We’ve constantly been sharing a list of our recent reads in our weekly emails for The Good Investors.

Do subscribe for our weekly updates through the orange box in the blog (it’s on the side if you’re using a computer, and all the way at the bottom if you’re using mobile) – it’s free!

But since our readership-audience for The Good Investors is wider than our subscriber base, we think sharing the reading list regularly on the blog itself can benefit even more people. The articles we share touch on a wide range of topics, including investing, business, and the world in general. 

Here are the articles for the week ending 20 July 2025:

1. Sweatshop data is over – Tamay Besiroglu, Matthew Barnett, Ege Erdil

Historically, the importance of data has been underrated in the field of AI. Decades ago, many assumed the key to AGI would come from devising the right “theory of intelligence”, which we could then implement by hand; the role of training data was sidelined.

Despite being trained on more compute than GPT-3, AlphaGo Zero could only play Go, while GPT-3 could write essays, code, translate languages, and assist with countless other tasks. The main difference was training data. AlphaGo Zero learned from Go games, whereas GPT-3 learned from natural language. This meant that while Google was playing games, OpenAI was able to seize the opportunity of a lifetime. What you train on matters.

We may soon witness a similar lesson if AI labs continue to scale up their models without similarly scaling up the quality of their training environments. Many have observed that pretraining is already saturating. GPT-4.5, while impressive in its own right, didn’t feel like a major generational leap in the way GPT-4 did over GPT-3.5.

The recent reinforcement learning with verifiable rewards (RLVR) paradigm seeks to revive progress by getting AIs to learn how to perform formally checkable reasoning inside contained environments. What we’ve seen so far is necessary for progress, but it is far from sufficient. Current methods will get us to the point where AIs can prove theorems and solve hard puzzles, but it won’t be enough to get models to deal with the open-ended nature of reality, where the quality of our actions cannot be so easily “verified” as either correct or incorrect.

To make progress, there’s no way around designing better rewards, and ultimately better RL environments.

2. Silk, Porcelain, Tea, Opium: 2000 Years of Trade Deficit with China – Tomas Pueyo

The West has had deficits with China for over 2,000 years, and they have had a massive impact on world history, from the opening of global trade routes, to the establishment of colonies, colonial policies, international wars, the emergence of nation-states, the politics of present-day China and the US…

…Romans loved luxury goods:

India, China and the Arabian peninsula take one hundred million sesterces1 from our empire per annum at a conservative estimate: that is what our luxuries and women cost us—Pliny the Elder, Natural History (77–79 AD).

Of these, silk was the biggest import from China. In 14 AD the Senate prohibited the wearing of silk by men!

To pay for it, Romans traded glassware, amber, wine, carpets, and other goods,2 but they didn’t make up for the value of what Romans bought from China. And in general, Chinese traders preferred money—mostly gold and silver—over other goods…

…Europeans obsessed about producing silk locally, but they didn’t know how to make it and didn’t have silkworms: China had protected its near-monopoly on silk for many centuries thanks to imperial orders to execute anybody caught trying to export silkworms or their eggs. The only way to succeed was by stealing them, and that’s precisely what two Christian monks did around 550 AD, risking their lives to smuggle silkworms hidden inside their canes.

This started silk production in the Eastern Roman Empire, which would slowly permeate through the rest of Europe.

This might have been the first time Chinese manufacturing prowess caused a trade imbalance in the West that required political intervention…

…Porcelain could only start reaching Europe in the 1500s,4 which is not a coincidence either: Porcelain was too heavy and fragile for overland routes, so it needed a maritime route to reach Europe. The Portuguese found a path to the Indies circumventing Africa just around 1500…

…Chinese porcelain was so much thinner, whiter and more translucent than local wares that European nobility really prized it…

…You know how nowadays Westerners design some products and then they send those designs to China for manufacture?

Porcelain is another example of China manufacturing products that Europeans craved, but again it didn’t need anything Europeans produced. Except for silver. So silver flowed from Europe to China. From 1500 to 1800, Bolivia and Mexico’s mines produced about 80% of the world’s silver; 30% of that eventually ended up in China!

Europeans hated that flow, as the silver disappeared as fast as it was produced, so they tried to stop it. Of course, the most incentivized were the countries who didn’t have access to either silver or trade with China. This is why the Italians tried to copy porcelain in the late 1500s with Medici porcelain, although they largely failed. By the early 1700s, Germans succeeded. A few years later, in 1712, the French Jesuit father Francois Xavier d’Entrecolles published the secrets of porcelain making in Europe, which he had read about and witnessed in China. In the following decades, the local production of porcelain increased and the import of Chinese porcelain fell…

…Tea’s ever-escalating trade imbalance with China became a serious economic problem, so much so that the British King George III sent an envoy to the Chinese Emperor to ask for more trade liberalization. These are excerpts of the Emperor’s response:

Our Celestial Empire possesses all things in prolific abundance and lacks no product within its own borders. There is therefore no need to import the manufactures of outside barbarians in exchange for our own produce. But as the tea, silk and porcelain which the Celestial Empire produces, are absolute necessities to European nations and to yourselves, we have permitted, as a signal mark of favor, that foreign merchants should be established at Canton, so that your wants might be supplied and your country thus participate in our beneficence.

So what did the British do to solve the trade imbalance? Two things. One is that the East India Company sent Scottish botanist Robert Fortune to China to purchase and export Chinese tea plants in the 1850s. This kick-started tea production in India, which grew over the following decades, reducing the share of Chinese tea consumed. Here we have, for the third time, a smuggling of Chinese production know-how to reduce trade imbalances…

…When the British conquered India8 in the late 1700s, they were very conscious about their trade imbalance with China, so they looked for any way to reduce it. They found the right tool in opium. They devised a plan to produce it in India and sell it in China. So the British drove local farmers in eastern India out of crop production and into poppies, from which opium is derived.

Then, the British introduced opium smoking in China…

…The Emperor Jiaqing noticed all this so he published an edict to stop it in 1810:

Opium has a harm. Opium is a poison, undermining our good customs and morality. Its use is prohibited by law.

But the government couldn’t enforce it. When the Chinese government finally cracked down on opium in 1839, the opium trade was paying for all the tea trade and then some, so the British reacted to protect the trade and attacked China; this was the First Opium War.

Britain won and bent China’s arm: It would be allowed to sell opium in China. It also took over Hong Kong.

There would be another Opium War, after which the British, and then other Westerners10 could reach far inland in China to sell opium. The deficit to China became a surplus. Over the following decades, opium addiction became widespread. By 1949, 4.4% of Chinese people were addicted. Local farmers replaced their crops with opium. Governments used opium taxes to finance themselves, and this lasted until the Communist Party had a strong enough chokehold on society and culture to finally ban opium.

This is what the Chinese call the century of humiliation, when China went from the richest and most advanced nation of the world to a dirt poor backwater.

3. The Codes AI Can’t Crack – Taras Grescoe

Since 2018, neural networks trained on cuneiform, the writing system of Mesopotamia, have been able to fill in lost verses from the story of Gilgamesh, the world’s earliest known epic poem. In 2023, a project known as the Vesuvius Challenge used 3D scanners and artificial intelligence to restore handwritten texts that hadn’t been read in 2,000 years, revealing previously unknown works by Epicurus and other philosophers. (The scrolls came from a luxurious villa in Herculaneum, buried during the same eruption of Mount Vesuvius that destroyed Pompeii. When scholars had previously tried to unroll them, the carbonized papyrus crumbled to dust.)

Yet despite these advances, a dozen or so ancient scripts — the writing systems used to transcribe spoken language — remain undeciphered. These include such mysteries as the one-of-a-kind Phaistos Disk, a spiral of 45 symbols found on a single sixteen-inch clay disk in a Minoan palace on Crete, and Proto-Elamite, a script used 5,000 years ago in what is now Iran, which may have consisted of a thousand distinct symbols. Some, like Cypro-Minoan — which transcribes a language spoken in the Late Bronze Age on Cyprus — are tantalizingly similar to early European scripts that have already been fully deciphered. Others, like the quipu of the Andes — intricately knotted ropes made of the wool of llamas, vicuñas, and alpacas — stretch our definitions of how speech can be transformed into writing…

…Cracking these ancient codes may seem like the kind of challenge AI is ideally suited to solve. After all, neural networks have already bested human champions at chess, as well as the most complex of all games, Go. They can detect cancer in medical images, predict protein structures, synthesize novel drugs, and converse fluently and persuasively in 200 languages. Given AI’s ability to find order in complex sets of data, surely assigning meaning to ancient symbols would be child’s play.

But if the example of Ithaca shows the promise of AI in the study of the past, these mystery scripts reveal its limitations. Artificial neural networks might prove a crucial tool, but true progress will come through collaboration between human neural networks: the intuitions and expertise stored in the heads of scholars, working in different disciplines in real-world settings…

…Ithaca was trained on ancient Greek, a language we’ve long known how to read, and whose entire corpus amounts to tens of thousands of inscriptions. The AI models that have filled in lost verses of Gilgamesh are trained on cuneiform, whose corpus is even larger: hundreds of thousands of cuneiform tablets can be found in the storerooms of the world’s museums, many of them still untranslated. The problem with mystery scripts like Linear A, Cypro-Minoan, Rongorongo, and Harappan is that the total number of known inscriptions can be counted in the thousands, and sometimes in the hundreds. Not only that, in most cases we have no idea what spoken language they’re meant to encode…

… Two of the greatest intellectual feats of the 20th century involved the decipherment of ancient writing systems. In 01952, when Michael Ventris, a young English architect, announced that he’d cracked the code of Linear B, a script used in Bronze Age Crete, newspapers likened the accomplishment to the scaling of Mount Everest. (Behind the scenes, the crucial grouping and classifying of characters on 180,000 index cards into common roots — the grunt work that would now be performed by AI — was done by Alice Kober, a chain-smoking instructor from Brooklyn College.)

The decipherment of the Maya script, which is capable of recording all human thought using bulbous jaguars, frogs, warriors’ heads, and other stylized glyphs, involved a decades-long collaboration between Yuri Knorozov, a Soviet epigrapher, and American scholars working on excavations in the jungles of Central America.

While the interpreting of Egyptian hieroglyphics is held up as a triumph of human ingenuity, the Linear B and Mayan codes were cracked without the help of a Rosetta Stone to point the way. With Linear B, the breakthrough came when Ventris broke with the established thinking, which held that it transcribed Etruscan — a script scholars can read aloud, but whose meaning still remains elusive — and realized that it corresponded to a form of archaic Greek spoken 500 years before Homer. In the case of ancient Mayan, long thought to be a cartoonish depiction of universal ideas, it was only when scholars acknowledged that it might transcribe the ancestors of the languages spoken by contemporary Maya people that the decipherment really began. Today, we can read 85% of the glyphs; it is even possible to translate Shakespeare’s Hamlet into ancient Mayan.

Collaborating across cultures and disciplines, and carrying out paradigm-shedding leaps of intuition, are not the strong points of existing artificial neural networks. But that doesn’t mean AI can’t play a role in decipherment of ancient writing systems. Miguel Valério, an epigrapher at the Autonomous University of Barcelona, has worked on Cypro-Minoan, the script used on Cyprus 3,500 years ago. Two hundred inscriptions, on golden jewelry, metal ingots, ivory plaques, and four broken clay tablets, have survived. Valério was suspicious of the scholarly orthodoxy, which attributed the great diversity in signs to the coexistence of three distinct forms of the language.

To test the theory that many of the signs were in fact allographs — that is, variants, like the capital letter “G” and “g,” its lower-case version — Valério worked with Michele Corazza, a computational linguist at the University of Bologna, to design a custom-built neural network they called Sign2Vecd. Because the model was unsupervised, it searched for patterns without applying human-imposed preconceptions to the data set.

“The machine learned how to cluster the signs,” says Valério, “but it didn’t do it simply on the basis of their resemblance, but also on the specific context of a sign in relation to other signs. It allowed us to create a three-dimensional plot of the results. We could see the signs floating in a sphere, and zoom in to see their relationship to each other, and whether they’d been written on clay or metal.”…

…A generation ago, most people were taught that writing was invented once, in Mesopotamia, about 5,500 years ago, as a tool of accountancy and state bureaucracy. From there, the standard thinking went, it spread to Egypt, and hieroglyphics were simplified into the alphabet that became the basis for recording most European languages…

…Monogenesis, the idea that the Ur-script diffused from Mesopotamia, has been replaced by the recognition that writing was invented independently in China, Egypt, Central America, and — though this remains controversial — in the Indus Valley, where 4,000 inscriptions been unearthed in sites that were home to one of the earliest large urban civilizations.

4. A 37,000-Year Chronicle of What Once Ailed Us – Carl Zimmer

On Wednesday, a team of scientists unveiled a new genetic chronicle, documenting the rise of 214 diseases across Europe and Asia over the past 37,000 years…

…The researchers examined the remains of 1,313 ancient individuals for the project. The large scale enabled the researchers to do more than just push back the earliest known occurrence of different diseases. They could also track the rise and fall of epidemics across centuries.

The oldest remains the researchers studied belonged to hunter-gatherers. Their bones and teeth contained a host of pathogens, such as hepatitis B, herpes virus and Helicobacter pylori, a stomach-dwelling bacterium.

“As far back as we go, humans have had infectious diseases,” said Eske Willerslev, a geneticist at the University of Copenhagen and an author of the new study…

…Initially, Dr. Willerslev and his colleagues assumed that they would see such diseases rise to prominence starting about 11,000 years ago. That’s when people started domesticating animals, from which new diseases could spread more easily…

…But the ancient DNA defied that expectation. The scientists found that plague and a number of other diseases jumped to people from animals thousands of years later, starting about 6,000 years ago. And those microbes did not jump into early farmers.

Instead, the new study points to nomadic tribes in Russia and Asia. Thousands of years after the dawn of agriculture, those nomads started rearing vast herds of cattle and other livestock.

Why diseases would have attacked those herders instead of earlier farmers, the scientists can’t say for sure. “We haven’t been able to come up with anything conclusive,” Dr. Willerslev said…

…The nomads expanded over the next few centuries across the steppes of Asia and eastern Europe. In that time, their pathogens thrived; the scientists frequently found several individuals in a single grave with DNA from plague or other diseases.

Those epidemics were so intense that they changed the genetic profile of the nomads. Last year, Dr. Willerslev and his colleagues found that the nomads experienced a spike in mutations that boosted their immune system and that may have helped them resist the diseases they contracted. But their active immune systems may have also attacked their own bodies, producing chronic diseases such as multiple sclerosis.

5. AI is killing the web. Can anything save it? – The Economist

Similarweb, which measures traffic to more than 100m web domains, estimates that worldwide search traffic (by humans) fell by about 15% in the year to June. Although some categories, such as hobbyists’ sites, are doing fine, others have been hit hard (see chart). Many of the most affected are just the kind that might have commonly answered search queries. Science and education sites have lost 10% of their visitors. Reference sites have lost 15%. Health sites have lost 31%.

For companies that sell advertising or subscriptions, lost visitors means lost revenue…

…Google has insisted that its use of others’ content is fair. But since it launched its AI overviews, the share of news-related searches resulting in no onward clicks has risen from 56% to 69%, estimates Similarweb. In other words, seven in ten people get their answer without visiting the page that supplied it…

…To keep the traffic and the money coming, many big content producers have negotiated licensing deals with AI companies, backed up by legal threats: what Robert Thomson, chief executive of News Corp, has dubbed “wooing and suing”. His company, which owns the Wall Street Journal and the New York Post, among other titles, has struck a deal with OpenAI. Two of its subsidiaries are suing Perplexity, another AI answer engine. The New York Times has done a deal with Amazon while suing OpenAI. Plenty of other transactions and lawsuits are going on…

…Reddit, an online forum, has licensed its user-generated content to Google for a reported $60m a year…

…The bigger problem, however, is that most of the internet’s hundreds of millions of domains are too small to either woo or sue the tech giants. Their content may be collectively essential to AI firms, but each site is individually dispensable. Even if they could join forces to bargain collectively, antitrust law would forbid it. They could block AI crawlers, and some do. But that means no search visibility at all…

…All of Cloudflare’s new customers will now be asked if they want to allow AI companies’ bots to scrape their site, and for what purpose. Cloudflare’s scale gives it a better chance than most of enabling something like a collective response by content sites that want to force AI firms to cough up. It is testing a pay-as-you-crawl system that would let sites charge bots an entry fee…

…An alternative is offered by Tollbit, which bills itself as a paywall for bots. It allows content sites to charge AI crawlers varying rates: for instance, a magazine could charge more for new stories than old ones. In the first quarter of this year Tollbit processed 15m micro-transactions of this sort, for 2,000 content producers including the Associated Press and Newsweek…

…One of Tollbit’s highest per-crawl rates is charged by a local newspaper.

Another model is being put forward by ProRata, a startup led by Bill Gross, a pioneer in the 1990s of the pay-as-you-click online ads that have powered much of the web ever since. He proposes that money from ads placed alongside AI-generated answers should be redistributed to sites in proportion to how much their content contributed to the answer. ProRata has its own answer engine, Gist.ai, which shares ad revenue with its 500-plus partners, which include the Financial Times and the Atlantic…

…As for the idea that Google is disseminating less human traffic than before, Mr Stein says the company has not noticed a dramatic decline in the number of outbound clicks, though it declines to make the number public. There are other reasons besides AI why people may be visiting sites less. Maybe they are scrolling social media. Maybe they are listening to podcasts.

---

Disclaimer: The Good Investors is the personal investing blog of two simple guys who are passionate about educating Singaporeans about stock market investing. By using this Site, you specifically agree that none of the information provided constitutes financial, investment, or other professional advice. It is only intended to provide education. Speak with a professional before making important decisions about your money, your professional life, or even your personal life. We currently have a vested interest in Alphabet (the company behind AlphaGo Zero and Google). Holdings are subject to change at any time.

We’ve constantly been sharing a list of our recent reads in our weekly emails for The Good Investors.

Do subscribe for our weekly updates through the orange box in the blog (it’s on the side if you’re using a computer, and all the way at the bottom if you’re using mobile) – it’s free!

But since our readership-audience for The Good Investors is wider than our subscriber base, we think sharing the reading list regularly on the blog itself can benefit even more people. The articles we share touch on a wide range of topics, including investing, business, and the world in general. 

Here are the articles for the week ending 13 July 2025:

1. Jim Chanos on the Nuttiness of ‘Bitcoin Treasury Companies’ | Odd Lots (Transcript Here) – Tracy Alloway, Joe Weisenthal, and Jim Chanos

Joe: All right, first question: Are Bitcoin treasury companies the stupidest thing you’ve ever seen in your entire life?

Jim Chanos: It’s rarely, rarely that I have to increase my personal security after a podcast which I had to do after our last podcast together when I said some intemperate things about Bitcoin treasury companies.

Here’s the thing. I get people very agitated about this and they point out just what a genius idea this is and I keep trying to point out to them I’m doing the same thing that guys like Michael Saylor are doing. I’m on the same side of the trade and I keep pointing out to my critics, “You’re on the opposite side of that trade and you don’t want to be on the opposite side of the trade, and the Bitcoin treasury paradox being that you are the one buying the pieces of paper that have infinite supply so that Michael Saylor and I can buy the digital asset with the limited supply and it makes kind of no sense.” So what will inevitably happen is happening, in that there’s nothing proprietary here – this is just simply raising capital to buy a financial asset and other companies will do this. In fact even since the podcast we last did, I think the number of companies that have announced this strategy is scores more. I think there’s over a hundred in the US and over 200 globally now…

…Jim: Because there’s a wonderful sales job that’s being done about the fact that this is an economic engine in and of itself, therefore terms like Bitcoin Yield are are used and I’ve called them financial gibberish – because they are. In fact, this will get arbed away ultimately by companies that will do this to try to capture that spread. In the case of Micro Strategy, it’s substantial. It’s still $50 billion, something like that, of the difference between the value of the enterprise value of the company and the value of their Bitcoin holdings. But the thing that really shot me into orbit on all this was when Saylor and others then said, “You can’t really value us on an NAV basis, a so-called MNAV, multiple of NAV. You actually have to also give us additional value for the amount of profit that we make every quarter from the appreciation in the asset.” I said, “Well that’s like saying my whole net worth is in a house that’s worth $400,000 that is now worth $500,000 a year or two later, and my net worth is not $500,000 now – it’s $2.5 million because it’s the value of the house plus a multiple on the increase in the profitability of the asset.”…

…Tracy: I have one more question why did Micro – I have to remember to call them Strategy but I can’t bring myself to do it. Why did they switch from issuing the convertible debt to preferred shares?

Jim: Because he realized that as he began to issue more and more common, it was putting pressure on the premium. Now the latest iteration is, “We’re going to do this quasi equity security, quasi debt, preferred stock and then we can lever up the the balance sheet.” This is a company whose selling point a year ago was “We’re not going to lever, because we have this wonderful equity that we can issue at a premium.” Now they’re saying, “Maybe if it trades above 2x we’ll issue equity, but if it’s between 1x and 2x, we’ll do preferred, and then if it’s below 1x we’ll buy back common and then what is Chanos going to do?” To which I said, “I’ll be out of the trade by then.” If it’s 1x NAV it’s not a trade. That’s the latest game plan – but stay tuned, it’ll change, I think. The narrative keeps changing…

…Jim: The legacy data centers – and there’s only a couple companies in the United States that really have legacy data centers. There’s Equinix, there’s Digital Realty, and then there’s old Colony Capital – it’s now called Digital Bridge and they own these things in fund format.

When we took a look at this with our partner back in ‘22 the idea was pretty simple. We did not see the AI explosion in mid-’22, but the idea was it was a pretty crummy business then, working on the cloud and SaaS demand. But it became a really bad business with the advent of AI because it just moved the hyperscalers to invest more in state-of-the-art data centers. These are older data centers that we’re short, the idea being that the new GPU-centric data centers need liquid cooling – they basically need all the infrastructure ripped out and replaced – and the business was not a high return on capital business before this. It’s getting even worse now.

What Equinix said yesterday at their Analyst Day was that revenues were not going to quite be what people thought they would be, but more ominously, capex was going to keep increasing. That’s what we’ve been saying, that these are not like warehouses where you just collect a check. These are actually operating businesses where you have to service the servers, you have to make sure there’s redundancy. It’s a business, a tech business, and they’re traded as REITs and that was the opportunity. That was the dichotomy in valuation. People added back the depreciation as they do with REITs and they valued them on a so-called FFO or AFFO, which is a cash flow metric. But in fact, unlike warehouses, shopping centers to a lesser extent, office buildings, the capex was real. Depreciation was a real expense. To give you an example, with Equinix yesterday, they said “Our capex is now going to bump up to between $4 billion and $5 billion a year.” The problem is their EBITDA this year is expected to be $4.5 billion, so all of that’s going to go to capex, meaning they’re going to have to basically borrow or issue equity to pay their interest and dividends. That’s just a definition of a bad business and it’s a business that’s not growing very fast. Unlike other really true AI companies which are growing 25%, 30%, 40% a year, these guys are growing 3%, 5%, 6% sort of with GDP. So there’s no growing their way out of this. So they’re just really bad businesses trading at just nosebleed valuations.

Tracy: On the topic of idiosyncratic opportunities I got to ask about Carvana because when my husband and I moved back to the States in 2022, we bought a used car through Carvana and that was a mistake. It took us about 6 months to actually get the car and they lost all our paperwork and it was just an absolute nightmare. I thought at the time this is a company whose entire business model is basically built on regulation, that’s what they’re doing and I thought they’re not going to have a future if they are this bad at it. Yet the stock is up.

Joe: It’s done insanely well.

Jim: It’s done a double round trip. It crashed 99% and now it’s up 100x, so it’s pretty interesting again. The reason it’s interesting is that if you go through the numbers, they are making more than 100% of their pre-tax profit from gain on sale of subprime loans and gain on sale of equity stakes in other companies. You ex those two out, they’re losing money and they’re losing money now right after the rebound, after the restructuring from 2022-2023. This is a company that is being valued again as a secular growth stock that saw its used car revenues drop 30% between 2022 and 2023, so it’s not necessarily a secular growth company. The accounting is abysmal. What people are really missing is that what’s happening in subprime auto securizations right now – and you can track it on your Bloomberg terminal – delinquencies are starting to skyrocket.

Tracy: We actually did an episode on this recently with Jim Egan.

Jim: So a huge amount of their profits comes from generating paper from customers and then selling it into the open market or to affiliates. This is a company that was spun out of a company called Drive Time Finance, which is their affiliated finance company which was originally called Ugly Duckling in the late ‘90s which was run by the current CEO’s father. That company collapsed in the first subprime blowup which was not the GFC – it was actually in the late ‘90s in subprime auto credit and consumer loans. It didn’t go bankrupt but it came close. He had to restructure it. He bought it in private and then restructured it, renamed it Drive Time Finance. But that’s the genesis of Carvana. That’s its DNA. It’s basically a subprime finance lead company, if you will. Those companies should not trade at 40x and 50x expected earnings – and they don’t by and large. They’re consumer finance companies. So it’s an odd bird. It’s still heavily leveraged, the stock is up a ton.

But what really got us interested again recently was the vast amount of insider selling that has just started in May and June in the company. If you go look at the insider selling in the company, it is just now a torrent of everybody selling pretty much every day. We just don’t think that’s a good sign given what’s happening in the subprime securization market…

…Jim: Every once in a while. There’s one other thing though I do want to mention. I was talking to someone earlier today and I think one of the things that’s underappreciated by investors right now and one of the things that’s been most interesting to me is how corporate profit margins have held up, which used to be very mean-reverting as you know. The more work we’ve done on this, the more we’re kind of convinced that the capital spending boom we’re seeing due to tech and specifically AI, is is looking very much akin to the global internet buildout networking buildout in the late ‘90s and the problem there of course is that if you buy my chips from NVIDIA or you were buying my networking equipment at Cisco and Lucent, that’s revenue for me and profit. But for you it’s a capitalized expense, it’s written off over time, and that adds a big, big boost until people pull their orders. That’s what we saw in 2001, 2002 that GDP dropped about 1% to 2% in the recession of ‘01-’02. Does anybody know what corporate profits did in that? That was an investment-driven recession. Consumers didn’t feel it at all. Earnings were down about 45% I think from peak to trough in the S&P. They were down about the same, a little bit more in the global financial crisis, but of course GDP collapsed.

Here’s a little interesting thought experiment. Right now NVIDIA’s revenues are about one-half of 1% of US GDP, about $140 billion and our GDP is about $29 trillion. Anyone tell me what Cisco and Lucent – the two companies that you needed when building out your internet network in ‘99, 2000 – did anybody know what their combined revenues as a percent of GDP was in 2000?

Tracy: No using your phones.

Joe: And ChatGPT.

Jim: It was a half a percent. It was roughly $50 billion total on GDP of $10 trillion. So those revenues stopped growing at some point shortly thereafter and actually shrunk a little bit. The investment boom we’re seeing right now, we’ve seen before. And it’s not just chips. It’s Caterpillar, it’s people building the data centers, it’s people building new utilities. There is an ecosystem around the AI boom that is considerable, as there was for TMT back in ‘99 and 2000. But it is a riskier revenue stream because if people pull back, they can pull back capex very easily, projects can get put on hold for six months or nine months, and that immediately shows up in disappointing revenues and earnings forecast if it happens. We’re not there yet but that’s one of the risks out there that I think a lot of people are underestimating.

2. Creating therapeutic abundance – Jacob Kimmel

Jack Scannell infamously predicted in 2012 that the number of drugs per billion dollars would decline two-fold every nine years. Unfortunately, our therapeutics industry has largely followed through…

…Drug program success rates are equally complex. Failures can be attributed to safety issues, failure of a drug to hit the desired biological target, or improper selection of the target for a given disease…

…We can bucket the failures into a two broad categories of safety and efficacy and make informed estimates.

1. Safety failures – ~20-30% of all candidates
A molecule was developed, but proved unsafe in patients. These are typically detected as failures in Phase 1 trials.

2. Efficacy failures – 70-80% of all candidates
The remainder of all drug candidates that fail – 63% of all drugs placed into trials period – fail due to a lack of efficacy. Even though the drugs are safe, they don’t provide benefit to the patients by treating their disease.

From these coarse numbers, it’s clear that the highest leverage point in our drug development process is increasing the efficacy rate of new candidate medicines…

…Efficacy failures can broadly occur for two reasons:

1. Engagement failures: We chose the right biology (“target”) to manipulate, but our drug candidate failed to achieve the desired manipulation. This is the closest thing drug development has to an engineering problem.
2. Target failures: The drug candidate manipulated our chosen biology exactly as expected. Unfortunately, the target failed to have the desired effect on the disease. This is a scientific or epistemic failure, rather than an engineering problem. We simply failed to understand the biology well enough to intervene and benefit patients.

It’s difficult to know exactly the exact frequency of these two failure modes, but we can infer from a few sources that target failures dominate.

• Success rates for biosimilar drugs hitting known targets are extremely high, >80%
• Drugs against targets with genetic evidence have a 2-3 fold higher success rate than those against targets lacking this evidence, suggesting that picking good targets is a high source of leverage
• Among organizations with meaningful internal data, picking the right target is considered the first priority of all programs (e.g. “Right target” is the first tenet of AstraZeneca’s “5Rs” framework).

The predominance of target failures has likewise led most companies working on new modalities to address a small set of targets with well-validated biology. This has led to dozens of potential medicines “crowding” on the same targets, and this trend is increasing over time…

…If searching for targets is the limiting reagent in our medicine production function, the difficulty of finding targets must increase over time in order to explain part of Eroom’s law. How could this be the case given all the improvements in underlying biomedical science?

In an influential paper “Are ideas getting harder to find?”, Nicholas Bloom and colleagues argue that many fields of invention suffer from diminishing returns to investment. Intuitively, the low hanging fruit in a given discipline is picked early and more investment is required merely to reap the same harvest from higher branches on the tree of ideas…

…Targets are getting harder to find not because we are getting worse at selection, but because many of the easy and obvious therapeutic hypotheses have already been exploited….

…While promising, human genetics can only reveal a certain class of targets. The larger the effect size of a genetic variant, the less frequently it appears in the population due to selective pressure. In effect, this means that the largest effects in biology are the least likely to be discovered using human genetics. Many of the best known targets have minimal genetic signal for this reason.

Our current methods are good at discovering individual genes that associate with health, but discovering combinations of genes is nascent at best. Human genetics cannot help us discover the combinatorial medicines or gene circuits to install in a cell therapy…

…Even with the best possible experimental methods, some of the most promising target biologies will never be searched exhaustively. There are a nearly infinite number of combinatorial genetic interventions we might drug, synthetic circuits we might engineer into cells, and changes in tissue composition we might engender.

Artificial intelligence models can learn general models from the data generated in functional genomics experiments of many flavors, predicting outcomes for the experiments we haven’t yet run. If we manage to construct a performant model for a given class of target biologies, we may be able to increase the efficiency of target discovery by many orders-of-magnitude. The cost of discovering a target could conceivably go from >$1B to <$1M.

There’s growing interest in the idea of combining these technologies to build “virtual cells,” models that can predict the outcomes of target discovery experiments in silico before they’re ever executed in the lab. The grand version of this vision spans all possible target biologies, from gene inhibitions to polypharmaceutical small molecule treatments. In the maximal form, it may take many years to realize.

More limited realizations though are tractable today. The initial versions of these models are already emerging within early Predictive Biology companies. As a few examples, Recursion is building models of genetic perturbations in cancer cells, Tahoe Tx is building models in oncology with a chemical biology approach, and NewLimit has developed models for reprogramming cell age across human cell types13. Focused models like these represent an early demonstration that this general approach can yield therapeutic value…

…We are entering an epoch of abundant intelligence. With these tools, we have the opportunity to discover & design target biologies at a rate that’s too cheap to meter. The therapies that emerge could serve as the counterexample that downgrades Eroom’s law to a historic conjecture.

3. What I learned watching 78 videos from Tesla’s Austin robotaxis – Timothy B. Lee

I’ve watched 78 videos posted by pro-Tesla influencers who got early access to the service. Those videos documented more than 16 hours of driving time across nearly 100 rides.

These videos exceeded my expectations. Tesla’s robotaxi rollout wasn’t perfect, but it went as well as anyone could have expected. A handful of minor glitches got outsized attention online, but a large majority of trips were completed without incident…

…Tesla’s robotaxis drove flawlessly during the vast majority of the 16 hours of driving footage I watched. They stayed in their lane, followed traffic laws, and interacted smoothly with other vehicles…

…Tesla’s most widely discussed error occurred around seven minutes into this video. The robotaxi approached an intersection and got into the left turn lane. But the robotaxi couldn’t make up its mind whether it wanted to turn left or go straight. The car’s steering wheel jerked back and forth several times. On the car’s display, the blue ribbon showing the car’s intended path jumped back and forth erratically between turning left or continuing straight. Finally, the Tesla decided to proceed straight but ended up driving the wrong way in the opposite left turn lane…

…But in a piece last year, I argued that they were misunderstanding the situation.

“Tesla hasn’t started driverless testing because its software isn’t ready,” I wrote. “For now, geographic restrictions and remote assistance aren’t needed because there’s always a human being behind the wheel. But I predict that when Tesla begins its driverless transition, it will realize that safety requires a Waymo-style incremental rollout.”

That’s exactly what’s happened:

• Just as Waymo launched its fully driverless service in 50 square miles near Phoenix in 2020, so Tesla launched its robotaxi service in about 30 square miles of Austin last month.
• Across 16 hours of driving, I never saw Tesla’s robotaxi drive on a freeway or go faster than 43 miles per hour. Waymo’s maximum speed is currently 50 miles per hour.
• Tesla has built a teleoperation capability for its robotaxis. One job posting last year advertised for an engineer to develop this capability. It stated that “our remote operators are transported into the device’s world using a state-of-the-art VR rig that allows them to remotely perform complex and intricate tasks.”

The launch of Tesla’s robotaxi service in Austin is a major step toward full autonomy. But the Austin launch also makes it clear that Tesla hasn’t discovered an alternative path for testing and deploying driverless vehicles. Instead, Tesla is following the same basic deployment strategy Waymo pioneered five to seven years ago.

Of course, this does not necessarily mean that Tesla will scale up its service as slowly as Waymo has. It took almost five years for Waymo to expand from its first commercial service (Phoenix in 2018) to its second (San Francisco in 2023). The best informed Tesla bulls acknowledge that Waymo is currently in the lead but believe Tesla is positioned to expand much faster than Waymo did…

…Last month, Waymo published a study demonstrating that self-driving software benefits from the same kind of “scaling laws” that have driven progress in large language models.

“Model performance improves as a power-law function of the total compute budget,” the Waymo researchers wrote. “As the training compute budget grows, optimal scaling requires increasing the model size 1.5x as fast as the dataset size.”

When Waymo published this study, Tesla fans immediately seized on it as a vindication of Tesla’s strategy. Waymo trained its experimental models using 500,000 miles of driving data harvested from Waymo safety drivers driving Waymo vehicles. That’s a lot of data by most standards, but it’s far less than the data Tesla could potentially harvest from its fleet of customer-owned vehicles…

…I posed this question to Dragomir Anguelov, the head of Waymo’s AI foundations team and a co-author of Waymo’s new scaling paper. He argued that the paper’s implications are more complicated than Tesla fans think.

“We are not driving a data center on wheels and you don’t have all the time in the world to think,” Anguelov told me in a Monday interview. “Under these fairly important constraints, how much you can scale and what are the optimal ways of scaling is limited.”

Anguelov also pointed to an issue that will be familiar to anyone who read last month’s explainer on reinforcement learning.

Waymo’s scaling paper—like OpenAI’s famous 2020 scaling law paper—focused on models trained with imitation learning…

…Anguelov was a co-author of a 2022 Waymo paper finding that self-driving models trained with a combination of imitation and reinforcement learning tend to perform better than models trained only with imitation learning.

Imitation learning is “not the most sophisticated thing you can do,” Anguelov told me. “Imitation learning has a lot of limitations.”

This is significant because demonstration data from human drivers—the kind of data Tesla has in abundance—isn’t very helpful for reinforcement learning. Reinforcement learning works by having a model try to solve a task and then judging whether it succeeded. For self-driving, this can mean having a model “drive” in simulation and then judging whether it caused a collision or other problems. Or it can mean running the software on real cars and having a safety driver intervene if the model makes a mistake. In either case, it’s not obvious that having vast amounts of human driving data is especially helpful.

One finding from that 2022 paper is particularly relevant for thinking about the performance of Tesla’s robotaxis. The Waymo researchers noted that models trained only with imitation learning tend to drive well in common situations but make mistakes in “more unusual or dangerous situations that occur only rarely in the data.”

In other words, if you rely too much on imitation learning, you can end up with a model that drives like an expert human most of the time but occasionally makes catastrophic mistakes…

…Since its 2018 launch, Waymo has acknowledged that it has remote operators who sometimes provide real-time assistance to its vehicles. But Waymo has also said that these remote operators never drive the vehicles in real time. Instead, they provide high-level feedback, while the vehicle always remains in control of second-by-second decisions.

In contrast, Tesla’s job posting stated that teleoperators can be “transported into the device’s world” so that they can “remotely perform complex and intricate tasks.” Could those “complex and intricate tasks” include driving the car for seconds or even minutes at a time?

In the videos I watched, a number of Tesla’s early customers commented on how human-like Tesla’s driving was. That might just be a tribute to the quality of Tesla’s AI model. But it’s also possible that sometimes a human driver is literally driving the vehicle from a remote location.

4. No Bad Risks, Only Bad Rates — And Other Lessons From National Indemnity Founder Jack Ringwalt – Kingswell

There are no bad risks in insurance — only bad rates

This maxim was Ringwalt’s north star, the iron-clad principle that allowed him to fearlessly pursue unusual and unwanted risks without driving himself right out of business. Almost anything can be intelligently insured, so long as you charge enough for the coverage.

(It’s also reminiscent of one of my favorite Warren Buffett lines. “I can go into an emergency ward and write life insurance,” he said in 1990, “if you let me charge enough of a premium.”)

When evaluating potential opportunities, Ringwalt’s open mind welcomed the weird and the wild — and he wrote many policies on offbeat ventures that others wouldn’t touch with a ten-foot pole. But, when it came to pricing, that flexibility vanished. If the market would not meet his rate, Ringwalt never blinked. He just waved goodbye to the deal with an indifferent shrug.

“When business is unprofitable to the companies in general,” wrote Ringwalt, “our premium volume has taken a very sharp spurt and when business has been profitable for most companies, we have run into very unintelligent competition and have had to cut down temporarily on our writings.”

The insurance merry-go-round is always the same: profitability lures rivals who slash rates to grab market share, only to crater when losses inevitably pile up. And when the industry bleeds, fly-by-night competitors vanish, prices climb back to normal, and the cycle starts spinning anew. “This pattern will keep repeating,” he wrote. “It makes no sense, but it’s human nature.”

Ringwalt steadfastly refused to play that sucker’s game — a tradition that continued under Berkshire’s aegis. From 1986 to 1999, National Indemnity’s revenue nosedived 85% as profitable premiums evaporated. But, rather than succumb to the pressure to write more business at any price, Buffett and co. urged employees to wait patiently for the right pitch (so to speak). Some things never change.

5. Why I don’t think AGI is right around the corner – Dwarkesh Patel

Sometimes people say that even if all AI progress totally stopped, the systems of today would still be far more economically transformative than the internet. I disagree. I think the LLMs of today are magical. But the reason that the Fortune 500 aren’t using them to transform their workflows isn’t because the management is too stodgy. Rather, I think it’s genuinely hard to get normal humanlike labor out of LLMs. And this has to do with some fundamental capabilities these models lack…

…But the fundamental problem is that LLMs don’t get better over time the way a human would. The lack of continual learning is a huge huge problem. The LLM baseline at many tasks might be higher than an average human’s. But there’s no way to give a model high level feedback. You’re stuck with the abilities you get out of the box. You can keep messing around with the system prompt. In practice this just doesn’t produce anything even close to the kind of learning and improvement that human employees experience.

The reason humans are so useful is not mainly their raw intelligence. It’s their ability to build up context, interrogate their own failures, and pick up small improvements and efficiencies as they practice a task.

How do you teach a kid to play a saxophone? You have her try to blow into one, listen to how it sounds, and adjust. Now imagine teaching saxophone this way instead: A student takes one attempt. The moment they make a mistake, you send them away and write detailed instructions about what went wrong. The next student reads your notes and tries to play Charlie Parker cold. When they fail, you refine the instructions for the next student.

This just wouldn’t work. No matter how well honed your prompt is, no kid is just going to learn how to play saxophone from just reading your instructions. But this is the only modality we as users have to ‘teach’ LLMs anything…

…When we do solve continuous learning, we’ll see a huge discontinuity in the value of the models. Even if there isn’t a software only singularity (with models rapidly building smarter and smarter successor systems), we might still see something that looks like a broadly deployed intelligence explosion. AIs will be getting broadly deployed through the economy, doing different jobs and learning while doing them in the way humans can. But unlike humans, these models can amalgamate their learnings across all their copies. So one AI is basically learning how to do every single job in the world. An AI that is capable of online learning might functionally become a superintelligence quite rapidly without any further algorithmic progrss…

…But here are the timelines where I’d take a 50/50 bet:

• AI can do taxes end-to-end for my small business as well as a competent general manager could in a week: including chasing down all the receipts on different websites, finding all the missing pieces, emailing back and forth with anyone we need to hassle for invoices, filling out the form, and sending it to the IRS: 2028
  I think we’re in the GPT 2 era for computer use. But we have no pretraining corpus, and the models are optimizing for a much sparser reward over a much longer time horizon using action primitives they’re unfamiliar with. That being said, the base model is decently smart and might have a good prior over computer use tasks, plus there’s a lot more compute and AI researchers in the world, so it might even out. Preparing taxes for a small business feels like for computer use what GPT 4 was for language. It took 4 years to get from GPT 2 to GPT 4. Just to clarify, I am not saying that we won’t have really cool computer use demos in 2026 and 2027 (GPT-3 was super cool, but not that practically useful). I’m saying that these models won’t be capable of end-to-end handling a week long and quite involved project which involves computer use.
• AI learns on the job as easily, organically, seamlessly, and quickly as a human, for any white collar work. For example, if I hire an AI video editor, after six months, it has as much actionable, deep understanding of my preferences, our channel, what works for the audience, etc as a human would: 2032
  While I don’t see an obvious way to slot in continuous online learning into current models, 7 years is a long time! GPT 1 had just come out this time 7 years ago. It doesn’t seem implausible to me that over the next 7 years, we’ll find some way for models to learn on the job.

You might react, “Wait you made this huge fuss about continual learning being such a handicap. But then your timeline is that we’re 7 years away from what would at minimum be a broadly deployed intelligence explosion.” And yeah, you’re right. I’m forecasting a pretty wild world within a relatively short amount of time.

AGI timelines are very lognormal. It’s either this decade or bust. (Not really bust, more like lower marginal probability per year – but that’s less catchy).AI progress over the last decade has been driven by scaling training compute of frontier systems (over 4x a year). This cannot continue beyond this decade, whether you look at chips, power, even fraction of raw GDP used on training. After 2030, AI progress has to mostly come from algorithmic progress. But even there the low hanging fruit will be plucked (at least under the deep learning paradigm). So the yearly probability of AGI craters.

---

Disclaimer: The Good Investors is the personal investing blog of two simple guys who are passionate about educating Singaporeans about stock market investing. By using this Site, you specifically agree that none of the information provided constitutes financial, investment, or other professional advice. It is only intended to provide education. Speak with a professional before making important decisions about your money, your professional life, or even your personal life. We currently have a vested interest in Alphabet (the company behind Waymo), and Tesla. Holdings are subject to change at any time.

We’ve constantly been sharing a list of our recent reads in our weekly emails for The Good Investors.

Do subscribe for our weekly updates through the orange box in the blog (it’s on the side if you’re using a computer, and all the way at the bottom if you’re using mobile) – it’s free!

But since our readership-audience for The Good Investors is wider than our subscriber base, we think sharing the reading list regularly on the blog itself can benefit even more people. The articles we share touch on a wide range of topics, including investing, business, and the world in general. 

Here are the articles for the week ending 06 July 2025:

1. Etched is Making the Biggest Bet in AI – Etched team

We’ve spent the past two years building Sohu, the world’s first specialized chip (ASIC) for transformers (the “T” in ChatGPT).

By burning the transformer architecture into our chip, we can’t run most traditional AI models: the DLRMs powering Instagram ads, protein-folding models like AlphaFold 2, or older image models like Stable Diffusion 2. We can’t run CNNs, RNNs, or LSTMs either.

But for transformers, Sohu is the fastest chip of all time. It’s not even close.

With over 500,000 tokens per second in Llama 70B throughput, Sohu lets you build products impossible on GPUs. Sohu is an order of magnitude faster and cheaper than even NVIDIA’s next-generation Blackwell (B200) GPUs…

…No one has ever built an algorithm-specific AI chip (ASIC). Chip projects cost $50-100M and take years to bring to production. When we started, there was no market.

Suddenly, that’s changed:

• Unprecedented Demand: Before ChatGPT, the market for transformer inference was ~$50M, and now it’s billions. All big tech companies use transformer models (OpenAI, Google, Amazon, Microsoft, Facebook, etc.).
• Convergence on Architecture: AI models used to change a lot. But since GPT-2, state-of-the-art model architectures have remained nearly identical! OpenAI’s GPT-family, Google’s PaLM, Facebook’s LLaMa, and even Tesla FSD are all transformers.

When models cost $1B+ to train and $10B+ for inference, specialized chips are inevitable. At this scale, a 1% improvement would justify a $50-100M custom chip project.

In reality, ASICs are orders of magnitude faster than GPUs. When bitcoin miners hit the market in 2014, it became cheaper to throw out GPUs than to use them to mine bitcoin…

…We believe in the hardware lottery: the models that win are the ones that can run the fastest and cheapest on hardware. Transformers are powerful, useful, and profitable enough to dominate every major AI compute market before alternatives are ready:

• Transformers power every large AI product: from agents to search to chat. AI labs have spent hundreds of millions of dollars in R&D to optimize GPUs for transformers. The current and next-generation state-of-the-art models are transformers.
• As models scale from $1B to $10B to $100B training runs in the next few years, the risk of testing new architectures skyrockets. Instead of re-testing scaling laws and performance, time is better spent building features on top of transformers, such as multi-token prediction.
• Today’s software stack is optimized for transformers. Every popular library (TensorRT-LLM, vLLM, Huggingface TGI, etc.) has special kernels for running transformer models on GPUs. Many features built on top of transformers aren’t easily supported in alternatives (ex. speculative decoding, tree search).
• Tomorrow’s hardware stack will be optimized for transformers. NVIDIA’s GB200s have special support for transformers (TransformerEngine). ASICs like Sohu entering the market mark the point of no return. Transformer killers will need to run on GPUs faster than transformers run on Sohu. If that happens, we’ll build an ASIC for that too!…

…Isn’t inference bottlenecked on memory bandwidth, not compute?

Actually, for modern models like Llama-3, no!

Let’s use NVIDIA and AMD’s standard benchmark13: 2048 input tokens and 128 output tokens. Most AI products have much longer prompts than completions (even a new Claude chat has 1,000+ tokens in the system prompt).

On GPUs and on Sohu, inference is run in batches. Each batch loads all of the model weights once, and re-uses them across every token in the batch. Generally, LLM inputs are compute-bound, and LLM outputs are memory-bound. When we combine input and output tokens with continuous batching, the workload becomes very compute bound…

…We can scale up the same trick to run Llama-3-70B with 2048 input tokens and 128 output tokens. Have each batch consist of 2048 input tokens for one sequence, and 127 output tokens for 127 different sequences.

If we do this, each batch will require about (2048 + 127) × 70B params × 2 bytes per param = 304 TFLOPs, while only needing to load 70B params × 2 bytes per param = 140 GB of model weights and about 127 × 64 × 8 × 128 × (2048 + 127) × 2 × 2 = 72GB of KV cache weights. That’s far more compute than memory bandwidth: an H200 would need 6.8 PFLOPS of compute in order to max out its memory bandwidth. And that’s at 100% utilization – if utilization was 30%, you’d need 3x more.

Since Sohu has so much compute with very high utilization, we can run enormous throughputs without bottlenecking on memory bandwidth…

…On GPUs and TPUs, software is a nightmare. Handling arbitrary CUDA and PyTorch code requires an incredibly complicated compiler. Third-party AI chips (AMD, Intel, AWS, etc.) have together spent billions on software to little avail.

But since Sohu only runs transformers, we only need to write software for transformers!

Most companies running open-source or internal models use a transformer-specific inference library like TensorRT-LLM, vLLM, or HuggingFace’s TGI. These frameworks are very rigid – while you can tweak model hyperparameters, changing the underlying model code is not really supported. But this is fine – since all transformer models are so similar (even text/image/video ones), tweaking the hyperparameters is all you really need.

2. Lots More on What’s Going On in Iran’s Markets (Transcript Here) – Tracy Alloway, Joe Weisenthal, and Maciej Wojtal

Maciej: If I can just comment on one thing, because the way you introduced Iran is the perfect way to show the country. It’s the size of Turkey in terms of population and actually geographical size as well. But if you compare the economy of Turkey and Iran, it’s around five times smaller. So Iran is around five times smaller and if you look at the composition of the economy, Turkey has no natural resources, so they have to import the whole energy commodities they consume. So Iran has a similar size of potential non-commodity GDP that it could grow to, from the current let’s say $250 billion to $1.1 trillion GDP that Turkey has. But also on top of this, has resources that are actually – if you combine gas and oil, they are bigger than Saudi Arabia’s and Saudi Arabia is another one, I think $1.3 trillion economy. This is a good way to just frame Iran as to show Iran, as it’s a big country that should really be having much bigger economy. Because of sanctions, various reasons and so on, it’s been underdeveloped. But the scale of this underdevelopment is like 10x.

Tracy: And because of the sanctions we can’t actually go and look up what’s happening in Tehran’s stock market. So why don’t you give us an overview of what it’s been like for the past week given geopolitical events?

Maciej: So for the past week it was difficult for everyone to check what was going on in Iran because internet was shut down basically. I could communicate with my team on the ground in Tehran once a day when they had signal and sometimes it was WhatsApp that was working, sometimes Telegram. But it was maybe once or twice per day. So what was going on in the market was simply nothing. The stock market hasn’t opened. The exchange of fire between Iran and Israel happened on a Friday, which is weekend in Iran, and then on the following Saturday there was an important religious holiday, so the market and actually the whole economy was supposed to be closed anyway. The economic activity, the market, was supposed to resume on a Sunday but they didn’t open. So the stock market, pretty much most of the currency market, has been closed for the last two weeks…

…Maciej: For example right now, when you have the stock market, the currency market were shut down, but you could track what’s going on with the exchange rate of the Iranian rial versus dollar either on Telegram chats but also on cryptocurrency exchanges. You have liquid market on stable coins versus Iranian rial inside of Iran where liquidity was limited during the last period anyway, but we could see the changes. So we knew that $1 before the war was at around 830,000 rials, then it went up roughly 15% to 950,000 and now after the ceasefire, it’s back down at 850,000. You can track the market, you can actually make transactions depending on the vol, on the liquidity, but it is possible. To be honest, when I saw those exchange rates moves 15% when you have a war where a lot of commentators were saying that this could turn into a massive worldwide conflict, that 15% in a country like Iran I would say that this is your usual volatility on the currency market…

…Maciej: In Tehran, a lot of residents were just relocating out of Tehran. Tehran is a big city, 12 million people, and they were moving mainly north to some smaller cities by the Caspian Sea. You had massive congestion. People were spending hours in traffic jams trying to get out of Tehran. There was not enough petrol on gas stations just because of this peak in demand. You had some petrol rationing.

Then I was asking them is the economy working, not working? Everything that was non-essential basically was closed. So you couldn’t build, buying materials or anything like this. But groceries, pharmaceuticals, gas stations, banks, this was all open and working properly with some disruptions. But those disruptions, for example, if you wanted to buy groceries in north of Iran where everyone has just relocated, you had some logistical bottlenecks. Distribution was not fast enough, so you had some shortages just for a little while. With banks, some branches were not operating at 100% capacity. Two banks got hacked. You had some cyber attacks on two banks in Iran and one cryptocurrency exchange. The rest of the banking sector was working without any disruptions. You could get cash from any ATM. There were no problems like these…

…Maciej: It’s interesting because there is information, very up-to-date detailed information on Iranian stocks available in Iran. But the majority of this information is not accessible if you’re trying to access it from a computer with your IP address outside of Iran. A lot of this information is restricted to Iran IP only. You cannot find anywhere on the whole internet. There is no website that shows the stock market index in dollars. When we send it out to our investors or just people who want to read news about the stock market in Iran, we are the only source of this information. This is quite amazing. It’s a country of 90 million people and stock market with 700 companies and there is no single place in internet that would show you the only important index…

…Maciej: In terms of oil, it is not really publicly traded. There is one Iranian monopoly called National Iranian Oil Corporation or company that is responsible for production. I think this is all centralized in one company and this is held by the government so it’s not publicly listed. You have some exposure to oil through oil refineries that are listed but refineries, they’re not sensitive to the price of oil. They are sensitive to the crack spread, which defines their refining margin, so they are not really a proxy to oil prices.

The whole stock market actually is well diversified. You have  large sectors such as chemicals – mainly these are like petrochemicals – companies that produce different products, different versions, use natural gas that is in in large supply as a cheap commodity and produce fertilizers or products like this. This is probably 20% of the stock market.

Then you have steel companies. The largest steel company in the Middle East is in Iran. You have car makers that produce more than 1 million cars a year. With car manufacturers, you have all the related industries, suppliers, to the car manufacturing businesses. You have banks – financials is an important sector – plus some consumer exposure, some building materials, cement companies are one of the best performers over the last few years actually…

…Maciej: I’ll get back to the potential for GDP in a moment. But the catalyst is absolutely clear. It must be the opening up of Iran as a country, and opening up of the economy, andthe US sanctions lifted. There must be an agreement between the US and Iran. What needs to happen? Some sort of political change. Political attitude must change on both sides. But to be honest, many analysts were expecting some big dramatic event that needs to happen in Iran for the country to properly open up.

When you look at Iran right now and you compare to let’s say even a few years ago when you had negotiations with the US, what were the biggest problem was, it was always about two things: (1) Iran enriching uranium too much basically, at a wrong level, and (2) Iranian regional policies, so financing proxies from Hezbollah to Hamas, Assad in Syria and so on. These two things were always the problem that they couldn’t negotiate over. When you look at it right now, to a large extent both obstacles are gone…

…Joe: Are there tech companies that trade on the Tehran stock market?

Maciej: There are tech companies. The ones that are listed are related to enterprise software, the Oracle or SAP, German SAP. But you have privately held companies that would like to IPO but they are just waiting for the approval from the regulator, and these are quite amazing companies. You have Snapp, which is like an Uber, but Snapp has more rides in Tehran than Uber in any city in the world. It’s a really world-class company. You have Digikala, which is like Amazon basically, also a large company, one of the biggest success stories.

3. Stablecoins might revolutionise payments, but what if they don’t? – Bryce Elder

That leaves payments:

While in a theoretical tokenized/blockchain based world, stablecoin-based payments would be faster, more efficient and interoperable, in practice at the moment these stablecoin based payments mostly start and finish with fiat, thus requiring on/off-ramps. This on/off ramp requirement adds significant friction/cost to the use of stablecoins for payments, making it less attractive compared to traditional financial systems, in particular if one takes into account the emergence of faster payment rails in the traditional financial system via fintech advancements in recent years. As a result, we find rather unrealistic the expectation of a massive increase in the use of stablecoins in payments. Indeed, our colleagues in US short-term rates research also note that market participants at the front end are skeptical of significant growth in the near term, in part due to the fact that the infrastructure/ecosystem for stablecoins remains underdeveloped. But even if one adopts an optimistic view and assumes, for example, a tenfold increase in the use of stablecoins in payments over the next couple of years, the stablecoin universe would only expand by $15bn x 10 = $150bn.

Stablecoin optimists point to the rapid adoption of the e-CNY, China’s central bank digital yuan, which has grown to a more than Rmb300bn market cap from Rmb13.6bn at the end of 2022. There’s no comparison, JPMorgan says:

First, the digital yuan is a central bank liability and thus it effectively replaces banknotes in circulation. While there does not appear to be a published target share of M0, there have been suggestions that a 10-15% share of M0 is a plausible medium-term goal, which would imply around RMB 1.3-2tr using current M0 levels. By contrast, stablecoins are a form of a tokenized MMF with zero interest, effectively a private sector liability rather than a central bank liability.

Second, the digital yuan does not operate through a fully decentralized blockchain-based ledger. Instead, it operates via a centralized network supervised by the PBoC and competes with other mobile/ electronic payment options in China such as Alipay and WeChat Pay.

Then is it better to think of stablecoins as global equivalents to Alipay and WeChat Pay? JPMorgan says no. Fintech payment companies offering collateralised electronic private money on their own platforms hasn’t proven the need for public blockchains; if anything, it proves the opposite:

Alipay/WeChat Pay digital money are private liabilities and are perhaps more similar to bank deposits in that regard which are also private liabilities. The difference between bank deposits and Alipay/WeChat balances is that the latter are backed by reserve funds that in turn hold public liabilities i.e. central bank reserves, while bank deposits are matched on the asset side by a mix of loans and debt securities, though they do have an additional guarantee via deposit protection arrangements.

In our mind, the strong expansion of Alipay and WeChat Pay should be viewed through the lens of a fintech payments revolution over the past decade in China that utilizes and increases the efficiency of traditional banking/financial system networks, rather than through the lens of a blockchain/crypto ecosystem revolution. In fact, it could be argued that the success and continued advancements in payments by fintechs, such as Alipay and WeChat Pay reduce the need for blockchain-based payment systems in the future.

4. Meet Project Rainier, Amazon’s one-of-a-kind machine ushering in the next generation of AI – Kirsteen Rodger

Project Rainier is designed as a massive “EC2 UltraCluster of Trainium2 UltraServers.” The first part refers to Amazon Elastic Compute Cloud (EC2), an AWS service that lets customers rent virtual computers in the cloud rather than buying and maintaining their own physical servers.

The more interesting bit is Trainium2, a custom-designed AWS computer chip built specifically for training AI systems. Unlike the general-purpose chips in your laptop or phone, Trainium2 is specialized for processing the enormous amounts of data required to teach AI models how to complete all manner of different and increasingly complex tasks—fast.

To put the power of Trainium2 in context: a single chip is capable of completing trillions of calculations a second. If, understandably, that’s a little hard to visualize: consider that it would take one person more than 31,700 years to count to one trillion. A task that would require millennia for a human to complete can be done in the blink of an eye with Trainium2…

…Traditionally, servers in a data center operate independently. If and when they need to share information, that data has to travel through external network switches. This introduces latency (i.e, delay), which is not ideal at such large scale.

AWS’s answer to this problem is the UltraServer. A new type of compute solution, an UltraServer combines four physical Trainium2 servers, each with 16 Trainium2 chips. They communicate via specialized high-speed connections called “NeuronLinks.” Identifiable by their distinctive blue cables, NeuronLinks are like dedicated express lanes, allowing data to move much faster within the system and significantly accelerating complex calculations across all 64 chips.

When you connect tens of thousands of these UltraServers and point them all at the same problem, you get Project Rainier—a mega “UltraCluster.”…

…Communication between components happens at two critical levels: the NeuronLinks provide high-bandwidth connections within UltraServers, while Elastic Fabric Adapter (EFA) networking technology (identified by its yellow cables) connects UltraServers inside and across data centers. This two-tier approach maximizes speed where it’s most needed while maintaining the flexibility to scale across multiple data center buildings.

5. OpenAI has started to form a “moat” – Rihard Jarc

I think anyone who follows the AI space knows about OpenAI and, more specifically, about ChatGPT. Even outside of investors and tech enthusiasts, the verb ChatGPT has gone viral, similar to how the verb Google started. What is even more surprising is that despite ChatGPT being out there for more than 2 years already, just recently, at the end of March, it came to another acceleration point in terms of adoption when the Ghibli photo trend emerged on ChatGPT:

The number of MAUs doubled from 400 million to 800 million in a matter of a few weeks. Looking at the adoption curves of other highly adopted technology platforms, such as TikTok, Facebook, Instagram; ChatGPT, is on a slope of its own.

Another factor to consider is that it is not just a “I must try it moment”. Looking at the number of minutes a user spends on ChatGPT, the minutes are constantly growing and have now reached the 29-minute daily mark.

Remember that at the start of ChatGPT and LLMs, many critics said that people tried it, had fun, and then didn’t use it again. This trend shows that that is not the case and that with each enhanced model version and UX improvement, the stickiness factor becomes bigger…

…OpenAI also now has serious hardware ambitions. In late May of this year, they acquired Jony Ive’s startup, a famous former Apple designer, for nearly $6.5 billion, who will now lead OpenAI’s hardware efforts. What is now almost a consensus opinion among big tech leaders is that AI will unlock the next computing platform, one that is not tied to the smartphone.

And if you listen to those conversations, everyone is calling for a similar device. A device that will be more like a companion system and will be less dependent on a screen. Proactive assistant who will run even when you don’t ask it.

---

Disclaimer: The Good Investors is the personal investing blog of two simple guys who are passionate about educating Singaporeans about stock market investing. By using this Site, you specifically agree that none of the information provided constitutes financial, investment, or other professional advice. It is only intended to provide education. Speak with a professional before making important decisions about your money, your professional life, or even your personal life. We currently have a vested interest in Alphabet (the company behind AlphaFold), Amazon (the company behind AWS), Apple, Meta Platforms (the company behind Facebook and Instagram), and Tencent (the company behind WeChat Pay). Holdings are subject to change at any time.

We’ve constantly been sharing a list of our recent reads in our weekly emails for The Good Investors.

Do subscribe for our weekly updates through the orange box in the blog (it’s on the side if you’re using a computer, and all the way at the bottom if you’re using mobile) – it’s free!

But since our readership-audience for The Good Investors is wider than our subscriber base, we think sharing the reading list regularly on the blog itself can benefit even more people. The articles we share touch on a wide range of topics, including investing, business, and the world in general. 

Here are the articles for the week ending 29 June 2025:

1. China’s rare earth choke hold – Amber Zhang

Rare earths comprise a group of 17 elements, typically categorized into light, medium, and heavy groups. These materials are indispensable for making high-performance magnets used in both civilian and military technologies. Among them, medium and heavy rare earths — critical for aerospace, defense, and other cutting-edge sectors — are particularly scarce and difficult to source.

Don’t be fooled by their size. Rare earth magnets are no larger than a stick of chewing gum, yet pack magnetic force 15 times stronger than traditional iron magnets. Heat-resistant and cost-efficient, they are essential components in electric motors — not only in EVs and hybrid vehicles, but also in robots, drones, offshore wind turbines, missiles, and fighter jets…

…According to the International Energy Agency, China accounted for over 60% of global rare earth mining output in 2023 — and an even more dominant 92% of the world’s refining capacity. According to the International Energy Agency, China accounted for over 60% of global rare earth mining output in 2023 — and controlled a staggering 92% of global refining capacity…

…Between 2020 and 2023, 70% of the rare earth compounds and metals used in the U.S. were imported from China, according to the U.S. Geological Survey…

…Ford recently halted production for a week at its Chicago plant due to rare earth shortages, affecting its Explorer SUV line. In early June, the Motor & Equipment Manufacturers Association (MEMA), along with General Motors, Toyota, Volkswagen, Hyundai, and other major automakers, issued a joint letter warning that without a stable supply of rare earth magnets, production of essential components could come to a standstill…

…The U.S. once boasted the world’s largest rare earth magnet industry. Its Mountain Pass mine in California had supplied most of the global market since 1965. But in 1998, the mine was shut down following a pipeline leak that released trace heavy metals and radioactive materials into the Mojave Desert. Chinese firms made three separate attempts to acquire the mine — all blocked by U.S. authorities.

Alarmed by Japan’s supply crisis, the Obama administration supported Hitachi Metals’ investment in a rare earth magnet plant in North Carolina, operational from 2011 to 2013. But the costs were prohibitively high compared to China’s vertically integrated, state-backed operations in cities like Ganzhou. U.S. buyers, ultimately unwilling to pay a “made-in-America premium,” continued sourcing from Chinese suppliers. In 2020, Hitachi shut down the facility and mothballed its equipment…

…Back in 2010, Mountain Pass — the U.S.’s only remaining rare earth mine — received over $1 billion in Pentagon funding just to stay afloat. But lacking commercial competitiveness, it shut down again the following year. In 2017, MP Materials acquired the site, restarted mining operations, and began exporting raw ore to China for processing. The company now plans to begin producing rare earth magnets at a new facility in Texas by the end of this year. Still, even at full capacity, its annual output would match just a single day of production in China…

…Domestically, the Round Top project in Texas has emerged as a cornerstone of America’s rare earth strategy. Operated by U.S. Rare Earths Inc., the site holds estimated reserves of 130,000 metric tons across 16 different elements and aims to supply 20% of U.S. rare earth demand by 2027. The company is also building a $100 million magnet manufacturing facility in Oklahoma, which is expected to process up to 2,000 metric tons of rare earth materials annually.

Meanwhile, the U.S. Department of Energy has launched the ReElement initiative, allocating $50 million to recover up to 90% of rare earth elements from electric vehicle batteries by 2025. But these recycling systems have yet to achieve commercial scale and remain economically marginal.

The National Defense Authorization Act for fiscal year 2025 earmarks $1.2 billion for strategic stockpiling and $350 million for domestic development. These funds are being channeled into American firms like MP Materials, aimed at accelerating the construction of a domestic rare earth processing infrastructure…

…According to the Center for Strategic and International Studies (CSIS), the Pentagon has invested over $439 million since 2020 to develop a rare earth industrial base — but most U.S. production remains in its early stages.

RAND Corporation estimates that it would take at least 10 years and $10–15 billion in investment to establish a fully independent domestic rare earth supply chain, factoring in infrastructure, permitting, environmental compliance, and workforce training.

2. The Great Decoupling (or Why Your Clicks Are Down and Impressions Up) – Ryan Law

Impressions are increasing because AI Overviews now give companies two chances to log an impression for a given keyword: once as a “traditional” blue link in the search results, and again as a citation in an AI Overview…

…At the same time, clicks are decreasing because AI Overviews are increasing zero-click searches. Searchers can get all the information they need to resolve their query without leaving the search results page.

When we studied this at scale across 300,000 keywords, we found that the presence of an AI Overview correlates with a 34.5% reduction in clickthrough rate…

…While our clicks are tanking because of AI search, recent data from Patrick Stox shows that—at least on the Ahrefs website—visits from AI search convert 23x better than visits from traditional search.

The way content marketing functions is very different, but guess what? There are more potential customers in the world, more demand for products and services. That is the real determinant of growth, not clicks to a blog. We’ll find different ways to reach those people.

3. A Cheeky Pint with Meta CFO Susan Li (Transcript Here) – John Collison and Susan Li

Susan: When I think about it, I go back to when I was IC4 and I joined in 2008, I’m building these first revenue models. I’d gone from banking – which is super organized, super structured, they don’t even need to know your name, they just train you to immediately figure out how to find the backup to everything, so that two years later someone else can do this and so on and so forth – to there was no infrastructure. So I’m hunting down the exact engineer who has built some ad server so that he can tell me what the parameters mean. And of course, the next time he changes them, he’s not gonna tell me, and I have to go find him again, and he’s like, “Oh, she’s coming. Don’t look her way.” A few months in, I got a meeting invite for power users of SQL and I thought, “My gosh. I’d been getting a good amount of feedback about how things could be better, and here was finally this moment of recognition that – I didn’t even know how to write queries in SQL when I started.” I show up to this meeting and there are five other people and the meeting organizer tells us that we have been called because we are the five users of SQL who consume too much power. And we have just been churning with our massive joint tables through the…

John: I love that you were all called to the principal’s office.

Susan: Basically, yes. But I often think back to this because this was a data analyst who didn’t know any of us that well, but had just generated his reports of who’s using the most infrastructure and looked at the top people on the list and thought, “Okay, this person in finance, it doesn’t make sense why she’s the third highest person on the list,” and called us in and then taught us to write better queries. No one I think specifically told him to do that. I think it’s a little awkward when you call people in to do this, but he did it because it would make us all better at our jobs…

…Susan: So, there’s this very measurable part of the company and we generally try to trade those things off against each other when we’re evaluating things within that bucket and we generally try to fund the things that are positive ROI. I’m usually the person who’s trying to make sure we understand, for every individual experiment, the expected return is something, but that’s where we are on the curve today, but what about 50 experiments later? Does the curve still have the same slope?

Then there’s a set of things which we constrain more in terms of, there’s some envelope of investment that we’re willing to make that’s not in this really ROI-driven bucket. It is very difficult to pencil out what the annual revenue forecast for Reality Labs is gonna look like over the next 20 years. For bets like that, we invert the problem. But when we talk about the return on the investment, the question that we pose, as a finance organization, to Mark – and make sure that Mark and the board understand – is what does this have to be worth to pencil out at the end? Does that pass the sanity check, the intuition, about what the size of these markets can be based on maybe some comparisons to markets that exist today, but of course in another 10, 20 years, you expect that the world will look different and maybe those markets should be bigger or smaller for whatever reason. That’s the guide, which is, for this thing to succeed at the rate at which we’re investing, it needs to be worth this at the end and does that make sense?…

…Susan: I am not a tech visionary. There are many things I’m good at, but envisioning the future of the world and what I want it to be like is not one of them. I’m a very happy beneficiary of the technology built by the world around me.

But Mark very much has a vision for what he wants that world to be. And for him, I think the strategic imperative is that we have to be building these next states of the world for us to again, be a good business, but also just be a compelling company that builds technology and puts it out in the world and builds incredible experiences for people.

I remind people in the finance organization all the time, we are very good at skeptically evaluating each bet. But the point is not that we have to look at every bet and be like, “This bet is going to work.” The point is there is a portfolio of bets, and some of them are going to pay off massively beyond, in fact, what the case on paper looks like when you make the bet. Many of them are going to not work out, but the ones that pay off are gonna more than justify the overall investment strategy or the overall roadmap that you’re building toward. If we just allowed ourselves to nix everything that the paper-case didn’t seem high-confidence, then we would never make a lot of the important bets that have been really important over the history of the company…

…Susan: That is the question that I assume all of my counterparts at these companies and I are all thinking about. For us, there are the drivers of the way we’re investing in capex today. Of course, we have, first of all, just a massively-scaled consumer business and core AI infrastructure that powers all the ranking and recommendations work and so on and so forth. That’s always been a reasonably big number for us, but also because it was getting more mature that we were driving to be more efficient over time. Then now you have, among many of our peers and ourselves, this big investment to train what we all aspire to be, frontier models. If you use those models to build great and scaled consumer experiences, then how much inference compute you’re gonna need on top of that? If compute required continues to scale up in this way forever, then you’re gonna run into some true problems of physics. But hopefully, there will be different kinds of research innovations along the way that will unlock things like being able to distribute the training so you don’t need one extremely large cluster somewhere and that will help with a lot of the energy and other challenges. So there’s some question about what that looks like over time.

Then there’s this question about, “Great, you can build all this capacity, and what do you do with them if it turns out you don’t need as much compute for either training or inference as you thought?” I think a lot of us have different backup use cases. So, up to some point, we would use a lot of compute very happily still, in the core business and what we expect the core business to be, three years from today. But frankly, we’d use more compute in the core business. Now, that doesn’t scale forever. So the real question is what happens in like two years if you’ve built so much compute that you cannot envision a reasonable ROI on the backup use case if what you’re building doesn’t come to fruition. That’s something I think we’re all gonna learn in the next few years…

…Susan:  As part of not wanting to miss the boat, we built out enough capacity for Reels but also for future things. We found that we were in fact able to put that capacity towards very good use – exactly as you said. So I do think an interesting question in the future will be allocating compute as a resource, It’s a muscle we’ve built later as a company, because we had gotten very good at allocating headcount as a resource, and headcount’s really easy to account for because you have org charts, you know exactly this person reports to this person, to this person, this person is incontrovertibly working on Facebook Marketplace, for example. GPUs don’t have that property. In fact, you often want to build out your infrastructure for it to be very fungible. Because you need to divert capacity to where – suddenly something has happened in India and you want a lot of compute to be available to be used there. So it’s not like this GPU is labeled for Facebook Marketplace, and this is labeled for – it’s actually quite a bit more difficult to account for where the capacity is being used at any given point in time. That means it’s harder to manage, and it’s harder to create the incentives around are you using GPUs efficiently?

John: You allow people to trade between people and GPUs, right?

Susan: In the budgeting process, we have allowed people to trade. Not too surprisingly, even though you’ll find that groups are often asking for compute, when that particular trade is on offer, people almost never trade for compute for exactly the reason I described, which is that if they get allocated 100 new headcount, there is no chance that 26 of those headcount will accidentally be working for something else.

4. My Trip to Washington to Get in Sync with Republican and Democratic Leaders on the Budget and Debt Situation – Ray Dalio

Everyone I spoke with on both sides agreed that:

• We are likely to have a big debt-economic crisis if we don’t get the budget deficit down to 3 percent of GDP, so 3 percent should be an agreed-on goal,
• Getting the deficit to 3 percent will require both spending cuts and tax revenue increases because if they come from just spending cuts or just tax revenue increases alone, the cuts or increases would be too big and shocking.
• It’s not possible for politicians to say these things publicly even though they believe them because they would be thrown out of office…

…So, our biggest problem is that our country’s political representatives can’t even say, let alone do, what they need to do to fix our debt issues because their constituents would throw them out of office if they did that. Such is the condition of our political decision-making system.

We discussed my idea of a “3 percent 3-part solution,” which would be to cut the budget deficit to 3 percent of GDP through a mix of spending cuts, tax revenue increases, and interest rate cuts. For example, cutting spending by 4 percent, increasing tax revenue by 4 percent, and lowering the real interest rate by 1%** so that the adjustments wouldn’t be unbearably large to achieve that 3% deficit goal. The leaders I spoke with said that they’d love to do this or something like it — in fact, they thought it would be wonderful if the “meme” of reducing the deficit in this way took hold in the electorate and there was public pressure to get it done.

As for where things are likely to go, there won’t be big enough changes to the current proposed budget to change the overall picture this tax year.

5. The Speed of Patience – Paul Higgins

To understand how patient preparation creates decisive speed, I’ll show you three different maps of the same territory I’ve found practical.

1. Pace layers reveal where to be patient and where to be urgent, showing how businesses operate across multiple timescales simultaneously, from seasonal fashion to generational culture.
2. S-curves illuminate when those layers will hit their inflection points, helping you recognize which growth curve you’re actually betting on.
3. Trust as a leading indicator – what emerges from ongoing interactions across and between layers (employees, communities, customers and processes), the invisible asset that compounds for decades…

…I like Stewart Brand’s pace layering framework for understanding how businesses operate across time. It reveals why this matters so profoundly. In most complex systems, different elements change at different speeds. Fashion moves seasonally, commerce shifts yearly, infrastructure evolves over decades, governance changes generationally, and culture moves so slowly it appears frozen in time…

…Layers don’t exist separately, they form a single, interconnected living system which is sometimes hard to see. We tend to see layers as independent parts to optimize separately, but in living systems, layers are how the whole organism breathes – each rhythm nested within another, each movement part of a larger dance. The fast movements at the surface and slow currents in the depths aren’t separate phenomena but the system’s way of being alive at every scale simultaneously. Speed doesn’t come from stability – they arise together from the coherence of the whole system…

…Apple master this temporal arbitrage. New iPhone colors arrive every season to satisfy the fashion layer, while annual product cycles drive the commerce layer with reliable predictability. But the iOS ecosystem, which represents their true competitive moat, took twenty years to build in the infrastructure layer, creating switching costs and network effects that compound with each passing year. Their App Store governance evolves with glacial deliberation, each change carefully considered for its long-term implications, while their design philosophy – the cultural layer that infuses everything they create – hasn’t fundamentally changed since Jobs articulated it decades ago. You just have to look at their cumulative cash reserves to see whether they have the capacity to keep it up or not.

Competitors try to destroy Apple’s fashion layer moat and assume that’s the game being played. They miss the insight that Apple’s speed in the fashion layer comes from stability in the infrastructure layer, that the layers aren’t independent but deeply interdependent, with the slow layers enabling the fast ones to move with confidence and clarity…

… In business, you’re never riding just one S-curve. You’re managing a portfolio of them, each operating at different speeds across different layers of your organization. Your product adoption might be hitting exponential growth (measured in months) while your infrastructure build-out is still in early grind (measured in years) and your culture formation hasn’t even begun its curve (measured in decades)…

…Netflix understood this with brutal clarity. In 2010, they were shipping 2 million DVDs daily – a massive operation at the peak of its S-curve. But Reed Hastings saw streaming was at the bottom of its S-curve, barely functional, with terrible selection and constant buffering. While Blockbuster optimized their mature retail model, Netflix deliberately cannibalized their profitable DVD business to ride the next wave. They moved $200 million from DVD operations into streaming content when streaming represented less than 20% of revenue. Today Netflix is worth $240 billion; Blockbuster is a cautionary tale…

…Kerry Group’s transformation from Ireland’s smallest dairy cooperative to a €6.3 billion ingredients empire illustrates how patience creates opportunities invisible to those focused on shorter horizons. Every dairy producer faced the same challenge with whey, the protein-rich liquid left over from cheese-making that represented both a disposal cost and a compliance headache. While the entire industry treated this as expensive waste, Kerry’s leadership recognized something profound: they were looking at two different S-curves operating on completely different timescales.

The dairy business that consumed everyone’s attention was approaching the top of its S-curve, with margins thinning and consolidation inevitable, while the ingredients business hadn’t even begun its exponential climb. For fifteen years, Kerry invested in extraction technology and scientific capabilities while competitors focused on optimizing dairy margins. By the time health consciousness and specialized nutrition exploded into mainstream consciousness, Kerry had spent two decades perfecting protein extraction, understanding molecular structures, and building relationships with food manufacturers who needed exactly these capabilities…

…Warren Buffett’s 2008 moves exemplified how trust operates across all three maps simultaneously. While others mocked Berkshire’s growing cash pile – $40 billion sitting “idle” – he was building in the infrastructure layer (pace layers), preparing for the inevitable down-cycle in financial services’ S-curve, and accumulating trust with every patient year. That cash pile represented more than financial capacity; it was trust crystallized into capital. Every year Buffett didn’t chase returns, every quarter he resisted leverage, every deal he walked away from, he was depositing into an invisible trust account. When 2008 hit, that patient accumulation enabled lightning-fast execution: $8 billion deployed to Goldman Sachs with one phone call. The $7.7 billion total return exceeded Coca-Cola’s entire 20-year dividend stream to Berkshire. Trust had compressed decades into days.

---

Disclaimer: The Good Investors is the personal investing blog of two simple guys who are passionate about educating Singaporeans about stock market investing. By using this Site, you specifically agree that none of the information provided constitutes financial, investment, or other professional advice. It is only intended to provide education. Speak with a professional before making important decisions about your money, your professional life, or even your personal life. We currently have a vested interest in Alphabet (the company behind AI Overviews), Apple, Meta Platforms, and Netflix. Holdings are subject to change at any time.

We’ve constantly been sharing a list of our recent reads in our weekly emails for The Good Investors.

Do subscribe for our weekly updates through the orange box in the blog (it’s on the side if you’re using a computer, and all the way at the bottom if you’re using mobile) – it’s free!

But since our readership-audience for The Good Investors is wider than our subscriber base, we think sharing the reading list regularly on the blog itself can benefit even more people. The articles we share touch on a wide range of topics, including investing, business, and the world in general. 

Here are the articles for the week ending 22 June 2025:

1. Message from CEO Andy Jassy: Some thoughts on Generative AI – Andy Jassy

Today, in virtually every corner of the company, we’re using Generative AI to make customers lives better and easier. What started as deep conviction that every customer experience would be reinvented using AI, and that altogether new experiences we’ve only dreamed of would become possible, is rapidly becoming reality. Technologies like Generative AI are rare; they come about once-in-a-lifetime, and completely change what’s possible for customers and businesses…

…You can see it in Advertising where we’ve built a suite of AI tools that make it easier for brands to plan, onboard, create and optimize campaigns. In Q1 alone, over 50K advertisers used these capabilities…

…We’re also using Generative AI broadly across our internal operations. In our fulfillment network, we’re using AI to improve inventory placement, demand forecasting, and the efficiency of our robots—all of which have improved cost to serve and delivery speed. We’ve rebuilt our Customer Service Chatbot with GenAI, providing an even better experience than we’d had before. And, we’re assembling more intelligent and compelling product detail pages from leveraging GenAI…

…First, we have strong conviction that AI agents will change how we all work and live. Think of agents as software systems that use AI to perform tasks on behalf of users or other systems. Agents let you tell them what you want (often in natural language), and do things like scour the web (and various data sources) and summarize results, engage in deep research, write code, find anomalies, highlight interesting insights, translate language and code into other variants, and automate a lot of tasks that consume our time. There will be billions of these agents, across every company and in every imaginable field. There will also be agents that routinely do things for you outside of work, from shopping to travel to daily chores and tasks. Many of these agents have yet to be built, but make no mistake, they’re coming, and coming fast.

Second, and what makes this agentic future so compelling for Amazon, is that these agents are going to change the scope and speed at which we can innovate for customers. Agents will allow us to start almost everything from a more advanced starting point…

…Today, we have over 1,000 Generative AI services and applications in progress or built, but at our scale, that’s a small fraction of what we will ultimately build. We’re going to lean in further in the coming months. We’re going to make it much easier to build agents, and then build (or partner) on several new agents across all of our business units and G&A areas.

As we roll out more Generative AI and agents, it should change the way our work is done. We will need fewer people doing some of the jobs that are being done today, and more people doing other types of jobs. It’s hard to know exactly where this nets out over time, but in the next few years, we expect that this will reduce our total corporate workforce as we get efficiency gains from using AI extensively across the company.

2. Experiencing the Real “Belt and Road” – Nina Chen

In early June, I traveled in Central Asia for 9 days, visiting two countries. I spent 2 days in Almaty, Kazakhstan, and 7 days in Uzbekistan, covering Tashkent, Samarkand, and Bukhara…

…We flew from Almaty, Kazakhstan, to Tashkent, the capital of Uzbekistan. Even before landing, it was clear that Uzbekistan and China have a close partnership. On the flight, there were many Chinese merchants and workers traveling in groups…

…When we arrived at the airport, the sense of close cooperation was even stronger. The airport signs had Chinese translations, and there was a billboard in the walkway advertising the “UZ-China Silk Road Free Trade Special Zone.”…

…While we didn’t meet any locals in Uzbekistan who’d been to China, in Kazakhstan, we met a Kazakh girl with fluent Chinese. We joined a day tour to the lakes and canyons near Almaty. With many Chinese tourists in our group, she translated for us when we couldn’t understand the guide. She studied in Chongqing(*) and worked in Yiwu, Zhejiang province (*), where her Chinese boss ran a company exporting goods from China to former Soviet countries like Moscow, Azerbaijan, and Central Asian cities.

This made me feel that trade between China and Central Asia is largely a one-way flow, from China to Central Asia, with China’s economic influence in the region being substantial…

…At the Tashkent City Mall, the premier shopping destination in Uzbekistan’s capital, I was surprised to find stores for well-known Chinese sportswear brands Anta, Li-Ning, and Xtep all located in close proximity.

I decided to explore the Anta store first. Picking up a pair of PG 7 running shoes (the PG 7 refers to the midsole technology), I noticed the price tag read 1,103,000 Uzbekistani som (approximately 612 Chinese yuan, US$87), which is significantly higher than the price in China (where it’s around 200-300 yuan, US$29–43 on Tmall). However, the store currently has a promotion: buy one pair and get the second at 50% off (effectively 459 yuan per pair, US$66) or buy two pairs and get the third free (bringing the cost down to 408 yuan per pair, US$58). Even with the discounts, the price is still higher than in China. When I asked the store manager if Anta is considered a premium brand in Uzbekistan, he confirmed it is. Surprised, I inquired if only the wealthy can afford it. He explained that due to the popularity of digital payments, many people, especially the youth, opt for installment plans…

… Central Asia has many Chinese-made beauty and skincare products that aren’t available in China.

An example is “Shanghai Song,” with packaging featuring a classic Chinese vintage design. The brand’s slogan states: “Inspired by myths and legends, it’s about Shanghai in the Song period, which ruled one of China’s most glorious cultural eras in the long-flowing Eastern cultural river.”

I found this puzzling. First, the specific myths or legends that served as inspiration aren’t clear, giving it a mysterious and abstract feel. Second, to the best of my knowledge, during the Song Dynasty, the economic and cultural centers of the Northern Song were in Kaifeng, and those of the Southern Song were in Hangzhou, not in Shanghai. Perhaps “Shanghai Song” represents a blend of the modern and the classical, or maybe the company behind the brand has a special affection for Shanghai.

When I picked up a bottle of cream and examined it closely, I found that the company is based in Guangzhou. Well, it’s likely that “Shanghai Song” is a brand from Guangzhou that embodies what Chinese people think Central Asians imagine about China and the East.

3. The Capital Cycle Way – Omar Malik

The capital cycle best explains how changes in the amount of capital employed within an industry will impact profits and future returns on capital.

Central to the capital cycle approach is the observation that an industry with high returns on capital tends to attract new entrants. For incumbents, high profitability loosens discipline because management incentives often align with growth. Therefore, both groups will increase spending to capture those high returns. The behavioural pattern of herding often means all the players in an industry invest simultaneously…

…A key characteristic of this cycle is the delay between the investment decision and the new supply coming online. By the time the new supply arrives, historical demand forecasts are often shown to have been overly optimistic, creating an overhang. This causes returns on capital to fall below the cost of capital. As profits collapse, management teams are changed, spending is slashed, and the industry begins to consolidate. That contraction in supply eventually paves the way for a recovery in returns…

…Supply dynamics are more certain than demand and therefore easier to forecast. This is because increases in industry supply are often well-flagged by management teams. In certain industries, such as aircraft manufacturing and shipbuilding, the supply pipelines are well-known. New entrants will noisily announce their arrival into an industry…

… Studying the supply side can help you identify companies that are likely to sustain their high returns for decades to come. The lack of competition due to a competitive moat prevents the supply side from shifting in response to high profitability and defies the typical mean revision in returns…

…Buffett’s investment cases are often predicated on a supply-side focus, and his acquisition of BNSF Railway is a good example. In his own words, the railroad industry had a ‘terrible century’ leading up to his investment. But after following the industry from a young age, he became interested in 2006, why?

The industry had rationalised from over 100 players in the 1960s to just five. In the 1990s, a final wave of consolidation led to the formation of today’s giants. The relative competitive position of railroads versus trucking had improved as oil prices rose, making the railroads the lowest-cost way to move heavy freight. No new capacity was being built. And after consolidating, driving efficiency became the focus, with the labour force falling by 90% and the introduction of new innovations, such as double stacking.

Putting that all together, as long as you believed that the US economy would grow over the coming decades, the structurally improved supply-side dynamics would lead to higher returns on capital in the future. He was not focused on demand because he acquired BNSF during the global financial crisis (GFC), the worst economic crisis since the Great Depression…

…We have held TSMC since Hosking Partners’ inception in 2013 — in fact, it dates back even earlier, if you include the years at Marathon.

The semiconductor industry is highly cyclical, and the news flow around the cycle is immense. Analysts are obsessed with questions such as: Are we at a peak or trough earnings cycle? Was that the last cut or the last beat? How many quarters will the trough last?

Our thesis for the last 15 years has been based on a simple insight: the foundry business would consolidate over time, given the ever-rising cost of advancing Moore’s law. And that TSMC had the superior model, as a pure-play foundry, creating a true alignment with the customer, completely agnostic to the end market. Today, we feel that insight still holds. The scale advantage of TSMC’s model has only grown as the industry has gone from over 20 players to just three…

…How a management team responds to the capital cycle in their industry is critical. If they can act counter-cyclically, pull back when others are adding supply, and take advantage of downturns, they can create significant value.

The way I think about it is if you find one of these outlier teams, you can subcontract the capital allocation decisions to them. You can trust them to navigate the cycles instead of trying to time the buy and sell decisions…

…Even if you have a fix on the supply side for the next decade and you trust management to allocate capital well, you still need to buy at the right price! That brings me to the fourth tenet – remember replacement value.

It is a simple concept: how much would it cost to reproduce or replicate this asset? It is the driving force of the capital cycle. When companies are valued at a premium to replacement cost in the equity market, it creates an incentive to invest and capture that arbitrage. That is why venture capital and private equity funding is tied to equity market valuations.

It is far easier to calculate replacement value in asset-intensive industries with readily available data. But it is more of an art in other sectors, where the model is asset-light with a greater share of intangibles. In such cases, a question I often think about is, “Should we compete with this business instead of buying it?”…

…The final point I’ll leave you with is that we are all guilty, including myself today, of singling out the parts of Buffett’s approach that appeal to us. It is natural, as we all look for confirmation in the tough pursuit of outperforming. I am convinced that the capital cycle lens is one of Buffett’s big mental models for the world.

But my ultimate takeaway from studying Buffett and attending these annual meetings is that he is the Swiss Army Knife of investing. Over his long career, Buffett has successfully invested in great compounders across a wide range of industries (i.e., Coke, Amex, Apple); deep value (i.e., PetroChina on a 3x P/E, as well as all the early partnership investments); activism (i.e., Sanborn maps, Berkshire Hathaway); baskets (i.e., Korean stocks, railroads, airlines, Japanese trading houses); merger arbitrage (i.e. Activision Blizzard); bonds (i.e., high-yield bonds in the fallout of the tech bubble); commodities (i.e., oil futures, silver, and more recently Occidental), among others.

4. A Moody’s Ratings Downgrade for the US: What now? – Aswath Damodaran

Through time, governments have often been dependent on debt to finance themselves, some in the local currency and much in a foreign currency. A large proportion of sovereign defaults have occurred with foreign currency sovereign borrowing, as the borrowing country finds itself short of the foreign currency to meet its obligations. However, those defaults, and especially so in recent years, have been supplemented by countries that have chosen to default on local currency borrowings. I use the word “chosen” because most countries have the capacity to avoid default on local currency debt, being able to print money in that currency to pay off debt, but chose not to do so, because they feared the consequences of the inflation that would follow more than the consequences of default…

…Researchers who have examined the aftermath of default have come to the following conclusions about the short-term and long-term effects of defaulting on debt:

1. Default has a negative impact on the economy, with real GDP dropping between 0.5% and 2%, but the bulk of the decline is in the first year after the default and seems to be short lived.
2. Default does affect a country’s long-term sovereign rating and borrowing costs. One study of credit ratings in 1995 found that the ratings for countries that had defaulted at least once since 1970 were one to two notches lower than otherwise similar countries that had not defaulted. In the same vein, defaulting countries have borrowing costs that are about 0.5 to 1% higher than countries that have not defaulted. Here again, though, the effects of default dissipate over time.
3. Sovereign default can cause trade retaliation. One study indicates a drop of 8% in bilateral trade after default, with the effects lasting for up to 15 years, and another one that uses industry level data finds that export-oriented industries are particularly hurt by sovereign default.
4. Sovereign default can make banking systems more fragile. A study of 149 countries between 1975 and 2000 indicates that the probability of a banking crisis is 14% in countries that have defaulted, an eleven percentage-point increase over non-defaulting countries…

…If sovereign ratings are designed to measure exposure to default risk, how well do they do? The answer depends on how you evaluate their performance…

…In sum, the evidence suggests that while sovereign ratings are good measures of country default risk, changes in ratings often lag changes on the ground, making them less useful to lenders and investors.

If the key limitation of sovereign ratings is that they are not timely assessors of country default risk, that failure is alleviated by the development of the sovereign CDS market, a market where investors can buy insurance against country default risk by paying an (annualized) price. While that market still has issues in terms of counterparty risk and legal questions about what comprises default, it has expanded in the last two decades, and at the start of 2025, there were about 80 countries with sovereign CDS available on them…

…At the start of 2025, the market was drawing a distinction between the safest Aaa-rated countries (Scandinavia, Switzerland, Australia and New Zealand), all with sovereign CDS spreads of 0.20% or below, and more risky Aaa-rated countries (US, Germany, Canada). During 2025, the market shocks from tariff and trade wars have had an effect, with sovereign CDS spreads increasing, especially in April. The US, which started 2025 with a sovereign CDS spread of 0.41%, saw a widening of the spread to 0.62% in late April, before dropping back a bit in May, with the Moody’s downgrade having almost no effect on the US sovereign CDS spread…

…The ramping up of US debt since 2008 is reflected in total federal debt rising from 80% of GDP in 2008 to more than 120% in 2024. While some of the surge in debt can be attributed to the exigencies caused by crises (the 2008 banking crisis and the 2020 COVID bailouts), the troubling truth is that the debt has outlasted the crises and blaming the crises for the debt levels today is disingenuous.

The problem with the debt-to-GDP measure of sovereign fiscal standing is that it is an imperfect indicator…

…Many of the countries with the highest debt to GDP ratios would be classified as safe and some have Aaa ratings, whereas very few of the countries on the lowest debt to GDP list would qualify as safe. Even if it is the high debt to GDP ratio for the US that triggered the Moody’s downgrade, the question is why Moody’s chose to do this in 2025 rather than a year or two or even a decade ago, and the answer to that lies, I think, in the political component. A sovereign default has both economic and political roots, since a government that is intent on preserving its credit standing will often find ways to pay its debt and avoid default. For decades now, the US has enjoyed special status with markets and institutions (like ratings agencies), built as much on its institutional stability (legal and regulatory) as it was on its economic power. The Moody’s downgrade seems to me a signal that those days might be winding down, and that the United States, like the rest of the world, will face more accountability for lack of discipline in its fiscal and monetary policy…

…The ratings downgrade was after close of trading on Friday, May 16, and there was concern about how it would play out in markets, when they opened on Monday, May 19. US equities were actually up on that day, though they lost ground in the subsequent days…

…If equity markets were relatively unscathed in the two weeks after the downgrade, what about bond markets, and specially, the US treasury market? After all, an issuer downgrade for any bond is bad news, and rates should be expected to rise to reflect higher default risk…

…While rates did go up in the the first few days after the downgrade, the effect was muddled by the passage of a reconciliation bill in the house that potentially could add to the deficit in future years. In fact, by the May 29, 2025, almost all of the downgrade effect had faded, with rates close to where they were at the start of the year…

…The expected return on the S&P 500 as of May 30, 2025, reflecting the index level then and the expected cash flows, is 8.64%. Incorporating the effects of the downgrade changes the composition of that expected return, resulting in a lower riskfree rate (4.01% instead of 4.41%) and a higher equity risk premium (4.63% instead of 4.23%). Thus, while the expected return for the average stock remains at 8.64%, the expected return increases slightly for riskier stocks and decreases slightly for safer stocks, but the effects are so small that investors will hardly notice. If there is a lesson for analysts here, it is that the downgrade’s effects on the discount rates (costs of equity and capital) are minimal, and that staying with the conventional approach (of using the ten-year US treasury bond rate as the riskfree rate and using that rate to compute the equity risk premium) will continue to work.

5. Contrary Research Rundown #140 – Contrary Research

Tesla has taken a fundamentally different approach. It does not use lidar or radar and instead relies entirely on eight cameras to make driving decisions. In contrast, Waymo’s fifth-generation car has 29 cameras, six radar sensors, and five lidar sensors…

…As early as 2013, Elon expressed skepticism about the need for lidar in autonomous vehicles. Elon framed the reason in a rather intuitive way in 2021: if humans can rely on their eyes and brain, then self-driving cars can rely on cameras and AI…

…Another reason Tesla has avoided using lidar is the cost. One 2024 report estimated Tesla’s sensor suite costs just $400 per vehicle, compared to an estimated $12.7K per vehicle for Waymo’s sensors on its fifth-generation Jaguar SUVs…

…Companies like Waymo follow a multi-step process where they first deploy vehicles with safety drivers to record and map the area, which can take months for each new city and requires continuous updates. Waymo and companies like it then use these predefined maps to complement their real-time sensor data from lidar/radar about the surrounding area. Tesla, by contrast, claims its software can operate anywhere without pre-mapped data, relying entirely on real-time camera input to understand road conditions…

…At Google I/O in May 2025, Waymo showed a few examples where its full suite of sensors successfully avoided pedestrians and where it claims a camera-only approach would have struggled.

In one example, Waymo’s lidar picked up the presence of a pedestrian in a Phoenix dust storm that was not visible on the camera…

…In another example, Waymo’s sensors were able to detect a pedestrian who was behind a bus and avoid a collision:

“We are detecting a pedestrian on the other side of the bus. That would be completely occluded to a human driver. So what’s happening here is that our sensors are able to pick up the movement of the person’s feet under the bus. And just that little bit of noisy and sparse signal is enough for the Waymo Driver to detect that there’s a pedestrian there and, furthermore, to predict what they’re going to do in the future, allowing us to take a defensive action early.”…

…Waymo has only had one fatal accident in its history, and not due to a Waymo error. In January 2025, a Tesla struck an unoccupied Waymo and other cars at a red light, killing one person. As we wrote in our last piece, one study by Swiss Re shows Waymo saw an 88% reduction in property damage claims and a 92% reduction in bodily injury claims when compared to human-driven vehicles…

…In 2023, a Tesla in Full Self Driving mode (FSD) hit a 71-year-old woman at highway speed, killing her. Video of the crash shows a sun glare appearing to blind the camera, and the National Highway Traffic Safety Administration (NHTSA) opened an investigation into Tesla in October 2024 for four total FSD collisions that occurred in low visibility situations…

…When Elon first called lidar too expensive in the early 2010s, it cost ~$75K per unit. Since then, costs have fallen dramatically, and some lidar units sold for personal vehicles (not robotaxis) are being priced in the hundreds of dollars…

…By one estimate, lidar costs have fallen by 99% since 2014.

---

Disclaimer: The Good Investors is the personal investing blog of two simple guys who are passionate about educating Singaporeans about stock market investing. By using this Site, you specifically agree that none of the information provided constitutes financial, investment, or other professional advice. It is only intended to provide education. Speak with a professional before making important decisions about your money, your professional life, or even your personal life. We currently have a vested interest in Alphabet (parent of Google), Amazon (company where Andy Jassy is the CEO), Apple, Tesla, and TSMC. Holdings are subject to change at any time.