What We’re Reading (Week Ending 29 January 2023) - 29 Jan 2023
Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes 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 29 January 2023):
1. TIP514: Permanent Supply Chain Disruptions That Will Sink The Economy w/ Jim Rickards – Trey Lockerbie and Jim Rickards
[00:25:05] Trey Lockerbie: I’d like to segue here and talk about supply chains, which is what this book is not all about, but at least half or so of the book is really this huge deep dive into how supply chains work and why they’re important.
[00:25:17] Trey Lockerbie: And I wanted to kind of call out what you, I don’t know if you came up with this phrase yourself, but you refer to it as this meta supply chain. That’s what we’ve evolved into now. So as we enter this new age of potential de-globalization, well first of all, explain what a meta supply chain is, but then also how does the meta supply chain unwind potentially in de-globalization, and what would be the risks in ramification of that?
[00:25:41] Jim Rickards: Sure. Well, let’s start with a simple supply chain and we’ll kind of build up from there. So you’re in a supermarket and somebody’s buying a loaf for bread, and you say to them, where’d the bread come from? And you go, oh, well, there’s a bakery on the other side of town and they bake it, and they send it over here on a truck and I buy the bread.
[00:25:55] Jim Rickards: Okay, that’s a simple supply chain. But even that, it’s not so simple because who made the truck? You know? Where’d the diesel come from? Where was that refinery? Where’d the truck driver get his training, et cetera? Oh, the law for bread? Well, it has a wrapper. Was it plastic or paper? Well, it could be either one, but that came from somewhere.
[00:26:11] Jim Rickards: And, and then you get over to the baker. But let’s just go further. So looking at the baker, how did they bake the bread? Well, they baked it in an oven. Where’d the oven come from? You know, it’s got tempered glass and steel and the semiconductors and thermostats and all kinds of parts. Might be from 15 or 20 different countries that was assembled and put together, and then the oven was produced.
[00:26:30] Jim Rickards: Well, how do you make bread? Well use flour Well, okay. Where’d the flower come from? Oh, it came from the middle. Okay. Well how did it get from the middle to the baker? Well, it came on a truck. Oh, another truck, another diesel, another driver, et c. How did the mill make the flour? Where did they get their ingredients?
[00:26:45] Jim Rickards: Well, they got wheat from the farmers. Really? How did it get there? Well, it came on a train. Well, trains run on diesel, who built the train? You know, et cetera. Then back to the farmer where the farmer get the seas, and by the way, the farmer needs tractors and diesel fuel and workers and gps and a lot of other scientific equipment, irrigation systems, and they need fertilizer nitrogen fertilizer to grow the wheat.
[00:27:04] Jim Rickards: And where does that come from? How does it come from Russia? Russia’s in a little war right now. We’re not buying their fertilizer, you know, so forth. So you can kinda keep going. So that’s what’s called the extended supply chain. So Baker does store is the simple supply chain, but you know, farmer fertilizer, on the one hand from Russia to the store with all those intermediate inputs is the extended supply.
[00:27:24] Jim Rickards: But then if you think about it, if you think of the supply chain as being horizontal from, you know, farmer to store with 10 stops in between the transportation lanes, every one of those intersecting points is a vertical supply chain. Again, all the components in the oven, all the components in the truck, et cetera.
[00:27:41] Jim Rickards: And you pretty quickly, this is where I, I say this in the book, the supply chain is not part of the economy. The supply chain is the economy and the meta supply chain is this vertically and horizontally expanded supply chain of supply chains that I described. And you can just kinda keep going in terms of inputs and all the way back to mines and semiconductor fabrication plants and so forth.
[00:28:02] Jim Rickards: And you realize that if it’s not literally infinite, it might as well be infinite because you cannot model it. You can model it theoretically, and you can do some computational work around it, but there’s not enough computing power in the world to end, nor is there all the data in the world, nor enough proper algorithms take everything I just described and put it into a computer.
[00:28:18] Jim Rickards: It can’t be done, but you can manage it in certain ways. So that’s the meta supply chain…
…[00:39:19] Trey Lockerbie: So, speaking of China, there a sentence in the new book stood out to me, which was that you claim China’s turn towards totalitarianism is a symptom of weakness.
[00:39:29] Trey Lockerbie: And you go as far as to say that we’ve just seen Peak China, if I’m not misquoting you there. And so this really was interesting because I know your older book, currency Wars was a huge influence on Ray Dalio. He gifted it to his entire company at one point, I believe. But he just wrote a new book as well.
[00:39:43] Trey Lockerbie: And this is the Changing World order where I think he’s alluding to a world where China is actually the rising power and as we’ve just yet to see them become the next world order. Right. So I’m curious where the disconnect is here because it seems to fly into the face of his theories.
[00:39:59] Jim Rickards: Well, look, I know Ray and he is a great guy and world’s greatest head of manager and deserves a lot of crazy smart guy.
[00:40:04] Jim Rickards: He’s still kind of coming up the curve in terms of history and geopolitics and so forth. But yeah, the conventional wisdom is the 20th century was the American century. The 21st century is going to be the Chinese century or the Asian century, and they’re going to blow past the United States in the matter of years in terms of being the world’s largest economy, higher G D P technology coming on stream, artificial intelligence, quantum computing, stronger military.
[00:40:27] Jim Rickards: It’ll be at at worst, Western Pacific hegemon, if not a global hegemon, and it’s all China and they’re going to roll the world. Everything I just said is wrong. But that is the conventional wisdom and you see variations of that all over the place. You know, Jeffrey Sachs, Richard Haas, you know, Ray Dalio, all smart people, but that’s fundamentally flawed.
[00:40:46] Jim Rickards: Now, the Peak China thesis, and to give credit, and I mentioned the names in the book, there’s been advanced by Michael Becky and I forget how’s last, he’s a scholar at the John Sal’s Club of Vater National Studies. Becky’s a scholar at Tufts University and they took a hard look at this and said, no, this is as good as it gets for China right now.
[00:41:05] Jim Rickards: They point to a number of reasons and I, I can kind of go down the same list. I’ve done the same research. Charles, half the water in China is poisoned. It’s not just dirty. You gotta clean it up before you can use it. It’s poisoned. I know a lot about the mining industry. I invested mines and I know that in the US and Canada, for example, if you use cyanide to extract gold from gold or which you do, that’s pretty standard.
[00:41:25] Jim Rickards: You gotta weigh the cyanide before you use it, then use it, case it, weigh it again, and it better be the same. Like none of that cyanide can escape, you know, careful control and disposal. In China, they do the same thing. They dump the cyanide into the rivers and a lot of ize in terms of mining, industrial output and so forth.
[00:41:41] Jim Rickards: So heth water is poisoned. They don’t have that much water to begin with, not enough of the size of the country. If you look at the geography of China, half of its desert or high plateau or mountains. People picture rice pads less about 20% of the land in like the southeastern corner. Most of it’s quite high and quite dry.
[00:41:57] Jim Rickards: They don’t have enough water to begin with. They’ve got a real estate collapse that makes what happened here in 2007 look like a picnic. They’ve got massive defaults. Not, I’ve been around China, like to say I got mud on my boots, but I was wearing Italian loafers. But I was out on construction sites looking at the ghost cities being built and.
[00:42:14] Jim Rickards: And just to give you one example, in the US when you buy a house, if you get a mortgage, the mortgage lender shows up for the closing and they give the seller the check. You sign the note and they record it. And you’ve got a mortgage in China, they have a mortgage system, but you take out the mortgage before the house is even built, and then you take the money and you give it to the developer and they use it to build the house.
[00:42:33] Jim Rickards: Well, guess what? The developers stole the money. They used it to cover out their debts. The houses never got built, but you still have the mortgage, you sign the note and the banks are trying to collect on mortgages from people who never got the houses. So this is leading to some, you know, that’s not rise demonstrations and social unrest.
[00:42:49] Jim Rickards: And you know, the government’s bail out the banks and the banks are billing out the lenders, but that’s a complete real estate collapse. So the water’s poisoned real estate sector, which is one of their biggest internal investment sectors, is collapsing. There’s a dollar shortage. You see the reserves coming down, treasury information available.
[00:43:04] Jim Rickards: You look at ’em, month by month of reserves are coming down sharply and they don’t have the technological edge. Anything they’ve got, they sold from us or firms in Europe, Siemens, or something like that. And that’s not being cut off. It’s worked for them so far. When I started developing economics in the 1970s and we thought that the hard part was to get from low income to middle income, but if you could do that, then it was straight path to high income.
[00:43:26] Jim Rickards: You would just kind of keep going. Turns out that’s not true. It’s actually kind of easy to get from low income to middle income. You don’t have too much corruption, which is you bring the population from the countryside of the city and you give them basically assembly type jobs. It’s like, people say iPhones are made in China.
[00:43:41] Jim Rickards: Not really. They’re assembled in China. Those parts come from 26 different countries. The semiconductors come from South Korea, but they assemble them in China, but that’s kind of Lego style manufacturing. And you can get there and you can get to $10,000 per capita annual income, although not evenly distributed, but getting from middle income to high income.
[00:43:58] Jim Rickards: That’s really hard and that requires technology and high value added production, and they can’t get there. They’re stuck in what is known as the middle income trap. But the biggest problem, while bigger than everything I just mentioned, is they are facing and is here now. It’s going to play out over a a 50 year, 55 year period, the greatest demographic collapse in history, worse than the Black death.
[00:44:21] Jim Rickards: Worse than the 30 years war, worse than the Spanish flu of 1918, they’re going to lose 600 million people in the next 50 or 60 years. Population’s going to go from 1.4 billion to about 800 million. Now, there are a lot of different equations for gdp, but the simplest one is workforce times productivity. How many people are working times?
[00:44:40] Jim Rickards: How productive are they? That there’s your gdp? How do you maintain any kind of economy if you’re going to lose 600 million people, which they are, and it’s worse than that because they’re losing them because their birth rate is so low. The magic number or the key number is 2.1. If two people have 2.1 kids, that’s enough to keep your population constant.
[00:45:02] Jim Rickards: Like why not to? Well, the answer is infant mortality and not every birth makes it to maturity so they can have. But on average, two people have 2.1 kids that’ll keep your population constant. The replacement rate, that’s the replacement rate earth rate in China right now, they say 1.7, but they always lie about their numbers.
[00:45:20] Jim Rickards: Other experts put it at kind of 1.2. Some people think it’s one that is behind the demographic disaster. But the reason it’s worse is that while you’re not getting new births to replace the population, the existing population is getting older and hundreds of millions are moving into their seventies, eighties, and nineties.
[00:45:37] Jim Rickards: Those age groups are highly age cohorts are highly correlated with Alzheimer’s, Parkinson’s, dementia, various kinds of cognitive decline, all of which are common at that age. They’re incurable and the progressive in the sense that they get worse. So they’re there, they’re alive, but they’re not the least bit productive.
[00:45:54] Jim Rickards: And then you need a large segment of the kind of what’s called working age population 25 to 54 as caregivers. To the people in their eighties and nineties who are suffering dementia. Now, that’s a very worthy occupation, but it does not lend itself to productivity gains. There’s been no, no increase in productivity in giving someone a bath in 5,000 years.
[00:46:15] Jim Rickards: I mean, maybe okay, 1870 indoor plumbing and hot water. Nice going, but that’s it. So you’re taking productive people, putting them as caregivers, which is a, which does not lend as self to productivity increases. A large segment of your population is not productive at all and some many suffering from a severe cognitive decline.
[00:46:35] Jim Rickards: So the portion that’s left who were actually productive working age people doing productive things, not caregivers, and not people in their eighties and nineties, keeps getting smaller. Some scholars estimate that that’s actually inflationary. Because you’re going to need to pay them more. And we did see this after the Black death in the late 14th century, early 15th century, returns to labor went up, wages went up because there weren’t enough workers.
[00:46:57] Jim Rickards: Now, it didn’t last, maybe last 75 years, but eventually the monarchs got the upper hand again. But it was a good, very good period for labor because the third of the European population was dead.
2. An Interview with Gregory C. Allen About the Past, Present, and Future of the China Chip Ban – Ben Thompson and Gregory C. Allen
Well, here’s the question. Let’s skip past the chip ban. We’ll circle back to it in a moment. Is self-sufficiency in your estimation possible? So even before, again, my argument would be integrating into the chain, becoming an essential piece, is the way that you should have actually gained leverage. Now, we fast forward to the chip ban, which I want to ask you more about how it came about or whatever. But the U.S.’s sort of explicit goal is not only can you not sort of buy our most advanced chips, but you can’t buy the equipment that goes into making the chips, which means if China wants to recreate this capability, they need to not just recreate the foundry model. They also need to recreate the lithography model, the etching model, the testing equipment. There are five companies that basically make all the equipment that goes into these factories. Do you think it’s even impossible?
GA: So I think to answer the question, you have to go in a scenario kind of probability tree. And can China do it full blown alone? I think that China could get to some degree of “self-sufficiency” on its own at a price of being nowhere remotely close to the technological state of the art. So if they are willing to take a massive hit in the competitiveness of their telecommunications equipment, of their computers, of their data centers, they could get to something called self-sufficiency. That’s like if aliens attack and blow up every country on earth except for China, they will have a semiconductor industry. It will do stuff.
However, they just don’t have a by-themselves path to economic competitiveness on price, quality, quantity, et cetera. The degree of technological sophistication that the U.S. and global semiconductor equipment companies have achieved. It’s not an overstatement to say that this is the most impressive technology that humans have created, period. It’s like this, the James Webspace telescope, the CERN Large Hadron Collider. This is really, really, really hard. And we are extraordinarily good at it, and Chinese companies are not close to where we are. So that’s the by-themselves theory of the case.
Now, there’s another option, and the other option is with foreign non-U.S. assistance, because the United States is a critical player in the semiconductor equipment value chain, the export controls are designed on the basis that there are roughly 11 technologies in which U.S. industry combined has basically a hundred percent market share. So while there are other companies in the equipment industry like ASML, Tokyo Electron, they don’t make the same stuff that U.S. companies make.
And also essential components of what they make are made by, well, once-U.S. companies, but now U.S. subsidiary.
GA: Yes. And so the path towards China getting away from U.S. dependency really relies upon persuading U.S. allies to, sorry for using this word, but betray the United States. So the negotiations China is having right now is they are going to governments like Japan, the Netherlands, and to the companies in these countries and saying, “If you start making products that the U.S. currently has a monopoly on, we will give you a boatload of money.” And it’s sort of like, this is not a perfect analogy, but if you want to make an airplane. And the United States makes the wheels and the avionics, but the Dutch make the engines, and the Japanese make the structures, well, maybe Japan doesn’t make landing gear today, but they’re way, way closer to being able to make landing gear than China is.
And so that is sort of the nature of the negotiations going on right now. Right now it’s a top U.S. diplomatic priority that these export controls, which are currently unilateral, become multilateral. And that’s most urgent in the case of the Netherlands and Japan, who have an extremely high degree of sophistication in semiconductor equipment and could start producing equipment that’s analogous to what the United States currently produces and has a monopoly on, in a matter of years. But then eventually, we’ve got to multilateral this even beyond those two countries. Countries like Germany and South Korea.
Germany is I think probably the most challenging ones.
GA: Exactly. They’re not as close as the Dutch and the Japanese, but they’re again way, way closer than China.
They make the lasers, they make the mirrors, some of the most difficult and essential inputs. And I think it’s fair to say they’ve been more difficult to get on board with U.S. diplomatic initiatives when it’s in direct conflict with economic opportunity for Germany…
…Actually, I want to dive deeper on that, the whole bit about why now? Could this even have happened a few years ago without Ukraine? Without COVID? Without the general frustration with China, without Xi Jinping’s diplomatic wolf warriors, and all this sort of stuff.
At the same time, I think one of the weirdest things about it was, in some places it seemed incredibly specific, and in some places it seemed to have huge gaping holes. And there certainly seemed to be a sense of, we have to get this out now to stop the hoarding issue, which sort of happened after the Huawei bit. A lot of these sort of specifics seemed to be like, let’s look at what’s in the market and then guide directly to that. But is that the best way to approach it? So what I’m hearing from you is this was sort of an initial step that was in many respects, not even necessarily directed at China, but was directed at the rest of the industry to say, you get on board, and we’re going to demonstrate how serious we are about you getting on board by putting this out. Is that a better way to understand?
GA: There’s a few things. First, the Department of Commerce is explicitly directed when writing export controls to consider the impact of foreign substitutions of U.S. goods. So there are export controls that the United States puts upon countries that we know will not work. For an example, when Syria is embarking upon massive human rights abuses, it is illegal to export handcuffs to Syria. Do we think that that’s going to stop the Syrian police from getting handcuffs? No. Of course they’re going to find somewhere else to buy handcuffs. We know those export controls aren’t going to work. We put them on any way as a signaling mechanism.
The China export controls are not that at all. These export controls are designed to work. They are designed to significantly degrade the capacity of the Chinese military in particular to adopt AI technology. And then sort of everything else that’s built into these are the sort of locking mechanisms that are designed to ensure that that overarching goal exists. The reason why this policy is geared towards restricting the progress of the Chinese semiconductor industry is because we don’t want China replacing the U.S. chips that are prohibited.
The second thing I would say is the Biden administration has a revealed preference for speed. And I would say that’s best demonstrated by the fact that, well, there’s different types of executive actions can move at different speeds. Having a new export control policy takes a long time to get that through the inter-agency process. Something that moves faster is what’s called an is-informed letter where you just send a letter to a company and it says like, “Hey, you’re no longer going to be allowed to sell this good, policy coming later.”
Which is basically what happened to Nvidia and AMD.
GA: This is exactly what happened. This is what I mean about the revealed preference for speed. So the Biden administration looked stupid for a full month. Because in September they sent an is-informed letter to Nvidia and AMD that said, you’re no longer going to be able to sell your high-end AI chips to China. And if that was the only policy, that would’ve been a hugely self-defeating policy, it would’ve given birth to a massive growth in the Chinese domestic GPU market for almost no gain whatsoever. And there was a whole month where everybody thought like, “Oh my God, the Biden administration just did the dumbest thing ever.” But the other shoe dropped with this October policy, which is all the sort of locking mechanisms that are designed to make that initial policy work. And that’s a revealed preference for speed. They cared about a month, they cared enough about that month to look really stupid.
I’m curious about Nvidia in particular in this. So one of the limitations in the chip ban is a combination of memory interconnect…
GA: 600 gigabytes-per-second.
600 gigabytes-per-second, which is the exact specification of Nvidia’s A100 chip. So they combined the exact specification of NVIDIA’s A100 chip with a certain level of compute, which all of NVIDIA’s chips sort of surpass at this point. So NVIDIA comes out with the A800, which seems to me to be some sort of hardware gimping of existing inventory of A100 chips, so it’s now 400 gigabytes-per-second. But obviously, it has the same sort of level of compute capacity. I’m curious what the response and view of that is. Is this a violation of the spirit, even though it’s allowed? Or is it really a sophisticated understanding of the importance of memory interconnect for AI?
These AI systems, these large ones at scale are systems problems. They’re not necessarily chip problems, right? We talked about moving up the stack before, and there’s an extent to which you’re treating an entire data center as a single chip in a certain respect, and this embarrassingly parallel process is running across all these things. The limiting factor is, can you get the data in and out to these chips. Hey, sell China the fastest chips you want as long as you can’t move that data in and out and the A800? No problem. That’s what we’re seeking to accomplish. Or is there irritation that, “Look, we’re trying to do something here and you’re just taking the shortest route possible to work around these sanctions.”
GA: No, I think NVIDIA was right to make this move. I mean, if the U.S. government does not want a company to engage in an activity for national security reasons, they have to tell them that. They can’t just ask the company to know that, and go on their own journey of determining what U.S. national security interests are. This is what compliance looks like. You follow the rules as they are written. That’s on the NVIDIA side of the equation. On the Biden administration, this policy is really about training AI models in data centers and supercomputing facilities. If you want a really beefy GPU to put in your video game console, have at it. That can totally go to China. But if you want to train really powerful AI models so that you can run an authoritarian surveillance network in Xinjiang, or so that you can train a model that is used in the guidance system of a hypersonic nuclear missile, sorry, the U.S. government cannot allow that economic transaction to occur.
I mean, that’s the thing is, the Chinese AI industry is incredibly sophisticated. If you go to NeurIPS, if you go to the big AI research conferences, there are Chinese representatives from companies like SenseTime, and iFLYTEK, and on just a pure research quality basis, they belong at these conferences. They’re doing great research. But in terms of what is paying the bills for these companies, it’s Chinese government authoritarian surveillance networks. When you combine that with China’s policy of civil-military fusion in which Chinese companies that are often assumed in Western media to be purely commercial entities, they definitely are not purely commercial entities. That’s why the Biden administration felt like they had to go for this new policy.
If you’ll indulge me for a second here, when we look back on 2022 from an international relations history perspective, there are two dates that are going to echo in history. February 24th when Russia invaded Ukraine, and October 7th, when the Biden administration dropped this new AI and chips export control policy.
This export control policy is like a total reversal of 25 years of U.S. government policy on trade in technology towards China. It’s a reversal in at least two ways. First, the prior basis of policy was, “Yes, you can engage in commercial trade with Chinese companies, but no, you cannot be a technology supplier to the Chinese military.” The new policy as a response to civil-military fusion basically does away with that and says, “For technologies above this performance threshold, it’s no longer restricted on a no-military end-user basis. Now it’s restricted on a no-China basis.” That’s a big, big change. The second way that this policy is a major reversal is, historically we were allowing the sale of technology to China, but it was the older technology, two generations-
Two generations behind, yeah.
GA: Yeah, two generations behind. That was designed to allow China to progress technologically, but to restrict the pace of technological advancement to ensure that the U.S. and our allies had a durable lead. Well, this policy, it not only restricts selling all the most advanced equipment to anywhere in China, but for Chinese companies that are already operating advanced facilities like SMIC’s 14-nanometer facility and the YMTC facility, for those facilities, this is not company-wide, it’s actually restricted to those facilities.
But for those facilities, you not only can’t sell advanced semiconductor manufacturing equipment, you can’t even sell the old stuff, and you can’t even provide software updates, and you can’t even provide spare parts. This policy is designed to put those facilities out of business, full stop. Moving from a policy of restricting the pace of advancement to actively degrading the status quo of technology in China, that’s a huge policy shift. That’s why even though this policy is somewhat narrowly targeted, it’s only going after the current state-of-the-art in AI chips and semiconductor manufacturing equipment above a certain threshold, the policy reversal that is embodied in this decision is so much larger than just AI and chips.
3. Are Declining Interest Rates Responsible for Stock Growth? – Nick Maggiulli
To get started, let’s examine how changes in interest rates have impacted U.S. stock prices throughout history. To do this, I plotted the total real return in U.S. stocks (over the prior five years) against the absolute change in the 10-Year Treasury rate (over the prior five years) since 1914 (when the Federal Reserve was first established)…
…As you can see, there seems to be somewhat of an inverse relationship between the change in Treasury rates and stock performance (at least at the extremes). When rates decline by a lot, stocks tend to rise, and vice versa… However, if you dig into the data a bit more, you’d realize that most of this relationship is derived from a single period—the 1980s…
…This suggests that most of the impact that declining interest rates had on stock prices occurred during this outlier period. Once you remove it, the connection between rates and stock performance isn’t as straight-forward…
…If declining interest rates don’t reliably impact stock prices, then what else is driving returns? One word—earnings.
To demonstrate this, let’s look at the percentage change in real price and real earnings of the S&P 500 from May 1997 to September 2022 (the latest data available):
As you can see, the total changes in real prices and real earnings of the S&P 500 are basically identical over this time period. This is true despite the fact that the 10-Year Treasury rate decreased from 6.7% in May 1997 to 3.5% by September 2022. This suggests that the increase in stock prices during this time can be attributed almost entirely to earnings growth and not necessarily to the decline in interest rates.
Of course, declining interest rates could increase earnings growth by stimulating economic activity, but that’s much harder to prove. However, there are times when declining interest rates lead to increased stock prices that aren’t hard to prove.
For example, if we were to plot the percentage change in real price and earnings of the S&P 500 from September 1982 to May 1997 (the period before the period above), we would see that earnings growth was not responsible for most of the increase in stock prices:
Over this time period, the 10-Year Treasury rate declined from 12.3% to 6.7% and stocks became more attractive as a result. And, as stocks became more attractive, investors started bidding up their prices more quickly than earnings were rising. This is known as multiple expansion or an increase in valuations. In other words, investors were willing to pay more for the same amount of earnings.
However, from the early 1980s to the mid-1990s is only period over the past four decades that I can say with certainty was influenced by a decline in interest rates. All of the growth in the stock market since this point (May 1997 onward) could, technically, be attributed to earnings growth (as demonstrated above).
While reality is far more complex than this, my analysis suggests that declining interest rates are far more important at the extremes. When the 10-Year Treasury declined from 15.3% in September 1981 to 6.7% by May 1997, that increased stock multiples much more than any rate decline that came after.
4. DeepMind’s CEO Helped Take AI Mainstream. Now He’s Urging Caution – Billy Perrigo
DeepMind—a subsidiary of Google’s parent company, Alphabet—is one of the world’s leading artificial intelligence labs. Last summer it announced that one of its algorithms, AlphaFold, had predicted the 3D structures of nearly all the proteins known to humanity, and that the company was making the technology behind it freely available. Scientists had long been familiar with the sequences of amino acids that make up proteins, the building blocks of life, but had never cracked how they fold up into the complex 3D shapes so crucial to their behavior in the human body. AlphaFold has already been a force multiplier for hundreds of thousands of scientists working on efforts such as developing malaria vaccines, fighting antibiotic resistance, and tackling plastic pollution, the company says. Now DeepMind is applying similar machine-learning techniques to the puzzle of nuclear fusion, hoping it helps yield an abundant source of cheap, zero-carbon energy that could wean the global economy off fossil fuels at a critical juncture in the climate crisis.
Hassabis says these efforts are just the beginning. He and his colleagues have been working toward a much grander ambition: creating artificial general intelligence, or AGI, by building machines that can think, learn, and be set to solve humanity’s toughest problems. Today’s AI is narrow, brittle, and often not very intelligent at all. But AGI, Hassabis believes, will be an “epoch-defining” technology—like the harnessing of electricity—that will change the very fabric of human life. If he’s right, it could earn him a place in history that would relegate the namesakes of his meeting rooms to mere footnotes.
But with AI’s promise also comes peril. In recent months, researchers building an AI system to design new drugs revealed that their tool could be easily repurposed to make deadly new chemicals. A separate AI model trained to spew out toxic hate speech went viral, exemplifying the risk to vulnerable communities online. And inside AI labs around the world, policy experts were grappling with near-term questions like what to do when an AI has the potential to be commandeered by rogue states to mount widespread hacking campaigns or infer state-level nuclear secrets. In December 2022, ChatGPT, a chatbot designed by DeepMind’s rival OpenAI, went viral for its seeming ability to write almost like a human—but faced criticism for its susceptibility to racism and misinformation. So did the tiny company Prisma Labs, for its Lensa app’s AI-enhanced selfies. But many users complained Lensa sexualized their images, revealing biases in its training data. What was once a field of a few deep-pocketed tech companies is becoming increasingly accessible. As computing power becomes cheaper and AI techniques become better known, you no longer need a high-walled cathedral to perform cutting-edge research.
It is in this uncertain climate that Hassabis agrees to a rare interview, to issue a stark warning about his growing concerns. “I would advocate not moving fast and breaking things,” he says, referring to an old Facebook motto that encouraged engineers to release their technologies into the world first and fix any problems that arose later. The phrase has since become synonymous with disruption. That culture, subsequently emulated by a generation of startups, helped Facebook rocket to 3 billion users. But it also left the company entirely unprepared when disinformation, hate speech, and even incitement to genocide began appearing on its platform. Hassabis sees a similarly worrying trend developing with AI. He says AI is now “on the cusp” of being able to make tools that could be deeply damaging to human civilization, and urges his competitors to proceed with more caution than before. “When it comes to very powerful technologies—and obviously AI is going to be one of the most powerful ever—we need to be careful,” he says. “Not everybody is thinking about those things. It’s like experimentalists, many of whom don’t realize they’re holding dangerous material.” Worse still, Hassabis points out, we are the guinea pigs…
…By 2013, when DeepMind was three years old, Google came knocking. A team of Google executives flew to London in a private jet, and Hassabis wowed them by showing them a prototype AI his team had taught to play the computer game Breakout. DeepMind’s signature technique behind the algorithm, reinforcement learning, was something Google wasn’t doing at the time. It was inspired by how the human brain learns, an understanding Hassabis had developed during his time as a neuroscientist. The AI would play the game millions of times, and was rewarded every time it scored some points. Through a process of points-based reinforcement, it would learn the optimum strategy. Hassabis and his colleagues fervently believed in training AI in game environments, and the dividends of the approach impressed the Google executives. “I loved them immediately,” says Alan Eustace, a former senior vice president at Google who led the scouting trip.
Hassabis’ focus on the dangers of AI was evident from his first conversation with Eustace. “He was thoughtful enough to understand that the technology had long-term societal implications, and he wanted to understand those before the technology was invented, not after the technology was deployed,” Eustace says. “It’s like chess. What’s the endgame? How is it going to develop, not just two steps ahead, but 20 steps ahead?”
Eustace assured Hassabis that Google shared those concerns, and that DeepMind’s interests were aligned with its own. Google’s mission, Eustace said, was to index all of humanity’s knowledge, make it accessible, and ultimately raise the IQ of the world. “I think that resonated,” he says. The following year, Google acquired DeepMind for some $500 million. Hassabis turned down a bigger offer from Facebook. One reason, he says, was that, unlike Facebook, Google was “very happy to accept” DeepMind’s ethical red lines “as part of the acquisition.” (There were reports at the time that Google agreed to set up an independent ethics board to ensure these lines were not crossed.) The founders of the fledgling AI lab also reasoned that the megacorporation’s deep pockets would allow them access to talent and computing power that they otherwise couldn’t afford.
In a glass cabinet spanning the far wall of the lobby at DeepMind’s London headquarters, among other memorabilia from the first 12 years of the company’s life, sits a large square of wood daubed with black scribbles. It’s a souvenir from DeepMind’s first major coup. Soon after the Google acquisition, the company had set itself the challenge of designing an algorithm that could beat the best player in the world at the ancient Chinese board game Go. Chess had long ago been conquered by brute-force computer programming, but Go was far more complex; the best AI algorithms were still no match for top human players. DeepMind tackled the problem the same way they’d cracked Breakout. It built a program that, after being taught the rules of the game by observing human play, would play virtually against itself millions of times. Through reinforcement learning, the algorithm would update itself, reducing the “weights” of decisions that made it more likely to lose the game, and increasing the “weights” that made it more likely to win. At a tournament in Korea in March 2016, the algorithm—called AlphaGo—went up against Lee Sedol, one of the world’s top Go players. AlphaGo beat him four games to one. With a black marker pen, the defeated Lee scrawled his signature on the back of the Go board on which the fateful game had been played. Hassabis signed on behalf of AlphaGo, and DeepMind kept the board as a trophy. Forecasters had not expected the milestone to be passed for a decade. It was a vindication of Hassabis’ pitch to Google: that the best way to push the frontier of AI was to focus on reinforcement learning in game environments.
But just as DeepMind was scaling new heights, things were beginning to get complicated. In 2015, two of its earliest investors, billionaires Peter Thiel and Elon Musk, symbolically turned their backs on DeepMind by funding rival startup OpenAI. That lab, subsequently bankrolled by $1 billion from Microsoft, also believed in the possibility of AGI, but it had a very different philosophy for how to get there. It wasn’t as interested in games. Much of its research focused not on reinforcement learning but on unsupervised learning, a different technique that involves scraping vast quantities of data from the internet and pumping it through neural networks. As computers became more powerful and data more abundant, those techniques appeared to be making huge strides in capability.
While DeepMind, Google, and other AI labs had been working on similar research behind closed doors, OpenAI was more willing to let the public use its tools. In late 2022 it launched DALL·E 2, which can generate an image of almost any search term imaginable, and the chatbot ChatGPT. Because both of these tools were trained on data scraped from the internet, they were plagued by structural biases and inaccuracies. DALL·E 2 is likely to illustrate “lawyers” as old white men and “flight attendants” as young beautiful women, while ChatGPT is prone to confident assertions of false information. In the wrong hands, a 2021 DeepMind research paper says, language-generation tools like ChatGPT and its predecessor GPT-3 could turbocharge the spread of disinformation, facilitate government censorship or surveillance, and perpetuate harmful stereotypes under the guise of objectivity. (OpenAI acknowledges its apps have limitations, including biases, but says that it’s working to minimize them and that its mission is to build safe AGI to benefit humanity.)
5. Analysis: Xi puts top brain in charge of Taiwan unification strategy – Katsuji Nakazawa
A source familiar with the inner workings of the Chinese Communist Party has pulled back the curtain on General Secretary Xi Jinping’s leadership reshuffle last October.
Why were some leaders retained to serve another term, while others were shown the door?
On the Politburo Standing Committee, there were three members who were 67 years old, technically under the retirement age of 68. All three of them could have stayed, but only one did.
The ones who stepped down were No. 2, Premier Li Keqiang and No. 4 Wang Yang. Only No. 5 Wang Huning stayed on and was promoted in the new lineup.
The source noted that this top leadership change hints at Xi’s political strategy as he aims for a fourth term. “Wang Huning’s mission is to lay the groundwork for Taiwan unification.”
If Wang Huning was retained to handle the Taiwan file, this would be the result of the failure of the “one country, two systems” in Hong Kong.
After massive pro-democracy demonstrations shook Hong Kong in 2019, Beijing quickly enacted a national security law for the special administrative region. It spelled the end of a free Hong Kong…
…On Jan. 18, state-run Xinhua News Agency announced the new members of the Chinese People’s Political Consultative Conference, the country’s top political advisory body. The inclusion of Wang Huning signaled that he would assume the role of CPPCC chairman, succeeding Wang Yang.
One of the CPPCC’s role is to set strategies for China’s “united front work,” including drawing Taiwan to the Chinese side.
Under this framework, Wang Huning is also expected to become the deputy director of the Central Leading Group for Taiwan Affairs, the party’s top decision-making body on China’s Taiwan policy. The top director is Xi.
So what role will Wang play in formulating a Taiwan policy during Xi’s third term?
One source knowledgeable of China-Taiwan relations noted that Wang will be tasked with writing a theoretical unification strategy fit for the Xi era.
“One may assume that a threat of China using force to unify Taiwan is imminent, but this is not the case. The first step is to launch a new theory that will replace Deng’s one country, two systems. Then pressure will be put on Taiwan based on it,” the source explained.
The source expects this theory to become a yardstick with which to measure progress and to decide if a military operation is necessary…
…Wang Huning will be supported by Wang Yi, the 69-year-old former foreign minister, who was promoted to the Politburo. His promotion went against the party’s traditional retirement rule that stipulates that officials do not assume new higher posts after they are 68.
Wang Yi also became director of the party’s Office of the Central Foreign Affairs Commission, making him China’s top-ranking diplomat.
Needless to say, the top diplomat reports to Xi on foreign affairs and security matters. But for policies involving Taiwan unification and relations with the U.S., Wang Huning is also in Wang Yi’s reporting line.
This is because Wang Yi will become secretary general of the Central Leading Group for Taiwan Affairs, where Wang Huning will serve as deputy director. Wang Yi once served as the director of the Taiwan Affairs Office of the State Council, China’s government.
As a Politburo Standing Committee member, Wang Huning in one of China’s top seven and has a much higher level of authority than Wang Yi, a Politburo member.
Xi wants to chalk up an achievement in regard to Taiwan at any cost over the next five years, which would help his quest to seek a fourth term as head of the party in 2027.
China’s policies related to Taiwan will be spearheaded by these two Wangs…
…Xi acquired ultimate power in October. While the use of force against Taiwan is not deemed imminent, Xi could launch an offensive at the snap of his fingers.
Last summer, China held military exercises around Taiwan and fired missiles. The display of force came in response to then U.S. House Speaker Nancy Pelosi’s visit to the island. Since then, Taiwan has become increasingly alarmed at the possibility of a military invasion by China.
Russia’s all-out invasion of Ukraine has also shocked the island.
China hopes to see the independence-leaning DPP ousted from power in 2024. But as relations between China and Taiwan are extremely tense, it is difficult to decide upon the timing of working out a new Taiwan unification strategy.
If the content of the new strategy is taken as merely a threat against Taiwan, it could backfire. Although China wants to support the KMT, it could end up saving the DPP.
“China will have no choice but to take a wait-and-see attitude for the time being,” one pundit said. “The timing of announcing a new Taiwan unification strategy is probably undetermined. It may be still a long way off.”
6. The forgotten mistake that killed Japan’s software industry – Tim Romero
No, for the sake of this podcast I’m going to assume that we are all in agreement that on average, Japanese software. is just … awful.
That way we can spend our time talking about something far more interesting. We are going to walk though the economic events and the political forces that made today’s poor quality of Japanese software almost inventible,
And by the end, I think it will give you a completely new way of looking at the Japanese software industry.
You see, the story of Japanese software, is not really about software. No, this is the story of Japanese innovation itself. The story of the ongoing struggle between disruption and control. It’s a story that involves, war, secret cartels, scrappy rebels, betrayal, rebirth, and perhaps redemption…
…In same way that the zaibatsu defined the economic miracle that was Japan’s Meji-era expansion, the keiretsu would come define the economic miracle that was Japan’s post war expansion.
Today there are six major and a couple dozen minor keiretsu groups, and during Japan’s economic expansion, as much as possible, they kept their business within the keiretsu family.
Projects were financed by the keiretsu bank, the materials and know-how were imported by the keiretsu trading company, and the final products would be assembled in the appropriate keiretsu brand’s factory. And supporting all of these flagship brands were, and still are, tens of thousands of very small, exclusive manufacturers that make up the keiretsu supply chain — and the bulk of the Japanese economy.
And with the exception of a tiny handful of true startup companies like Honda and Sony, all of Japan’s brands that were famous before the year 2000 or so, are keiretsu brands.
And for those of you who think big companies can’t innovate, let me remind you that from the 50s to the 70s, these keiretsu groups began innovating, disrupting, and dominating almost every industry on the planet; from cars, to cameras, to machine parts, to steel, to semiconductors, to watches, to home electronics, Japan’s keiretsu simply rewrote the rules.
But how did the keiretsu do in the world of software development? Well, pretty darn well, actually.
It’s important to remember, though, that the software industry in the 60s and 70s was very different than it is today. The software development process itself was actually rather similar. Fred Brooks wrote The Mythical Man Month about his experience during this era, and it remains as one of best books on software engineering and project management today.
But the way software was bought and sold was completely different. In the 60s and 70s, software was written for specific and very expensive hardware, and the software requirements were negotiated as part of the overall purchase contract. Software was not viewed so much as a product, but more like a service, similar to integration, training, and ongoing support and maintenance. It was usually sold on a time-and-materials basis, and sometimes it was just thrown in for free to sweeten the deal. The real money was in the hardware.
Software in this time (both in Japan and globally) was written to meet the spec. It did not matter if it was creative, innovative, easy to use, or elegant, it just had to meet the spec. In fact, trying to build exceptional software in this era was considered a waste of resources. After all, the product had already been sold and the contracts had already been signed. The goal back then, just like many system integration projects today, was to build software that was just good enough to get the client to sign off on it as complete.
Software that met the customer’s spec was, by definition, good software.
Japan’s keiretsu did well in the age of big-iron. Although Fujitsu, NEC, and Hitachi never seriously challenged IBM and Univac’s global dominance in the 60s and 70s, they did pretty well in mini-computers and large office systems.
They were innovators.
However, when the PC revolution arrived in the late 1980s, Japanese industry as a whole was hopelessly unprepared, and not for the reasons you might think.
The reason Japanese software development stopped advancing in the 1980s had nothing to do with a lack of talented software developers. It was a result of Japan’s new economic structure as a whole, and the keiretsu in particular.
As a market, personal computers were something fundamentally new. Sure, the core technology and the hardware were direct continuations from the previous era, but this new market was completely different.
The PC market quickly coalesced around a small number of standardized operating systems and hardware architectures. The keiretsu did pretty well in hardware side of this market, making some really impressive machines, particularly laptops.
But a market for non-spec or “shrink-wrap” software was something new to everyone. It required delighting the customer, and knowing what they wanted before they did. It was the kind of challenge that the keiretsu of the 60s and 70s would have thrown themselves into whole-heartedly, innovated aggressively, and then dominated.
But things in Japan had become very different in the 1980s.
Here was a chance to define and lead a new global industry. A chance for the keiretsu to build a software industry from the ground up.
But, wait a minute. Why should they?
Sure, back in the 60s when Japan’s economy was small, survival required looking outwards, competing globally, making long-term investments, and innovating to make the best products in the world.
But this was the 80s! Japan was the second-largest economy on the planet and in the middle of the largest economic boom the world had ever seen. This was the era of Japan as Number 1, with economists predicting Japan’s GNP would be larger than America’s within a decade.
With such a lucrative, and pretty well protected, market right at their fingertips it made much more sense for for the keiretsu to focus on the easy money rather than to take risky and expensive bets on an uncertain and emerging global market.
Each keiretsu group had their own technology firm who started selling PCs and software, some to consumers, but the big money was in corporate sales. And since the keiretsu liked to keep the business in the family, these technology companies grew and profited by selling to their captive customers within their keiretsu group. And just like before, they made the real money integration, and customization.
An unfortunate result of this is that the big Systems Integration companies or “SIs” emerged as powerful players, and Japan’s software firms never had to compete globally, or even with each other.
Japan simply missed the opportunity to develop a globally relevant PC software industry…
…I mark 2010 as the year Japan’s software developers finally started stepping into the spotlight, although things starting moving a bit before that.
There were two triggers that led to this development. First, the emergence of cloud computing and second, the introduction of the smartphone. Although these were both technological developments, it was not the technology itself that led to the change.
Cloud computing drastically reduced the capital and time required to start a startup. In the dot-com era a decade before, starting an internet startup required purchasing racks of servers and paying system administrators to keep them running, but suddenly fully configured, maintained, and secure serves could be had for a few cents per minute — pay as you go.
Suddenly Japan’s software developers didn’t need to explain their idea to a VC and convince them that it would sell. They could just build things and get people to start using them and start paying for them. And that’s just what they did.
The other important development was the introduction of the iPhone in 2007 and Android a year later. Not just because of the technology, but because of how it changed the software business model…
…As we talk here together at the start of 2023, what does the future look like for Japanese software?
Japan has had a lot of catching up to do over the past fifteen years. After basically sitting out the global PC and dot-com revolutions, Japanese software developers have been making up for lost time and in the startup space. Japan is developing a competitive software market in some areas, but on average, there is still a long way to go.
Japan’s once dominant Systems Integrators will continue to see their power decline. Their customer lock-in is fading fast, and B2B SaaS software startups are letting Japanese enterprises leapfrog to modern IT systems for less than costs to maintain their SI-run legacy systems.
The SIs won’t disappear, of course. There will always be a need for good systems integrators, and the more forward thinking ones are already trying to reinvent themselves. However, the days when the SIs dictated their clients’ IT strategy are coming to a close. That is a very good thing for Japanese software, Japanese startups, and Japanese competitiveness as a whole.
The Kishida administration has made startups a national priority, and the importance of quality software and software startups in Japan has never been higher.
7. The Bank that Never Sold – Marc Rubinstein
Standard Chartered traces its roots back to the height of the British Empire. In order to finance expansion overseas, specialist banks were set up to facilitate trade. One of them, the Chartered Bank of India, Australia and China was founded to serve the markets of … India, Australia and China. The bank ended up not getting a charter for Australia, but succeeded in establishing a foothold in the other two fledgling markets.
Chartered’s model was that of an “exchange bank”. Capital was raised in the City of London and shipped out, often as gold or silver coins in wooden crates, to support currency transactions for British companies across the main ports of the East from Bombay to Shanghai. To mitigate against risk, the bank employed a portfolio approach, opening up over 20 overseas branches. By 1928, Chartered Bank ranked alongside HSBC as one of the largest overseas banks launched out of the UK, focused on trade finance and foreign exchange services.
The growth of Chartered and other overseas banks caught the eye of UK domestic banks. By now, the market at home had consolidated around five main banks. Previously cautious that “there would be something mildly improper about using their UK depositors’ money to fund lending in distant climes,” they began to revise their opinion.
But fortunes turned as the 1930s augured a collapse in international trade. Chartered Bank was additionally shut out of some of its core markets, in particular China, following political upheaval after the end of the Second World War. Retreating to Hong Kong, the bank managed to carve out a profitable niche. It increasingly dealt with local companies, rather than just British agency firms, and grew its loan book. To support its franchise, it later established a network of local retail outlets in order to accumulate local deposits. By the mid 1960s, Chartered Bank was adding branches at a rate of two or three a month. The success of its business in Hong Kong marked out a new future for the bank, no longer dependent on the traditional trade links of the Empire.
By the 1960s, most vestiges of the British Empire had faded. A devaluation of the pound ended its role as a reserve currency and led to the breakup of the sterling system that had underpinned the UK overseas banking model for years. Competition from US banks increased. In response, Chartered Bank merged with Standard Bank of South Africa to create a more global overseas bank with operations across Asia, the Middle East and Africa.
Like Chartered Bank, Standard Bank had been established to facilitate trade flows, in its case to Africa. Its legacy was similar. British Prime Minister John Major, a former employee of Standard Bank, wrote that both “relied for many decades on adventurous young recruits from Britain who were keen to work overseas”. (He worked for Standard Bank in Nigeria.)
For Standard, the diamond industry provided a historic route to good fortune; by the late nineteenth century, it operated almost 100 branches in South Africa, practising an almost central banking role in the country. But it, too, had to adapt to the shifting macro climate. The isolation of South Africa as Apartheid became entrenched prompted Standard to spin off its South Africa business and focus on other markets in Africa, which it consolidated through its acquisition of the Bank of British West Africa.
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 Alphabet and Meta Platforms (parent of Facebook). Holdings are subject to change at any time.
