Actionable insights — These Commodities Are Mind-Bogglingly Underpriced

32:35 1. The NPV "free-warrant" screen — buy the long-tail reserve years for free

1. Value a resource company on the net present value of its proven reserves: project the cash flows at current and forecast commodity prices, then discount at 8%.
2. Note where the discount zeroes out the future: at 8%, any cash flow past year ~11–12 has no present value — so a 30-year reserve life means the last ~18 years are valued at zero in today's price.
3. Buy when the market price sits at or below that truncated NPV — you then own four stacked "free warrants": the back reserve years, the exploration upside, the commodity-price upside, and (newly) the AI-efficiency upside.
4. Re-run the same calculation in five years: you get the same NPV answer, having banked five years of cash flow "for nothing" — confirming it's a time game, not a price call.

• Any reserve-rich producer trading below the 8%-discounted NPV of proven reserves; long reserve lives (20yr+) maximize the "free" back-end.

9:45 2. The "three sins" floor — past underinvestment + future demand + a falling dollar

1. Tally sin #1 (supply): years of exploration/production underinvestment that can't be reversed inside the supply-response lead time.
2. Tally sin #2 (demand): structural future demand — growth, lifting the world's poor, plus the AI/electrification overlay — layered on top of existing demand.
3. Tally sin #3 (the unit of measurement): commodities are priced nominally in dollars; a debasing dollar raises nominal prices even with flat real demand (the 1970s dollar fell ~75% in real terms).
4. Conclude that all three push the same direction — a nominal price floor — so the only question is timing, not direction.

• Commodities where all three sins line up; treat dollar debasement as a standalone tailwind, not a footnote.

6:56 3. The supply-lag / ration-by-price test — price the lead time, not the headline

1. Estimate the supply-response lead time end-to-end: exploration to first success (~10yr statistically), +3yr to drill the deposit off, +3yr to permit/build — i.e. 16–17 years for copper.
2. Compare it to the demand timeline: if demand arrives well inside the lead time, supply cannot respond and the market clears by rationing-by-price.
3. Identify the only off-ramp: a demand collapse (synchronized global depression). Absent that, the price must rise — so size the bet to your view on recession odds, not on supply optimism.

• A demand timeline that falls inside the supply lead time; recession/depression as the single thing that voids the thesis.

38:55 4. Play big markets, not small ones — a single deposit can't "crack" a copper market

1. For a thematic commodity bet, screen on market size first: pick a deep market (energy, copper), not a thin one (titanium, vanadium).
2. Reject thin markets even when the story is right: one big new deposit or a technology/supply surprise floods a small market and "you get your ass kicked."
3. Accept that big markets move slower but are durable — supply can't be conjured, so the theme survives a single mine coming online.

• Market depth vs. single-deposit risk; the right theme in a thin market is still a trap.

40:20 5. The energy-density / energy-security screen — what survives a supply shock

1. When energy security re-enters the conversation (a supply shock, an embargo, a chokepoint), ask which fuel a resource-poor nation can actually stockpile.
2. Score candidates on energy density: you can't warehouse enough coal, oil, LNG, wind or sunshine — uranium's density lets one warehouse hold years of national power.
3. Cross-check the demand side: AI needs 24/7 (not intermittent) and non-carbon power — uranium uniquely satisfies both — so the security buyer and the AI buyer converge on the same molecule.
4. Use history as the analog: the 1973 Arab oil embargo built the French (#4) and Japanese (#3) nuclear fleets — energy insecurity, not climate, drove the last great build-out.

• Energy-security rhetoric returning after a supply shock; the fuel that is both storable (dense) and dispatchable (24/7) is the structural winner.

1:01:06 6. The capital-allocation tell — buy the producer that reinvests, not the one that cannibalizes

1. Read the capital-allocation policy: is free cash flow going mostly to buybacks/dividends (subsidizing today's shareholders) or into sustaining + new production for 3–5 years out?
2. Treat heavy buybacks during a structural-shortage thesis as "cannibalizing the company" — borrowing future output to flatter the present.
3. Use the analyst-pressure signal as the macro tell: the message has landed only when analysts start telling companies to cut payouts and reinvest in production. Until then, the supply shortage stays intact (bullish the commodity).
4. Prefer the rare disciplined reinvestor as the single-name expression.

• The day analysts pivot to "cut buybacks, grow production" — that's the top tell for the commodity and bottom tell for the producers' multiples; until then, hold the disciplined reinvestor.

24:54 7. Use AI as an analyst force-multiplier — but constrain the database and the question

1. Point AI at large, well-bounded datasets (years of financials; thousands of well logs / seismic / completions) to surface correlations and "coincident anomalies" no human can compute by hand.
2. Constrain the inputs: feed it only the right database — "if you don't constrain the database, the results are worse than useless."
3. Know the question first: "if you don't know the questions to ask, AI stands for artificial ignorance." Domain expertise sets the prompt; AI does the grind.
4. Convert the time saved into edge: the six-week by-hand comparative-financials job becomes a two-minute task, freeing judgment for the parts that need it.

• Companies (and your own workflow) sitting on large proprietary datasets that AI can mine; the moat is the constrained data plus the right question, not the model.

53:33 8. Fade the hyperbolic chart — vertical moves don't end by going sideways

1. Identify a near-vertical "hockey-stick" move (up or down) — the kind that resolves "almost always."
2. Bet against it symmetrically: sell into hyperbolic up-moves, buy into hyperbolic crashes. "The back side of a hockey stick is just as steep as the front side."
3. Require a fundamental confirmation alongside the chart before acting, and accept being early/unpopular — "whenever I do something immediately unpopular, I'm invariably right."

• Parabolic price action as a sell trigger for longs and a capitulation crash as the mirror-image buy trigger; pair it with a fundamental check.

55:25 9. Save vs. speculate, and rotate risk by what the crowd is selling

1. Separate a price-insensitive savings bucket (gold, bought on every liquidity event) from a speculation bucket you actively size for return.
2. In a risk-off market, lean into the riskiest names — they fall the most and indiscriminately — rather than the safest, reversing the prior "biggest and best" stance.
3. Gate the rotation on improving fundamentals: only step up risk when the operational data is getting better (e.g. spectacular drill results after years of exploration spend) even as prices fall.
4. Deploy new money rather than rotating out of the majors — change the risk profile of incremental capital, not the core.

• Indiscriminate risk-off selling that drags quality down with junk, plus improving fundamentals — that combination is the cue to up risk with new capital.

1:01:37 10. Defer outside your competence — name a better expert instead of opining

1. Define what you can actually value (for him: natural-resource companies; the rock, not the market).
2. When a question lands outside it — market-correction size, oil-inventory dynamics — decline and explicitly hand it to a better-qualified expert rather than manufacture a view.
3. Still contribute the piece you do own: e.g. "the cure for high prices is high prices," and the marginal (poorest) buyer sets the price.

• The urge to opine outside your toolkit — treat it as the signal to defer and name the better source, not to stretch.