So I think Bitcoin's been going through this kind of like normalization phase where we kind of get away from the election bump that it enjoyed and back toward what is it structurally doing or not doing.

Right. So Lynn, you said that Bitcoin mining companies have a way more opex, less capex than than data centers. But I do think that in the maybe the the the crypto bull market 2020, 2021, there was a lot of speculative activity in Bitcoin mining. Like and there were loans that were made collateralized by Nvidia chips. This is something that we are seeing now. And there's a debate in in the the stock market community about is the appropriate weighted life for depreciation appropriate. So Microsoft and and Google and Amazon Meta are depreciating these chips at 5 years, 7 years. What if the real life should be two years? So actually the depreciation expense should be way way higher, which is going to suck out of earnings. Do you think that let's say the Jim Chanos argument, not talking about MicroStrategy, but talking about the the depreciation argument, do you think that that argument is correct and that the artificially long weighted average lives of depreciation for these chips and data centers is artificially boosting earnings? And if it if so, could could end disastrously.

So my intuition leads toward that yes, that those cycles are probably shorter than they're currently being accounted for. I would generally leave that to the experts that that have focused more time on that. I will say that in Bitcoin mining where where I have a more of an example, there was a well-known critic of Bitcoin that put forth estimates for how short the supposed kind of capex cycle is for those like how long the lifetime was. I think he estimated like like less than 18 months or something. And it was like demonstrably false in the Bitcoin mining space. That if anything, a lot of machines go five, seven years because of that thing I mentioned before where first they're bought by these kind of big miners and then after after two or three years when they're no longer the best models, they they go out on the secondary market to those that are like getting natural gas mining with them and stuff like that, these kind of scrappier ones. And that the full life cycle ended up being longer than people thought. But that's a that's inherently kind of scrappier type of chip than GPUs. So from what from what I understand from the industry, they do burn out pretty quickly because there's so much demand. Now, I don't know what happens when we reach more of a steady state. Obviously, things are way different when you're in explosive growth phase versus when you kind of rationalize and get to a more steady state. So maybe maybe they're shorter now than accounting expects, but then when the dust settles, they lengthen again perhaps. I I don't know. But I would leave that to to them. I would say that in the Bitcoin's case, the the capex the cycles were actually longer than some of the the bears said.

That's interesting. Again, I I don't know what I'm talking about. But you said that Bitcoin can kind of go off-grid whereas these data center AI chips are running all the time. That's why they have to have liquid cooling because it just gets so so hot in there. So Lynn, I'm not going to ask you the question and try and corner you and say, "When is this AI bubble going to pop?" Because that would be about share prices and that's so hard to predict. I mean, it's an interesting question. If you want to give me an answer, I'd be delighted to hear it. But maybe something a little bit more tangible. When do you think the turnover in AI capital expenditure is going to be? And I I'll share what I mean. Like it the capital expenditure on telecom was higher in '98 than it was in 1997. Higher

In 1999 than it was in 1998, higher in 2000 than in 1999. But 2000 was the peak and then it from it went down from 2000 to 2001, then from 2001 to 2002. So that was the decline and that coincided with the popping of the dot bubble.

Do do you have a sense of how long this explosive growth is going to be? When might this thing roll over? Is it going to be in six years? Is it going to be in in six months? And how are you what are your how are you thinking about that question? What are what are your inputs for that question?

I think that the share price and the capex are are actually somewhat intertwined so that they're not just two different things because and an example of that is when when Facebook kind of went quote unquote all in on the metaverse even changed their name to Meta was doing a lot of capex on VR. I mean their share price took a really big hit and and they kind of had to to pivot. Basically the market saying we don't love this. I think that as long as the market loves it, these really big AI companies will most likely keep pretty aggressively pushing it. But I do think that the the price probably rolls over first and then capex. It's it's quite pro it's quite possible that we're in the rolled over phase for price now. I again like you said that's super hard to predict and so I I'm not going to try to make that call but I do think that in the next 12 months or so we we could see the rollover in kind of the for the price and then capex maybe follows by a year or two that's if if I had to give a base case it might be something like that but that but that wouldn't necessarily be like quote unquote the forever like rollover that could be kind of like how back when Ethereum was using GPUs used to mine. So in in say let's say 2017, Nvidia had this really big boom and then it cooled off, but that's now like a micro blip on the Nvidia chart. Actually, that I I end up trading that one at the time. I I kind of bought into that dip and it ended up selling too early, but it was like a useful trade. I I think it it we probably see one of those in that time frame that I mentioned. I I don't have super high confidence on the details, but I it's not doesn't mean that it's it's like over forever. It just means that basically probably go through a cooling cycle to see okay where is this actually working where is this actually have a chance of being profitable where is this malinvestment recalibrate from there and then probably have another cycle for all I know. So I I view it more as in terms of kind of an intermediate kind of rollover rather than trying to call like a gigattop some sort of generational change in the whole space because like I said I think that it's real and I think that the amount of capex over the next 5 10 15 years is going to be immense. I just think it won't be a straight line. I think it'll have some sine wave to it.

That's really interesting, Lin that you think that your base case could be that like this is the top and that the share prices are decline. Obviously, not the perma top, but like the the top for this cycle. That's really interesting to me. I feel like if you have that view, isn't it almost a necessary view, a following view that you are bearish on the S&P over the next year? Because as we agreed upon in the beginning, this stock market cycle is the AI cycle and this econom economic cycle is the AI cycle.

>> The way I would phrase it is I it's not that I expected a giant rollover, but I think we could see a consolidation. So I I think it could have a a flat year. I think it could be up down 10%. That's that's still pretty weak for the kind of the recent trend in the S&P. And even when you enter a kind of a gradual rollover or a consolidation, uh that can lead investors and companies to scrutinize their spending more. So it doesn't take necessary a crash like in Meta's case or kind of a partial crash. Just a period of kind of more weakness. I mean, we've already seen credit default swaps of some of these AI related bonds. That's the that's an early signal of the market saying, are you are you guys sure about this? And there's and that's even without a a major draw down in in in kind of the broad space. some individual names. So I think you can see individual draw downs overall kind of choppiness and just kind of less stellar returns kind of like what we saw in Bitcoin this year, right? So it's it's down but it's really kind of just a a volatile flat for the year. November 2024 to November 25 you touch 100K and then you're still like now below 100K is really choppy environment. You could easily see something like that with AI names where it's not it's not the big crash that the all the bears want, but that it's just the the the rate of change and the euphoria is behind itself and then all those decisions get more scrutinized. Take a breather, which again is different than being like over and we already saw like recently if you look at like the AMD chart, the Nvidia chart, they had these really big kind of pullbacks and then new highs. So like and I I mean I was I got into AMD in my research service when it had that really big pullback got out which probably proves to be too early after like roughly a double and so the like I these things happen they kind of come and go in waves now so far the waves have not been big enough that aggregate has rolled over and the question is can you get enough roll over once but e even as kind of that skeptical kind of near-term I I mean I would be a potentially a buyer of of of pullbacks on at least certain parts of the market should we get it. So, I'm not like a structural bear on this. I just think that it's it's kind of near-term pretty heated.

>> Interesting. But what but if that the decline I I don't know if you gave numbers on it, but the decline in these uh AI stocks, let's call them, is is going to cause a decline in capex. I feel like then that is a reason to be very bearish. like I I don't think if capex next year is lower than this year, I don't think the uh these AI stocks are going to do well at all. I'm I'm particularly thinking of Nvidia, Broadcom and stuff and uh you know, those are those are a lot of the S&P.

>> I agree with you, but that that'd still be tighter than what I said because I think that it could potentially roll over price-wise like in in this year like in the next 6 12 months, but then with a two-year delay potentially you have capex also then start to struggle. So, I I'd be surprised if capex struggles in the next 12 months. Even though I I do think that I think you said six years or six months, I think that it's probably somewhere in the middle. I don't I think we'll see a cooling probably well before six six years is my expectation. I I do think that maybe that two to threeyear time frame is is reasonable, but again, a lot of this is new. So, I mean, it's it's like a opinion loosely held. that's what I'm kind of willing to say now, but I am regularly monitoring this type of thing. And so I'd be kind of looking for signs that I'm right or wrong and adjusting rather than just saying here's my opinion. I'm never going to change my opinion and then see what happens.

>> I completely agree. Yeah, you can't be fixed in in your your view. So Lynn, do you think that AI is in a bubble? Do you think the stock market is in a bubble? If not in whole, in part?

>> I think it's in part. I think it's in part a bubble but based on a real trend that is that is worth trillions and that will notably impact things over 5 10 plus years but that much like Nvidia was in a kind of a local bubble in 2017 cooled off got into another bubble cooled off got into another bubble I I think that kind of the industry as a whole is is a little bit bubbishious now especially with some of these like recursive deals between each other kind of a little bit incestuous financing going on signs not of a bottom That's that's for sure. Bubbles tend to go on longer than people think. So just because you're in a bubble doesn't mean you're at the top of the bubble. But yeah, I do think that it's a it's a crowded space that it will at some point need a shakeout to say, okay, what what percentage of this is real? What percentage of this is just kind of along for the ride and need and the fat needs to be trimmed off. Any growth industry needs to go through cycles like that. Otherwise, it just goes to the moon? Like it's just basically it needs kind of that speculative froth where someone's willing to finance. anytime someone puts AI into a name if if it gets financed that is not sustainable that needs to get shaken off and then say okay well that wasn't sustainable but this pocket is and so kind of capital gets a moment to recalibrate and go toward what's actually working and so I do think that I would be surprised not to see some sort of consolidation or shakeout in that kind of two to threeyear period cool off to then recalibrate and see what's okay what's still growing what's still structural what's real

>> yes I mean I just it seems to me like that the CEOs of these giant trillion dollar companies, they are so dedicated to this project that they're just going to spend as much money as they can as quickly as possible. And uh you know, their stock price is going to have to go down a lot more than 20% for them to stop doing that. And you know, no insult to to uh anyone, but it's kind it's kind of like you know, Michael Sailor is such a believer in in Bitcoin. He is going to be issuing stock and financing stuff, raising money to buy as much Bitcoin as possible. like hell hell or high water. And I think it the same is true of these tech CEOs. They're going to be giving money to Nvidia and buying chips and buying data centers, you know, and they don't care about the the post on Twitter showing Oracle CDS credits full swap. Like they they don't care. Uh that's my view.

>> I I do think it would take quite a bit of pain to slow them down. And I I also think that again going from growing to flat is a big deal or going from growing sharply to then growing mildly and all the analysts kind of reducing their estimates is a big deal. And I think that's a more realistic deal than oh like basically we have a giant capex boom and then it turns out the next year is like nominally lower by a substantial amount. I think that kind of sharp reversal to your point is very unlikely because I think that so many CEOs view this as structural existential. They're not going to say, "Oh, our stock price down 20%." Or, "Our our our bond market's getting jittery. We're going to complete change." They might come out with a statement. We we want to clarify we're pursuing profitable growth. We are being responsible and maybe they say, "Okay, we'll dial back our expectations to some extent." That's a more realistic scenario. The rate of change rolling over and and things consolidating more so than just a sharp reversal. At least

>> that makes a lot of sense. And again, people are pointing to the Oracle credit default swap. Like that's true, but also like Pimco did a deal with Blue Owl and Facebook Meta and they made $2 billion in a day by by funding this bond. So there's so much money in the fixed income world that is fee based as well as uh carry, you know, profit, you know, incentive based that um even if they are bad deals, I don't know. I don't have the like the it's happening. It's h that's my view.

>> Yeah, I think for a while. Yeah, I agree.

>> Yeah. Well, Lynn, thanks so much for joining us on Monetary Matters. As everyone knows, people can find you on Twitter at Lyn Alden Contact. Your website, Lyn Alden Investment Strategy is fantastic. That's lynalden.com. So, people know about your Twitter, your research service. Tell us about Ego Death Capital and what you're doing there.

>> Sure. So I'm a general partner at Egoath Capital and we are a series A and seed investor in Bitcoin related startups. So v the venture space building out some of the rails that we discussed earlier. We also some of some of our companies touch stable coins, things like that. Not really the broader crypto space, but anything kind of related to Bitcoin, moving money around, kind of focusing on high utility, low speculation elements. And that's kind of my appetite for for the unprofitable investments, these these long-term growth things. In public markets, I tend to focus on profitable companies. And in that private work is when I kind of focus on these longer term hyperrowth types of bets.

>> Well, that sounds great because if there's one thing that crypto needs, it is high utility, low speculation. So, congrats on that. Thank you everyone for watching. Please subscribe to the Monetary Matters YouTube channel and leave a rating and review for Apple Podcast and Spotify. I know the show gets a lot of listeners, but there aren't that many reviews. I think that's something to do with the maybe the viewership, the demographic. But please, please leave rating review. It really helps helps the show. Thanks for tuning in. Looking into REMX by Van? Head to van.com/remxjack to learn more. That's van.com/remxjack. Until next time. Thank you. Just close the door.