What We’re Reading (Week Ending 27 April 2025) - 27 Apr 2025
Reading helps us learn about the world and it is a really important aspect of investing. The late Charlie Munger even went so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 27 April 2025):
1. Rebalancing the world economy: Right idea but wrong approach – Richard Samans
The Trump “reciprocal” trade initiative identifies a long-standing and legitimate concern of U.S. trade policy, but the administration has mistakenly conflated differences in tariffs and other trade practices with the larger, more macroeconomically driven phenomenon of the level and persistence of the country’s trade deficit. Unfair trade practices are a contributing factor, and the large general tariff increases will lead to some degree of import substitution. But undue reliance on this blunt instrument will lead at least as much to a diversion of imports from one trading partner to another and compression of demand due to the higher prices brought by import substitution and lower income resulting from lost exports due to foreign retaliation…
…The administration is effectively treating the superficial symptoms (bilateral trade deficits) of the underlying problem (macroeconomic imbalances) and doing so with an overprescription of an outmoded medicine (a 19th century-like tariff wall) that runs the risk of precipitating a cascading failure of the patient’s (the U.S. and world economy’s) vital functions…
…The aim of rebalancing the world economy and modernizing its cooperative architecture remains a valid one, and the incoherence and incompleteness of U.S. policy in the face of it across multiple U.S. administrations of both parties runs even deeper than the considerations outlined above. The world economy has changed—geopolitically, technologically, environmentally, and most of all in terms of the distribution of industrial production and middle-class purchasing power—such that the time has come to think more seriously beyond existing international economic arrangements that were formed in response to 20th-century circumstances and challenges…
…The time is ripe for a new “deal” to be struck among the major economic powers aimed at strengthening the growth and stability of the world economy, an outcome that would be greatly in the national interest of each. The U.S. is eager and brandishing a big stick in the form of its new “reciprocal” tariff initiative; Europe is already preparing to do its part, albeit for unrelated reasons; and China appears to have finally convinced itself that, after wave upon wave of decreasingly effective supply-side investment-cum-export stimulus measures, it has no choice but to boost domestic consumption in order to maintain sufficient growth in output and employment.
However, to be fully effective the bargain must be a truly grand one, firing on all of the cylinders of international economic cooperation. It will need to include major initiatives in fiscal, monetary, development, and trade policy in order to yield demonstrable net benefits for each of the main protagonists as well as the international community as a whole…
…With respect to fiscal policy, China would need to agree in concrete terms to implement reforms sufficient to raise domestic consumption as a proportion of GDP to a level commensurate with its share of responsibility for maintaining the global economy’s momentum and stability, for example, by 10 percentage points of GDP over the next decade. Its final consumption expenditure relative to GDP is about 56%, which is well below that of other industrializing economies (e.g., 71% in India, 72% in Malaysia, and 82% in Brazil) and the global average (76%)…
…At the same time, Germany and the rest of Europe would need to commit not to offset the fiscal impact of their reflationary defense-related spending increases (an estimated additional 1% of GDP over the next few years); and the U.S. would need to agree on a target for the orderly reduction of its fiscal deficit to a more sustainable and cyclically appropriate level, for example from the current 6%+ to about 2.5% of GDP over the next four years, not far below the 3% of GDP target Secretary of the Treasury Bessent has advocated…
…With respect to monetary policy, all three partners should signal a contingent willingness to undertake coordinated intervention in foreign exchange markets to support a limited and orderly (e.g., 10% to 15%) depreciation of the dollar to levels more consistent with the progressive and symmetrical adjustment of global economic imbalances, reinforcing the expected effects of the fiscal policy measures outlined above, if required. In addition, they should request the IMF to calculate and publish independently (i.e., not subject to prior Board approval) estimates of exchange rate reference ranges on a semi-annual basis that it deems consistent with this immediate objective and the larger, ongoing one of avoiding large and persistent global economic imbalances…
…Regarding international trade policy, the three preceding components of this global accord would make possible the wider “balance of concessions”—i.e., positive-sum game political outcome—that was missing in the Doha Round and continues to frustrate attempts to update the WTO’s norms and dispute settlement system. The institution has been in a state of semi-suspended animation since the first Trump administration, and the Biden administration did little to breathe new life into it. The carrot of a large and sustained increase in development financing, as outlined above, and the stick of potential unilateral adjustments in U.S. tariffs could help to create the conditions necessary for a constructive, negotiated reset of the institution and multilateral trade system.
By now, it should be abundantly clear to the world that there is a durable bipartisan consensus in the U.S. that the country no longer enjoys the enormous advantage in economic size and technological leadership which led it to take a generally non-reciprocal, foreign-policy-first, and short-term-shareholder-return-first approach to trade policy in the latter half of the 20th century. In addition, market-based growth and development, or capitalism, has evolved into many shades of “mixed economy.” Thus, a one-time negotiated rebalancing of tariff schedules combined with modernization of rules and procedures to take better account of the changing nature of trade and industrial policy as well as corresponding reform of the dispute settlement system, paired with a big sustained push on development and climate finance and refocusing of trade preferences on low-income countries, might provide the political basis for a new modus operandi for the WTO, especially if it takes place in the context of a process of macroeconomic rebalancing among the largest players. To enhance the prospects for success of all of this, the U.S. should express a willingness as part of the overall accord to discuss its tariff rebalancing objectives vis-à-vis the countries with which it has the most legitimate concerns (based on their actual practices rather than bilateral balances) on a best-efforts and time-limited basis, employing the procedures authorized by the organization’s charter for this purpose.
In sum, such a four-part international economic accord would be far more likely to spur major and enduring adjustment of global economic imbalances than the blunt and risky instrument the Trump administration has deployed…
…The U.S. would arguably be the biggest beneficiary of a 21st-century project to update multilateral norms and institutions that were established in its image during the last century. Judging from recent pronouncements, there appears to be little prospect of it acting along these lines at present. For the time being, the responsibility falls to Europe, which has greater agency to lead the international community in this direction than it may imagine.
Particularly if financial market volatility and fears of a recession in the real economy persist as a result of the trade policy shock, the current administration may begin looking for a way out that would still enable it to claim partial credit for having revived global and U.S. economic growth prospects as well as modernized international economic relations and institutions. The proposed G3 accord offers such an opportunity.
2. Johnson & Johnson Pivots Its AI Strategy – Isabelle Bousquette
The “thousand flowers” approach involved a number of use case ideas germinating from across the company, which made their way through a centralized governance board. At one point, employees were pursuing nearly 900 individual use cases, many that were redundant or simply didn’t work, he said. And as the company tracked the broad value of AI, including generative AI, data science and intelligent automation, it found that only 10% to 15% of use cases were driving about 80% of the value, he added.
Now J&J is drilling down into high-value generative AI use cases around drug discovery and supply chains, as well as an internal chatbot to answer questions on company policy…
…J&J began its pivot last year, removing a centralized governance board responsible for vetting employee GenAI ideas. It then distributed governance responsibilities to various corporate functions, including commercial, supply chain and research, that had a better handle on whether the use cases were actually driving value in their area. Those groups were able to shut down or consolidate redundant use cases and focus resources into the ones that were working.
One example that is working is a “Rep Copilot,” which helps coach sales representatives on how to engage with healthcare professionals about new treatments. The company is piloting this in its Innovative Medicine business segment, which develops new treatments for oncology and other areas, and is now working to expand that pilot to its MedTech segment, which sells robotics and hardware like hip replacements and lenses.
GenAI also is being used for an internal chatbot that ingests information about company policies and benefits to help reduce the some 10 million interactions employees have every year with the services team.
In drug discovery, the company is looking at whether GenAI can help researchers find the optimal moment to add a solvent to turn a liquid molecule into a solid. Swanson said J&J also is testing how AI can help identify and mitigate supply-chain risks, including the impact of a shortage of a given raw material.
3. An Interview with Dan Kim and Hassan Khan About CHIPS – Ben Thompson, Dan Kim, and Hassan Khan
To your point, South Korea was very attuned to the importance of semiconductors, obviously.
DK: Oh, no doubt. It was decades in the making, and a lot of efforts, and actually a lot of what South Korea did has been a bit misunderstood, and if you would ask my former CEO — I also used to work at SK Hynix, I’m jumping around a little bit here, as their chief economist — and I would ask their CEO very bluntly, I said, “What does the government do that is the most important for South Korean companies in the semiconductor space?”, and no hesitations, “That we invest in the workforce, workforce is the most important thing, and without the workforce, nothing else matters”, and so I learned a lot there.
Is the one sentence summary, government invests in the workforce, unlocks these large companies, and then I think the thumbnail version of South Korea’s rise to dominance is really investing from downturns driving the Japanese out of the memory market and that’s sort of been the foundation. Is that a good synopsis of how they got to where they are?
DK: The way that I think it made sense to me is there are a couple things. One is on the workforce aspect of it, if you wanted to study electrical engineering or material sciences that were relevant to semiconductors as a PhD student or a Master’s student, there was a lot of availability of dollars for you to get into that and then you could get into the companies thereafter.
The other, of course, is yeah, they went into the memory sector first and they were part of this consolidation trend at the time, and so it really was, “How do you innovate and achieve scale and not die along the way?”. I think there was a couple of interventions by the government in terms of guaranteeing loans and things like that, but by and large, it was just scratch-and-claw to survive, and I think if you have ever lived in Asia, if you’ve ever lived in Korea, scratching-and-clawing to survive, it’s actually not a bad way to describe their economic efforts for the past 50 years or so. So I think it sort of played right in their wheel house to be in the memory space…
…DK: So one is Capacity. Can we allocate resources to build significant production scale to make a difference in this country to actually close the cost gap between a North America and East Asia?
And what is the fundamental driver of that cost gap?
DK: Yeah, that’s a good question, I would break it down into two aspects. One is that what we found is that the manufacturing ecosystem in this country, particularly when it comes to semiconductors, had atrophied quite a bit over the past three decades or so, and so everything was a little bit inefficient and that contributed to higher operating costs, that was partially due too because of labor cost differences between East Asia and here. We would have anecdotes of experiences where there was a lack of specialized plumbers, immediate availability. When you need to fix something in Taiwan or South Korea, you can call them at 3AM and they would come immediately and fix that thing so that your fabs could operate.
And literally every minute that a line’s not running is costing you a tremendous amount of money.
DK: Exactly right. And in the US, that’s not necessarily the case. Plus, the US is a very big place, so you have to create an ecosystem within a limited geography that has all your suppliers at an efficient scale, has all that labor and talent at an efficient scale. There are some advantages in the US, especially that the energy costs tend to be lower than it is in East Asia, but the labor costs are higher, everything is a bit slower. The equipment themselves, which if you’re building a new fab just to get it up and running, somewhere between 60 and 70% of a new fab, the equipment themselves will be the same. But the cost of installing it actually might be higher because of the labor costs and the service contract you have to enter into on top of that.
HK: And the timeline to get those up and running is a major cost driver. So even if you’re buying the same tools, the fact that our fabs, everything that Dan said about the ecosystem being weaker, that compounds during the construction phase, and every additional day in the construction phase blows up your CapEx budget. And you saw that I think most acutely in some of the first fabs that were coming online, they were very publicly known to take much longer than the rest of the world.
One of the data points that’s maybe less obvious, I think, to folks on how the ecosystem is maturing in a lot of the ways that Dan pointed out, is you don’t hear as much noise now as firms like TSMC is on its second fab, Texas Instruments is moving its Sherman complex forward, because they’ve trained up the workers and they’re building replicas of the last ones, they’re moving faster.
How much is it training the workers versus importing the workers?
HK: Well, I think for construction it’s a lot of training because you want that workforce. But for a lot of the talent to run the fab, there’s a portion of, you need to get that talent in and Dan can probably speak to that more.
DK: Yeah, this is a good question. Especially when it comes to foreign companies building in the United States, yes, there’s a lot of companies that are working at a company like TSMC or Samsung or SK Hynix, but there’s a select group of people within that company that is actually doing process engineering, integration engineering, that makes up for the very special and very secret sauce that differentiates that company versus everywhere else. That’s not something that you can replicate immediately in a different place, especially in a foreign country, and so to some degree you have to be able to import specialized knowledge and that could only be done with people that are coming here and that they need then to be able to train folks…
…What we mentioned earlier, that is the industry that just in the very structure of the industry is a total priority on efficiency because so much is out of your control.
DK: Correct. So there is a business model innovation that we missed out on, and then there’s the efficiency, brute force efficiency and scale game that we purposely let go, in a sense. When you miss those two things, that’s two thirds, three fourths of the total capacity there is in the world if you add up memory and foundry. By the way, when you mature a foundry fab, then you have all the current and mature technologies that serves hundreds of customers, not just the leading edge, so we missed out on a lot of that. When we had to diagnose what went and wrong here, the easy answer that an industry association gave was, “Oh, it’s because foreign governments subsidize their production and that’s why it went elsewhere”, and I think that told a very incomplete story as to what happened.
No one in the US wanted to do a foundry. That was the whole reason why TSMC had a market opening.
DK: Correct, yeah. And so because of that, we needed to think about what business models should we support and what kind of capacity do we need to build? And so that’s the first C, Capability.
No, that’s good. That was a very, I think, fruitful discussion. And the feeling with the CHIPS program, was foundry capability the top priority?
DK: It was among the top priorities, and if you read our notice of funding carefully, there is actually a very explicit statement in there that says that there is a priority for business models that can serve multiple customers. What else could it be other than a foundry model that can serve multiple customers?
Of course being able to serve multiple customers not only gets you scale, but to your point about the difference between IDM and foundry is that the inherent advantage of a foundry is that all your customers are now invested in the success of your process and so they have engineers there that’ll fly over to your fabs, that’ll meet with your teams and iterate and so if you’re an IDM and you’re no longer competing against a foundry, you’re competing against that foundry and all of their customers that are invested in that process.
Yeah, it’s a shelling point for everyone to invest in the next process, which is getting astronomically more expensive.
DK: Correct. Because of that, it lowers the risk on everyone too. So from the foundry’s perspective, you have customers who are buying corridors ahead of time, but also who are invested in the success of that, so to break into that actually is very challenging.
Hassan, you wanted to follow onto that?
HK: I was going to just add, when Dan talked about the preference for the foundry model, remember too, we were writing this right after the COVID crisis where the emphasis on supply chain resilience came back and said, “We really want you to be able to service a range of customers and make that capacity available broadly”.
And actually another comment that I realized after Dan brought this up, I think the point that he made about ecosystems and your customers investing in your processes is the biggest overlooked advantage that TSMC had over Intel. Intel was fighting at one point — it continues to fight — an ecosystem, and that’s another value add of why you want to have the entire ecosystem invested in here. If you just built fabs for the four or five IDMs that continue to operate here, you would not get the global ecosystem to be investing alongside them to make them a success.
This is one of the biggest questions, you gave me a segue to Capability, the second C. I’m going to come back to that in a moment. But this to me is the biggest issue and it’s not a critique of you, it’s a zoom out. I’m also not sure how to fix this, which the key here is the demand side, and you guys are on the supply side. The problem for Intel is a lack of buyers, and some of that is on Intel, some of it is just the historical development of the reality that you’re not fighting TSMC, you’re fighting TSMC plus Apple, plus Nvidia plus Qualcomm plus everyone else in this ecosystem. Is that just an intractable reality that you couldn’t address because that wasn’t your remit? How did you think about this balance between supply and demand? And actually spurring, “We’re not just going to give money one time, but we’re going to have a virtuous cycle that helps us achieve our goals”?
HK: I think there is a real challenge for those. I think those firms, all of them, if you gave them truth serums would say, “We love TSMC and we are afraid of our full reliance on them”. Not just from a geopolitical perspective, but from a pure business perspective, you don’t want your entire business hinging on one supplier, right? The same way TSMC feels that discomfort regarding ASML and the reliance on EUV for all these advanced nodes.
Well, I think that’s underappreciated, about the whole thing with the semiconductor equipment manufacturing is it used to be all these equipment manufacturers did one thing and they were different parts of the chain, and TSMC basically said, “No, you all have to learn how to do everything,” so they can set you off against each other. But it’s happened because they’ve had the most power, but everyone depending on TSMC feels the same way, but has had less leverage.
HK: But go back to the fabless firms, and I know Dan had thoughts on this too, but at the end of the day, they have to be able to make a bet, it’s a minimum $500 million bet and they have to be able to go look their shareholders in the eye and say, “I’m going to spend $500 million taping out a chip with a foundry that’s not TSMC”, and they’re going to ask them why.
Well, I mean Apple tried. They had a generation where customers were looking, “Is this a Samsung chip or a TSMC chip?”, an unacceptable outcome to them. Nvidia’s tried more than anyone, they’ve tried to balance, and that ended up costing them because they weren’t a favorite customer of TSMC because TSMC’s like, “Oh, you want to flirt with Samsung? Not so good”.
HK: And Dan will tell you, we had these conversations, and I think the tension is here. Our remit was to create, exactly as we talked about, the capacity needs to be economically sustainable and viable so we can’t then go beat people over the head and say, “Hey, you need to go invest in a foundry that you can’t look your shareholders in the eye that you say that is not there yet”, which they were public. But I do think it put us in a tough spot of saying, we all looked around in the room and said, “These efforts aren’t going to be successful if other firms don’t buy in”, and we were trying to create the conditions to reduce their risk. But at the end of the day, that’s on the firms to deliver that technology and for them to get comfortable with each other.
Is this the bit where, you guys come in after the law is passed, and if you could wave a magic wand and actually reshape the law, and maybe this gets into the broader industrial policies perspective, my critique is that the law only addressed the supply side. There wasn’t a generation of demand, whatever that’ll be, I feel like spending billions of dollars to buy chips that you throw in a landfill would actually be tremendously beneficial because it’s a guaranteed buyer to get this stuff on the lines. Was that just a real hole in this whole program?
DK: That’s a good question. You’re right that we focus on supply because that was our responsibility in our remit.
That was the law, right. But if you could go back and there was a different law?
DK: That was the law, but we grappled with demand quite a bit, and that was one of the factors that we absolutely required companies to prove to us that there was a business case to be made. Meaning who are your customers? Are they bought into this? Is the government essentially investing in a space in which you have no hope of a customer, in which case the funds wouldn’t be given?
One of the many things that our amazing investments’ office, these finance and other professionals were doing, was doing a lot of due diligence into reaching out to customers to the extent that we had permission to do that and saying, “Okay, what is your plan to use this company and this fab?”, and you can think of it in terms of calling an Apple or the Nvidia’s of the world, and of course all of that had to happen and that’s not a secret.
But take a small MEMS producer, of course we’ve had those in our portfolio as well, we’ve had to contact their customers to say, “What’s your long-term plan with this company?”, and so we had to do that. In some ways then yes, we were only focused on supply, but what we found is that just by calling the customers from the US government’s perspective saying, “We are willing to give money to this supplier if you are willing to take on them as a customer”, did so much to put that potential customer at ease and buy into that corridor.
So what I’m trying to say is that this is an untapped power that we had if we were given more authority to do that.
To give them security that they should take the risk of getting the supplier.
DK: Right. But we have to be careful here too because at some point how much a government should be dictating what demanders should be doing, is a really tough question…
…DK: I could talk for another two hours about this particular story, but let me boil it down to a very quick one. Believe it or not, my daughter’s life was saved by a semiconductor technology at birth, while we were negotiating with TSMC, Intel and Samsung. She was born in December 2023, and so we were really in the thick of it, and the only way that her life was saved is that there was a new pacemaker that came on the market, on an emergency authorization, that was miniaturized enough to save premature infant babies with heart troubles, and I knew that if there’s anything to do with miniaturization of any electronic device, it has to do with the semiconductor technology, there’s nothing else that’s going to drive it.
So I contacted the manufacturer of this electronic device and I said, “I want to know everything about that semiconductor technology,” which is not something that parents usually ask, but because I was in CHIPS, I wanted to know, how did you do it? Who makes it for you? What are your options? Can you make it in the US? Did you have a shortage of this during COVID? How can we fix this? Because it really distilled for me in a really tangible way, in a personal way, what we were trying to accomplish. Not just, “Can we make iPhones?”, and “Can we make server chips?”, but, “Can we make a life-saving device that’s using the latest technology that can get there?”.
They explained to me that they had searched everywhere for a foundry partner, because they themselves could no longer go down the innovation cycle of chips themselves, they have their own fab, but it’s outdated, so they needed a foundry partner. No one would take them on except for TSMC. Why? Because they knew how to do the 3D packaging. They had fully depreciated six inch, eight inch fabs, and they were so maniacally focused on serving their customers that they didn’t mind taking on 1,000 devices, not wafers, 1,000 devices. Morris Chang himself apparently said, “We need to do this for this company to save these kids”, when nobody else would take it on.
That’s the kind of foundry we’re talking about here. And so we could talk about abstractions, about whether the foundry model is superior and IDMs, and there are clear superiorities and differences in business models, but I think we are talking about a very unique company and the culture that they have that have allowed the world to be served by it and served so well, but now we are exposed to the risks of it.
Now we come back to the question that you’re asking, which is, “How do we now de-risk that, not only through supply chain, supply-based policies, but is there demand-based policies that we could get to?”. My hunch is that there absolutely are demand-based policies to get at that, and there is a hunger for it from the customers’ perspective. To Hassan’s point, we have heard so much glowing feedback from TSMC’s customers about how good they are at delivering, they under-promise and over-deliver. It’s the theme that we hear over and over, and over again from the customers and they’re saying, “But if they can build in the US or if there could be US alternatives, of course we will take a look at it because we are not stupid, we know the risks here”.
So if you look at the enthusiasm for the Arizona fabs at TSMC, I think that tells you what you need to know about that company and what company is willing to do, but it’s not a complete de-risking. As an economist, I would have to say, if you’re looking for an insurance policy that completely de-risks, then that’s a very expensive insurance policy, almost too expensive for the world to handle…
…HK: But things like Datacom transceivers that are being made on indium phosphide, if you want to build mega AI data center clusters that can communicate across campuses, you need to be able to push the frontier on indium phosphide technology. That’s a couple $100 million investment to move up to a six-inch wafer. That doesn’t grab headlines, but it’s one of those key linchpin technologies for being able to build something that really does matter to the government if you want to be able to deploy the best AI models…
…Well, to me, this has been one of my biggest critiques of CHIPS, in that I felt it was too focused on the leading edge, when the single biggest contrast between national security or resiliency concerns and economic incentives was the trailing edge. The reason we have trailing edge is it used to be leading edge a long time ago, so it was economical because these are fully depreciated assets. It’s impossible to rebuild that economically because you have to pay off your equipment costs, and you’re competing with fabs that don’t, and China can do that because they have to move their way up the learning curve, and so they’re going to dominate all this trailing edge. TSMC has that because they already built it, even they in the response to China are specializing all their trailing edge fabs too, because just general purpose, 28nm chips or 90nm chips or however back you want to go, China’s just going to inevitably flood the market there. There is no one that can solve this other than the government, the economics never pencil out otherwise. How much did you think about that and how did you balance this? You mentioned Hassan, ChatGPT takes the world, everyone’s thinking about AI, but actually where the government can arguably we have the biggest impact is in this area.
HK: I think the most debates that we had on portfolio construction were essentially around the question of how much to reserve for the trailing edge, because the big guys came in early and we knew what their asks were and their asks, and the Secretary said this, if you looked at just the Big Four, they totaled more than the $39 billion that we had. So there’s a version of it where we could have just said, “Hey, you all get your ask or something close to your ask and we’re all going home”.
We chose not to do that, but we then did have a long set of conversations on how much do you reserve. The problem, Ben, was the firms in the projects they were proposing, their assumption was there’s no world in which the economics of a new build here are going to work, unless the government is willing to lean in at such a level, almost the majority comes from the CHIPS program on a $10 billion plus fab. But there was also, we had given guidance that our expected funding ranges were in the 5% to 15% plus the 25% ITC [Investment Tax Credit]. So, a lot of these firms read the tea leaves.
So, this was sort embedded in the law. The law itself didn’t really allow for it, because a trailing edge project needs 100% government funding. That’s just the reality of it.
HK: And I think it was sort of anticipated that there was not an appetite for the government to lean in and basically say, “We are building this fab”, unless the DOD wants to do it on special circumstances, which is a separate case than what the Commerce, how people viewed our authority in the Commerce Department. It was to go fund facilities that would be economically viable with dollars on the margin, coming from the federal government…
…HK: I really do believe both sides share an earnest belief in the urgency to address the problem, and I think if we can recognize that shared earnestness to address it, I am more optimistic. Our system feels more chaotic, but to your point, amidst the chaos, very smart capable people pull the good ideas to the forefront and they make them work. And so we may seem more chaotic and less organized than say the rest of the world, but that’s kind of simultaneously our superpower.
That is the US in a nutshell, absolutely.
HK: It is a little our superpower because some weirdo comes out with something that no one considered and turns out that’s a more viable disruptive path, and I do think that noise, you’ve got to kind of be able to be zen about it at some level and say something good may yet come hopefully…
…DK: The last thing I would leave you with this, Ben, is this thought that I think about this every day as I listen to my daughter’s heartbeat, I talked to the chip designers at the medical device company. They’re called Medtronic, you might’ve heard of them, they have life-saving devices. I asked them about what node they’re made on, how it’s packaged, what fab it’s made from and everything else and they gave me all technical details. And the lead designer of that just kind of paused and said, “I just want you to understand this device, this pacemaker detects your daughter’s heartbeat, how it wants to beat. If she’s sleeping, it knows that. If she’s trying to walk or run, it knows that. If it needs to speed up or down, it knows that, it’s intelligent”. And he said, “Right now, I know there’s a lot of focus on AI, but that device in your daughter’s heart that’s keeping her alive, that’s AI”.
A cynical analyst might come along and say, “Actually, no, that’s an IoT device, AI is really in language models”. But I took that to heart to say, “Right, the best of what America has to the world is in part innovation, technology innovation that enables all of us to live longer, to live better, to live more peacefully and with more safety and health in our lives”, that’s AI there, it’s not just language models. It’s that too, and semiconductors and manufacturing thereof is the foundational building blocks of all of that, and we should never forget that, and it’s something that we have given to the world that we’re trying to strengthen. So it shouldn’t be a zero-sum game, it should be something that we approach with a lot of enthusiasm, but with a lot of care.
4. Mitu Gulati on Whether Trump Could Restructure US Debt (Transcript here) – Tracy Alloway, Joe Weisenthal, and Mitu Gulati
Tracy: Yes, to put it mildly. Explain this further. You said earlier a debt swap could be basically the equivalent of extending maturities on existing Treasuries. But I would hope that there is some clause in the bond documents that would rule that out. Am I wrong? I guess I’m wrong.
Mitu: Alas, you are wrong. So anybody and everybody who holds US Treasuries should go and look at the contract terms for their US Treasuries and ask the question, “Do my contract terms restrict the US Treasury Department from saying to me tomorrow, “We need a little more money, so we’re just extending the maturity of your debt by another 20 years at the interest rate that you lent to us,” – is there anything restricting that?” I don’t think you’ll find anything, really.
Joe: This blows my mind because to me, if I’m a holder of US Treasuries and the creditor says, “You know what, I’m just paying you back a long time.” To me that sounds like default. You’re saying that in the research that you’ve done, this would not trigger credit default swaps. Because to my mind my assumption would be “This breaks the entire system. This is a default, and we can’t have a default on the risk-free assets.” You’re saying that actually in the wording of the document, it’s not there.
Mitu: We have to be clear and I have to be geeky here in my law professor mode. There are at least three different types of default. So there’s a default on the contract, something you could sue somebody for breach of contract. It would not be a breach of contract. US government is allowed to do this. Then there’s default in terms of would it trigger the handful of credit default swaps that are written on US Treasuries? That sort of depends on what the credit rating agencies decide. Based on what we saw in Greece 2012, they probably would say it was a default for credit default swaps. But that doesn’t apply to everyone. So those are the two big default scenarios.
Joe: What do you mean that doesn’t apply to everyone?
Mitu: The default for a credit default swap really only applies to the people who are holding credit defaults protection on US Treasuries. So they would be able to get their money back from somebody who had provided them insurance. But the rest of us dupes like me would just have to sit with the US extending the maturities…
…Mitu: My students asked me this in class a couple of days ago so I had to sketch it out for them. I said, Step One, the US Treasury Department and the Secretary of the Treasury have authority to manage the maturities of US Treasuries. What does manage the maturities mean? It really does mean issuing bonds of different maturities, managing your yield curve. But could it include unilaterally extending the maturities? Seems implausible. But this government has pushed its legal authority in many ways. Now what is most likely to happen, the US Treasury, if they ever went down this path because, say, they needed money and rates had gone up and they wanted to take advantage of the fact that their old borrowing was at low rates, what they would do I think, the pattern we have seen is that they would extend the maturities and then Congress would quickly pass a law confirming it. That’s what we’ve seen in all of the other Trump executive orders plus Congress quickly passing a law. And then there would be lawsuits. There would be lawsuits left and right saying, “This is a violation of the Constitution.” Because remember, there’s no contractual protection. So now you have to say, “You’ve somehow taken my property and I have an implicit, moralistic right to having my money paid back at the time when you said you would pay back.”
Now that’s really tough. We have historical precedent for this going back to the 1930s when we were in deep trouble because of gold. We didn’t have enough gold to pay everyone in case they invoked their gold clauses, which entitled holders of certain US Treasuries to get paid in gold. If they had gotten paid in gold or asked to get paid in gold, US would have essentially gone broke. So the President, backed by Congress, abrogated the gold clause protection in contracts. It was thought that surely the Supreme Court would say “This is not allowed. You cannot just take away people’s contractual rights.” The Supreme Court in one of the most famous cases of that era said, “It was okay.” The markets, I don’t wanna say this, but I’m gonna say it because it’s true – the markets didn’t crash.
Joe: What year was this?
Mitu: I think it’s around 1935. I’m gonna mess up which year Congress did the abrogation and then when the Supreme Court decision came out. But the predictions were this will destroy the US ability to ever borrow in the future and that did not happen. There are some famous articles about this.
Tracy: What’s your read then on why this didn’t happen? Why did the market seem to just go, “Okay, this is unusual, but fine.”
Mitu: My read, with no proof, is that there are these rare instances where the market thinks, “This abrogation of contractual rights, while it looks like a violation of the rule of law in every which way possible, is necessary to make us all better. Therefore, instead of penalizing the government that does it, we’re going to reward them and we’re going to lend even more.” Arguably, Greece in 2012, where Greece also legislatively abrogated contractual rights and did something very similar, is a similar situation where the market didn’t penalize them anywhere near the amount that many sages on Wall Street were saying would happen. I’m not saying that that’s what would happen now. I mean, this administration seems crazy.
5. Epizone AI: Outside the Code Stack – Kevin Kelly
Our newest invention – artificial intelligence – is usually viewed in genetic terms. The binary code of AI is copied, deployed, and improved upon. New models are bred from the code of former leading models – inheriting their abilities –, and then distributed to users. One of the first significant uses for this AI is in facilitating the art of coding, and in particular helping programmers to code new and better AIs. So this DNA-like code experiences compounding improvement as it spreads into human society. We can trace the traits and abilities of AI by following its inheritance in code.
However, this genetic version of AI has been limited in its influence on humans so far. While the frontier of AI research runs fast, its adoption and diffusion runs slow. Despite some unexpected abilities, AI so far has not penetrated very deep into society. By 2025 it has disrupted our collective attention, but it has not disrupted our economy, or jobs, or our daily lives (with very few exceptions).
I propose that AI will not disrupt human daily life until it also migrates from a genetic-ish code-based substrate to a widespread, heterodox culture-like platform. AI needs to have its own culture in order to evolve faster, just as humans did. It cannot remain just a thread of improving software/hardware functions; it must become an embedded ecosystem of entities that adapt, learn, and improve outside of the code stack. This AI epizone will enable its cultural evolution, just as the human society did for humans…
…AI civilization requires a similar epizone running outside the tech stack. It begins with humans using AI everyday, and an emerging skill set of AI collaboration taught by the AI whisperers.There will be alignment protocols, and schools for shaping the moralities of AIs. There will be shamans and doctors to monitor and nurture the mental health of the AIs. There needs to be corporate best practices for internal AIs, and review committees overseeing their roles. New institutions for reviewing, hiring and recommending various species of AI. Associations of AIs that work best together. Whole departments are needed to train AIs for certain roles and applications, as some kinds of training will take time (not just downloaded). The AIs themselves will evolve AI-only interlinguals, which needs mechanisms to preserve and archive. There’ll be ecosystems of AIs co-dependent on each other. AIs that police other AIs. The AIs need libraries of content and intermediate weights, latent spaces, and petabytes of data that need to be remembered rather than re-invented. There are the human agents that have to manage the purchase of, and maintenance of, this AI epizone, at local, national and global levels. This is a civilization of AIs.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Apple and TSMC. Holdings are subject to change at any time.
