# The World Is on Sale, America Charges a Premium: Notes on Dave Iben

Source: Realpha Blog (blog.getrealpha.com)
Original article and charts: https://blog.getrealpha.com/en/blog/mebfaber-2026-09-25-dave-iben-the-world-is-on-sale-while-america-trade/

> Listening notes on The Meb Faber Show #652 (2026-09-25): Kopernik co-CIO Dave Iben on the US holding two-thirds of world market cap, commodities down to single-digit index weight, platinum's incentive price, and the night Korea's president tried to seize power. Educational notes and personal observation, not investment advice, and no stock recommendations.

Published: 2026-09-25
Locale: en
Tags: value investing, global allocation, scarcity, valuation discipline, podcast notes


![A mine stockyard at dawn, a line of ore cars running from the foreground deep into a far valley where glass towers catch hard light](/covers/mebfaber-2026-09-25-dave-iben-the-world-is-on-sale-while-america-trade-cover.png)

> Men, it has been well said, think in herds; it will be seen that they go mad in herds, while they recover their senses slowly, and one by one.
>
> — Charles Mackay, *Extraordinary Popular Delusions and the Madness of Crowds* (1841)

On The Meb Faber Show #652, released 25 September 2026, Kopernik Global Investors co-CIO Dave Iben points out that the United States accounts for two-thirds of global equity market value while producing roughly a fifth of global output, and that oil, copper and gold together still add up to a single-digit share of the indexes. He spent two months in Brazil this summer doing research and met zero people who were bullish on their own stock market. He states the condition on his own argument in the episode: you need five or ten years to settle the bet. Someone who bought his way in 1972 spent the next decade getting tortured and watching 75% of the money go.

![Three horizontal proportion bars run from long to very short: the top bar, "US share of global stock market value," fills two-thirds; the middle bar, "US share of the global economy," fills only one-fifth; and the bottom bar, "oil, copper and gold inside the index," is barely a stub.](/figures/market-cap-vs-economy-mismatch-en.svg)

## What the episode covers

Iben has been in the business for nearly forty-five years, runs over ten billion dollars, and has stamps from 108 countries. His habit now is to move the team into a market and live there for a stretch; this summer it was Brazil, twenty-some people for two months. Faber opens with a blunt question: in an age of digital abundance, does scarcity still matter?

The next hour runs from copper cash-flow models to platinum's incentive price, through the night Korea's president tried to take over, on to Chinese nuclear plants and Ukrainian black soil, and closes on whether AI makes human judgement cheap. One line holds the whole thing together: bubbles always happen around the real deal. What stayed with me is that he spends most of his energy not on where things are cheap, but on how people mis-model them.

## The points worth keeping

**One: commodities are the only industry discounted backwards.** Every other sector's DCF assumes the dollar loses value over time. In copper, oil and gold, Iben says, the model pins the dollar and then assumes the commodity falls in price every year. Once that assumption is fixed, the longer the reserve life, the cheaper they'll sell it to you. He has been harvesting that gap for years.

![A line chart with mining licence years on the horizontal axis and the future selling price assumed by the model on the vertical axis: one line rises to the right, showing that other industries assume the dollar thins out over time, while the other line falls to the right, showing commodities are assumed to sell for less each year.](/figures/reverse-discount-assumption-en.svg)

**Two: forty years of reserves, killed by a three-month delay.** A mine with forty years of copper underground announces production slips by three or four months, and the stock gets destroyed. His read is that not one speck of that forty years went away. That is mis-analysis, a different source of opportunity from the one that comes from emotion.

**Three: the platinum show of hands.** At a conference a year ago, after he laid out the case for platinum, someone asked the table to raise a hand if they liked platinum. Every hand went up. Then: raise your hand if you own platinum. Every hand went down — nobody among the thirty-odd people held any. At the V&A in London he saw a minerals exhibit whose panel still said platinum always trades at a premium to gold because it is scarcer; standing there, platinum needed to more than triple to catch gold. That morning he had spoken with the largest platinum producer, who put the price that gets someone to consider building a new mine at around 2,500, against 1,800 that day.

![A very long horizontal bar at the top stands for forty years of reserves, with a barely visible sliver cut off at its right end for a three-to-four-month delay; below, two share-price columns show the later one clearly shorter than the earlier one.](/figures/forty-years-priced-by-four-months-en.svg)

**Four: bubbles never form around junk.** The current line is that this time is different, AI is the real deal. His answer: canals, railroads, telephones, semiconductors, the internet — every one of them was real, every one is still with us, and every one produced a bubble. Money pours in precisely because people believe the thing is good. He was right in 1999 about Amazon and Microsoft; the price of being right was watching 90% of the Amazon position disappear inside two and a half years, and waiting until recently for Cisco to come back.

**Five: he was buying the night of the Korean coup attempt.** Two and a half years ago they could already buy good Korean companies at a third of book. Then the president tried to seize control, the military backed the democracy, he was taken away, and the crisis was over in days — but the market stayed down 25%. They took Korea to 19% of the portfolio, on unglamorous reasoning: Hyundai makes good cars, LG makes good products across many lines, Korea Telecom and LG U+ run businesses as sound as Verizon's and AT&T's, and he was paying two-thirds less. The market snapped back by a third, and when people asked why it rose so much he asked why it fell. Worth noting: much of the later run came from two semiconductor giants that now make up about half that market, and he owns neither.

![A dashed line sits high across the chart as the price needed to justify building a new mine, a solid column below falls well short of it, and the blank space between the two is labelled as the gap.](/figures/platinum-incentive-price-gap-en.svg)

**Six: falling prices grow the position, rising prices shrink it.** He appraises the business, then risk-adjusts for the country, the debt and the management: worth 50, adjusted to 30. At 20 there's 50% upside, so he buys a little. At 15, some firms would say the market is telling them they're wrong; his reading is that 50% upside became 100%, so he buys 2%. At 10 it becomes one of his 4% positions. Bought at 20, gone at 30, trimming all the way up. Low turnover in names, constant trimming and adding. Gold was once over 20% of the book and is down by two-thirds, with the money moved into platinum and copper.

**Seven: AI as servant, never master.** He cites an experiment by Peter Kaufman: ask AI for a bull report on Costco written from Charlie Munger's point of view, and out comes something thick, excellent and persuasive. Then ask for the bear case from Warren Buffett's point of view — also persuasive. Iben's conclusion is that trying to out-calculate AI with your left brain is a losing game, while having it grade your reports after you've decided what you think is workable. He tells his analysts their jobs are safe if they delegate the shallow analysis and keep the deep part. The line I wrote down: if everyone uses AI to find what everyone else is doing and then does the same thing, mediocrity is the best case.

## Going further

### "Cheap things stay cheap — am I just catching a falling knife?"

This is the pit I've fallen into most often. You buy an 80% discount, sit in it for three years, sell at a loss, and learn nothing, because cheapness by itself never tells you when it ends.

Iben splits cheap into two kinds, and the split is usable straight away. The first has a physical clock on it: platinum below the price that would justify a new mine means nobody adds supply, so supply only shrinks. That kind of cheap has a door that closes by itself — if the price doesn't rise, the material runs short. The second kind is emotional: Korea's 25% had nothing to do with what those companies earn, the political risk cleared within days, and the price stayed put.

Both are cheap, and their catalysts differ in kind. In the first, time works for you; in the second, you wait for the crowd. What lets him sit through it isn't conviction, it's the schedule — which price maps to which position size, written down in advance. Listening to this, I realised the names I failed to hold were the ones where I never wrote "at this price I own this much," so every decline turned into an improvised vote of feelings.

### "My money is spread across US large-cap tech — isn't that diversified?"

Two numbers from the episode answer it. The US is two-thirds of global market value and about a fifth of global output. Add the mechanism of cap weighting: the largest weights in an index are the positions that have had the most money thrown at them.

Korea makes the same point in miniature. Iben says half the recent run came from two semiconductor companies that are now about half that market. Buy the Korean index and you have bought two semiconductor names, with several hundred others as decoration. He went around the index and held telecoms and appliances at a third of book, carrying a risk profile with almost nothing in common with the index buyer's.

![A square stands for the whole market: its left half is filled by two large blocks of colour for two semiconductor names, while its right half is cut into many small grey cells for the several hundred other companies.](/figures/index-half-is-two-names-en.svg)

So the distance between "I hold five different US tech names" and "I have one bet on" is shorter than the account screen suggests. What I do now is check the common denominator once a month: if the same variable — AI capex, cloud budgets, memory prices — decides the fate of all five, it's one bet. No model needed; write the holdings on paper and note beside each one what it gets paid for.

### "I use AI for research every day — does that make me sharper or more average?"

The force of Kaufman's experiment is that AI will write a persuasive report for any position. Bring a position and it hands you ammunition. Bring none and it hands you the most mainstream position in its training data — the one everybody else also receives.

Iben frames it as deep versus shallow. I'd put it more plainly: AI is good at surfacing what has already been said, and cannot decide which sentence is worth betting on. He describes his own edge without decoration — when buying makes him sick to his stomach, he can still buy. That isn't an information advantage.

I changed how I use it. I write down my judgement and the thing that would overturn it, then tell the AI to attack. What comes back is a stress test. Ask it first what it thinks, and what comes back is consensus. Same tool; the order decides whether it helps you or decides for you.

## Sources worth your time

- The Meb Faber Show #652 (25 September 2026), interview with Dave Iben.
- Kopernik Global Investors' quarterly letters to investors. Faber notes at the end of the episode that they mix monetary theory with rock lyrics.
- Richard Cantillon on the non-neutrality of money (eighteenth century). Iben uses the river: pour water in at the top, the people near the top get plenty, those further down get it late and thinned, some never get a drop.
- John Templeton on "this time it's different" being the most dangerous words in the English language, and his history of buying Japan at three times earnings.
- Charles Mackay, *Extraordinary Popular Delusions and the Madness of Crowds* (1841), on how crowds come to believe one thing together.

## One thing to take with you

**Liking something and putting resources into it are two different acts.** Thirty-odd people raised their hands for platinum and not one of them owned any. That gap never hurts, because verbal support costs nothing, and you file "I support this" in memory as "I'm doing this." Every opportunity in this episode traces back to that gap — a room agreeing something matters while nobody pays for it.

![On the left, thirty filled circles stand for everyone who raised a hand to say they like it; on the right, the same grid of thirty circles is entirely hollow, showing that nobody actually owns any.](/figures/hands-up-versus-holdings-en.svg)

Here's something I tried that takes under ten minutes and has nothing to do with stocks. Take a sheet of paper. On the left, write three things you've been telling people lately that you care about — your body, a particular friend, something you want to learn; keep investing off the list. On the right, write the hours or the money you actually put in this month, real numbers, not "working on it." The sheet will tell you which one you care about and which one you raised a hand for.