What We’re Reading (Week Ending 21 September 2025) - 21 Sep 2025
Reading helps us learn about the world and it is a really important aspect of investing. The late Charlie Munger even went so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 21 September 2025):
1. AI Will Not Make You Rich – Jerry Neumann
Fortunes are made by entrepreneurs and investors when revolutionary technologies enable waves of innovative, investable companies. Think of the railroad, the Bessemer process, electric power, the internal combustion engine, or the microprocessor—each of which, like a stray spark in a fireworks factory, set off decades of follow-on innovations, permeated every part of society, and catapulted a new set of inventors and investors into power, influence, and wealth.
Yet some technological innovations, though societally transformative, generate little in the way of new wealth; instead, they reinforce the status quo. Fifteen years before the microprocessor, another revolutionary idea, shipping containerization, arrived at a less propitious time, when technological advancement was a Red Queen’s race, and inventors and investors were left no better off for non-stop running…
…1. Containerization History: The benefits of the tech are obvious, leading many companies to enter. AI Rhyme: The idea that AI is the next big thing is widespread, and entrepreneurs and tech companies quickly enter.
2. Containerization History: There is immediate government and social attention, leading to pushback. AI Rhyme: The debate over AI that immediately surfaces in society, the media, and government limits experimentation.
3. Containerization History: Shipbuilders and other infrastructure companies get a quick boost, but not a long-lasting one. AI Rhyme: Chip makers, data center builders, and data providers get a quick boost, but not a long-lasting one.
4. Containerization History: Competitive intensity makes it difficult to keep prices high or lower costs, and forces high spending on capex, R&D, and talent. AI Rhyme: Prices start to drop even as companies spend heavily on capex, R&D, and talent. Companies will not be especially profitable.
5. Containerization History: The industry searches for ways to limit competition through cartels and regulatory bodies. AI Rhyme: Investors become alarmed and push for rationalization, resulting in consolidation and convergence on a few business models.
6. Containerization History: The value created by the innovation is zero-sum: who captures it (provider vs customer) determines the structure of the resulting industry. AI Rhyme: Companies vertically integrate into their customers’ businesses. Companies built on another company’s model have their margins or business model subsumed. Model companies become generalized AI providers.
7. Containerization History: The longer-term beneficiaries of increased productivity are existing companies that dramatically reduce prices or open new markets to their products. Most incumbents don’t do this. AI Rhyme: The beneficiaries of increased productivity in “thinking” are existing knowledge-industry service providers. Those that won’t adapt will die.
In the “AI rhymes” column, the first four items are already underway. How you should invest depends on whether you believe Nos. 5–7 are next…
…The high capex of AI companies will primarily be spent with the infrastructure companies. These companies are already valued with this expectation, so there won’t be an upside surprise. But consider that shipbuilding benefited from containerization from 1965 until demand collapsed after about 1973.[19 If AI companies consolidate or otherwise act in concert, even a slight downturn that forces them to conserve cash could turn into a serious, sudden, and long-lasting decline in infrastructure spending. This would leave companies like Nvidia and its emerging competitors—who must all make long-term commitments to suppliers and for capacity expansion—unable to lower costs to match the new, smaller market size. Companies priced for an s-curve are overpriced if there’s a peak and decline.
All of which means that investors shouldn’t swim upstream, but fish downstream: companies whose products rely on achieving high-quality results from somewhat ambiguous information will see increased productivity and higher profits. These sectors include professional services, healthcare, education, financial services, and creative services, which together account for between a third and a half of global GDP and have not seen much increased productivity from automation. AI can help lower costs, but as with containerization, how individual businesses incorporate lower costs into their strategies—and what they decide to do with the savings—will determine success. To put it bluntly, using cost savings to increase profits rather than grow revenue is a loser’s game.
The companies that will benefit most rapidly are those whose strategies are already conditional on lowering costs. IKEA’s longtime strategy was to sell quality furniture for low prices and make it up on volume. After containerization made it possible for them to go worldwide, IKEA became the world’s largest retailer and Ingvar Kamprad (the IK of IKEA) became a billionaire. Similarly, Walmart, whose strategy was high volume and low prices in underserved markets, benefited from both cost savings and just-in-time supply chains, allowing increased product variety and lower inventory costs.
2. Getting Rich on Rocks – Joe Raymond
35% per year for 19 years results in a 300x return.
This is a Hall of Fame result. It’s an incredible feat in only two decades for a single stock…
…But what if I told you there was an obscure OTC stock that returned more than 35% annually from 1993 to its acquisition in 2012?
You almost certainly haven’t heard of this company. Its executives aren’t on the covers of any magazines and haven’t written any bestselling books. And its shareholders quietly made their fortune without anybody noticing.
To make matters even more interesting, this was an aggregates business. That’s right, the company sold rocks…
…Let me tell you about Western Lime…
…Western Lime was an unremarkable business in the ’90s.
Growth was around 5-6% per year, and ROE hovered around 10%. Decent, but not particularly noteworthy.
What was noteworthy was the price.
For much of the ’90s, WLC traded between $150 and $160 per share. Trades were very infrequent. The stock only changed hands a few times a year.
At $155 per share in 1993, Western Lime had a market capitalization of only $2 million. The company earned $1.3 million after-tax that year, good for a P/E ratio of 1.7x.
Shareholders’ equity was $12.7 million, so the P/B was 0.16x.
The company had no debt and paid a small quarterly dividend…
…In addition to being incredibly cheap, the company itself was repurchasing shares in private transactions at $550 (more than triple the OTC price). I don’t know if anyone was arbing this, but I bet somebody was…
…By 2010, Tweedy owned 27% of Western Lime’s outstanding shares…
…By this point, word had started to get out on WLC. It was no longer a completely undiscovered stock selling for less than 2x earnings. It was then trading for $5,600 per share…
…Performance had been solid from 1993 to 2009.
Net income grew at 13% per year, the share count was cut in half, and the P/E multiple more than doubled from 1.7x to 3.7x.
The result was a 25% CAGR before dividends from 1993 to 2009…
…In late 2010, we received a string of correspondence between the company and Tweedy, Browne. It was sent to all shareholders. And it made for compelling reading.
The company had offered Tweedy $7,600 per share to acquire their 27% interest (36% above the prevailing $5,600 share price).
This equated to about 5x trailing earnings and 86% of tangible book value…
…Tweedy pegged the intrinsic value of WLC at somewhere between $24,000 and $33,600 per share. This equated to 8.5x EBITDA (15.9x earnings) on the low end and 11.9x EBITDA (22.0x earnings) on the high end…
…Tweedy ultimately rejected the bid, saying that they would much rather buy shares at $7,600 than sell them…
…What’s interesting is that the company upped their bid to $10,300 per share based on “an independent valuation of WLC’s stock” which includes “a discount for lack of marketability of minority blocks of stock.”…
…WLC traded for less than $200 per share 15 years prior. The current market was around $5,600. The company was offering $10,300. And they showed no signs of getting serious about selling the entire company or uplisting the stock.
In other words, there was no other clear “catalyst” on the horizon, other than this seemingly juicy offer from the company.
But Tweedy stuck to their core principles and refused to sell below intrinsic value.
They declined the bid and continued to hold their shares…
…Western Lime ended up selling to Graymont a little over a year later in March 2012…
…Shareholders received $52,000 per share.
That’s more than 5x the price offered to Tweedy less than two years prior, and a 36% CAGR from the 1993 price of $155 (before dividends).
3. From flops to fortune: How tech’s biggest failures create tomorrow’s winners – Chin Hui Leong
Ever since OpenAI launched ChatGPT in November 2022, Alphabet has found itself in an unfamiliar situation – playing second fiddle to OpenAI’s popular artificial intelligence (AI) assistant.
But with the recent launch of Gemini 2.5 Flash Image, Google is starting to look innovative again. The new image feature (code-named Nano Banana) attracted more than 10 million new users in a week, with over 200 million images edited.
Here’s what most people don’t realise: Nano Banana’s success was about 15 years in the making. The story begins with Google+, the company’s catastrophic attempt to challenge Facebook. Launched in 2011, Google+ burned through hundreds of millions before being shuttered in 2019.
But buried within that failed social network was a gem – Google Photos. When Google Photos became a standalone product in 2015, it brought along the image editing and organisation capabilities developed for Google+. Those capabilities – during its failed social network experiment – would give Google’s image AI the headstart it needed.
Fast forward to today, the technology that couldn’t save a social network now powers Google’s comeback in the AI race. Nano Banana’s overnight success took about 15 years of patient failure…
…For investors, the lessons are:
- High-profile failures may signal opportunity, not disaster.
- Watch how executives handle failure. Do they admit mistakes openly like Nadella?
- Look for companies with “failure labs” – autonomous labs, experiment budgets that embrace Bezos’ brutal math of taking a bet that has a 10 per cent chance of a pay-off of 100 times.
4. The bloom is off: the start of the DAT crash? – Andrew Walker
When I was writing my series ~a month ago, MSTR was trading for ~2x mNAV, and every company that announced a DAT [Digital Asset Treasury] deal with any crypto was seeing their stock price skyrocket.
Today, things have changed dramatically. Yes, you’ll still get an occasional squeeze on a buzzy deal in a company with a tiny float (see: OCTO jumping ~2500% on a worldcoin DAT strategy), but for the most part things have cooled down. Just take a look at the king of DATs: MSTR1 has traded down to ~1.5x mNAV….
… and the market seems to be looking at their strategy with increasing skepticism; most of their preferreds are trading below par (in the case of STRD, well below par), and, despite the drop in MSTR’s mNAV, MSTR has been forced to shift most of their capital raise to their ATM program in order to continue to buy bitcoin…
…And we’re already seeing formerly hot DATs need to pivot their strategy as their stocks trade below mNAV. For example, SBET has announced a share repurchase program as their stock slipped below mNAV, and they’re not alone. My favorite is Empery Digital, which announced a share repurchase program and had their CEO make an impassioned plea to shareholders about buying their stock to get discounted access to BTC…
…Despite the shareholder friendliness of the buybacks, I suspect they are a band aid on a bullet wound for most DATs.
Why?
Most of these DATs have fully deployed all of the proceeds they raised into their underlying assets. SBET, for example, has purchased over $3.5B of ETH and had just ~$72m in cash on their balance sheet at their last update; that’s a pittance versus their >$3B market cap…
…I think what’s really interesting about the bloom coming off DATs (the premiums fading away) is that it’s happened while crypto is still generally in favor. ETH is up ~70% over the past three months, while Bitcoin is up ~5%.
If DATs are starting to go out of favor will the underlying crypto is still doing reasonably well, what would happen if we hit another crypto winter and crypto prices traded down meaningfully?
And, if I might speculate a bit, if a lot of the recent rise in crypto has been caused by the huge rush of capital into DATs (which then gets deployed into the crypto, thus supporting the price), what would happen if that unwound for some reason? What if a bunch of DATs said “we’re trading at a discount to NAV; let’s practice good corporate governance, sell crypto, and buy our stock back (option 3 above)”? Or what if a bunch of DATs practice option 2 (leveraging crypto to buy back stock) and get margin called?
I suspect the underlying crypto could go a lot lower real fast as the same flywheel effect that’s sent crypto up recently unwinds.
5. What the Pentagon’s Rare Earths Deal Gets Right and Wrong – Tracy Alloway, Joe Weisenthal, Arnab Datta, and Peter Harrell
Rare earth elements and magnets manufactured from them are used across defense and industrial applications: An F-35 fighter jet, for example, requires more than 900 pounds of rare earths, and in cars they are used for everything from batteries to power seats. Apple uses a rare earth magnet in the iPhone’s “haptic” engine that makes a user feel buzzes and other vibrations.
China’s dominance of rare earths (it processes nearly 90% of rare earths globally) is relatively recent. For much of the 20th century the U.S. produced both rare earths and rare earths magnets domestically. Indeed, MP’s mine in Mountain Pass, California, located near Las Vegas, started production in 1952.
In the early 2000s, however, low-cost Chinese producers came to dominate global markets, driving most non-Chinese companies out of business: the Mountain Pass mine, for example, stopped operations in 2002. By the time it reopened in 2012, China had built a market infrastructure to dominate all aspects of the trade. The mine closed again in 2015. GM sold America’s leading rare earth magnet manufacturer to Chinese companies in the 1990s. By 2004 it, too, had shuttered U.S. manufacturing. Even after MP acquired the Mountain Pass mine and restarted operations in 2017 it exported most of its product to China to be processed and turned into magnets.
The Defense Department’s deal with MP Materials is designed to end America’s dependency on China with respect to two specific rare earths, neodymium (Nd) and praseodymium (Pr). In addition to expanding mining and processing of the raw metals, the deal is intended to build up America’s capacity to manufacture the metals into magnets, specifically neodymium iron boron (NdFeB) permanent magnets, one of the most important types of rare earths defense and industrial magnets…
…First, MP committed to expand U.S. mining, processing, and magnet manufacturing facilities. The company will increase mining and processing operations, including possibly in heavy rare earths; expand its existing magnet manufacturing facility in California to be able produce 3,000 tons of NdFeB permanent magnets annually (up from 1,000 tons annually currently), and construct a new “10X” facility in Texas that will enable MP to produce a total of 10,000 tons of magnets annually after 2028. Combined, the facilities should be able to meet a substantial portion of U.S. demand for NdFeB magnets, including all of our defense needs.
Second, DoD set a guaranteed price floor of $110 per kilo of MP’s NdPr products, running for 10 years. If the market price, currently below $60/kilo, remains below $110, DoD will pay MP the difference between the market and $110/kilo. If market prices exceed $110/kilo, DoD is entitled to 30% of MP’s extra profits. This ensures that MP can make money on its mining and processing operations even if it has to sell minerals below cost to compete with Chinese producers.
Third, DoD has guaranteed that either it or commercial buyers will purchase all of the 10X facility’s NdFeB magnets, estimated at 7,000 tons a year for the next decade. DoD will pay MP its realized cost of production of the magnets, plus $140 million per year to guarantee MP a profit, with a 2% annual inflation increase in the guaranteed profit figure. With DoD’s consent, MP can sell some of its magnets to commercial buyers, in which case, DoD will take the first $30 million in MP magnet profits exceeding $140 million. Additional profits beyond that will be split 50/50 between MP and DoD. Similar to the price floor, the magnet offtake agreement ensures that MP can profitably make magnets even if low global prices would undercut MP’s manufacturing…
…Beneath the deal’s ambition, its structure raises significant policy design questions. The first is a fundamental question about the extent to which the government (versus the private sector) should bear the costs associated with addressing critical U.S. supply chain risks. The MP deal essentially puts the U.S. taxpayer on the hook for developing a reliable U.S. supplier of rare earths and NdPrB magnets. And while the U.S. government can share in the upside if global prices for rare earths and the magnets exceed expected levels, if the price trajectory looks similar to the last decade, the U.S. government could be on the hook for billions. Potential costs include $1.4 billion in guaranteed profits for MP ($140 million per year, adjusted up at 2% per year). The price floor alone could cost billions over ten years if MP hits their announced capacity of 6,075 metric tons and prevailing market prices stay constant…
…The deal elevates MP Materials as America’s de facto magnet champion, despite having no track record of commercial success in magnet production. By contrast, China’s national champions typically emerge through fierce domestic competition. Firms like CATL did not rise to global leadership through political selection alone; they fought their way to the top by outperforming rivals on innovation and scale: CATL remains one of the top patent recipients globally while leveraging partnerships with major automakers like Tesla and BMW. Government support was structured to spur this competition. Subsidies and pilot programs were spread across multiple firms before consolidating behind the winners.
The U.S. decision to back MP sidesteps this competitive process, effectively granting a monopoly franchise in magnet production. This risks locking the U.S. into a suboptimal path if MP fails to deliver on cost or performance, while crowding out rivals that could prove more innovative…
…The deal also hardwires U.S. fiscal exposure to the same market infrastructure that China uses to determine prices and stifle investment in competitors. Under the price‑protection term, DoD pays the difference between $110/kg and a reference price — specifically the Asian Metal Market price. A substitute ex‑China Index is only allowed at DoD’s election and with the company’s consent. That means core cash flows for a decade depend on a benchmark whose prints reflect Chinese production costs, market structure, trade flows, policy choices, and tax treatment.
This creates three, compounding problems: (1) basis risk; (2) manipulability; and (3) path dependence. NdPr is sold as concentrate, oxide, and metal with varying specs, impurities, tenors, and delivery terms; Asian Metal quotations often embed VAT regimes, logistics premia, and buyer restrictions that diverge from U.S. realizations. Even with upside sharing, those mismatches can cap clawbacks in booms and invite arbitrage in busts. Relatedly, when public payments hinge on a single, quarterly external print, an actor with market power can manipulate spreads, restrict eligible buyers, or flood spot supply to push the index below U.S. breakevens and eat DoD appropriations. And locking federal contracts to Asian Metal deepens liquidity and legitimacy in that price‑discovery ecosystem. The U.S. ends up validating the very benchmark that concentrates market power abroad, raising the fiscal cost of preserving domestic capacity and making future decoupling harder.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Alphabet (parent of Google), Apple, Meta Platforms (parent of Facebook), Microsoft (its CEO is Satya Nadella), and Amazon (its founder is Jeff Bezos). Holdings are subject to change at any time.
