Metallurgical Coal, Blast Furnaces, and the Physics of Inelastic Demand
A Deep-Dive Investment Memo on Warrior Met Coal (HCC) and Alpha Metallurgical Resources (AMR)
A Note on Mistakes, and Why This Memo Exists
Before getting into metallurgical coal, blast furnaces, or company specifics, I need to start with something uncomfortable.
I made a mistake.
Early in my journey as a value investor, I acted like, as Charlie Munger would say, a horse’s ass.
In early 2024, I purchased Mohnish Pabrai’s position in Alpha Metallurgical Resources (AMR) after seeing it disclosed in one of his 13F filings. I did not understand the business. I did not understand the industry. I did not understand the cost curve, the cyclicality, the failure modes, or the capital allocation dynamics.
I saw a great investor buying - and I followed.
That was the entirety of my process.
I purchased AMR around ~$275 per share. I then rode it all the way down into the ~$105 range. Along the way, I made decisions driven by price action, not understanding. I eventually tax-loss harvested half of the position around ~$150 because I needed to shore up capital for a genuinely outstanding opportunity elsewhere.
Every decision I made in that sequence was reactive. None of it was grounded in real analysis.
As AMR rallied sharply through 2025, something shifted. Instead of feeling vindicated or frustrated, I felt exposed. I realized that this security had become a mirror - not of market volatility, but of my own lack of preparation.
If I had done the work two years earlier - the work you are about to read - I would have recognized mid-2025 as a once-in-a-cycle opportunity. I wasn’t unlucky. I was unprepared.
That is on me.
This memo exists because of that mistake.
It is not an attempt to justify a past decision. It is an attempt to stress-test my own decision-making process, to rebuild it from first principles, and to ensure that I never again outsource understanding to reputation.
I will never blindly clone a trade again.
Not from Mohnish.
Not from anyone.
The lesson was expensive (in both type 1 and type 2 errors) - but it was invaluable.
The analysis that follows is the result of that lesson. It is what proper preparation looks like. And next time the opportunity comes - because it always does in cyclical industries - I will be ready.
What follows is my full investment memo on metallurgical coal and two companies at the center of the cycle: Warrior Met Coal (HCC) and Alpha Metallurgical Resources (AMR).
Executive Summary
Metallurgical coal is not misunderstood because investors are lazy.
It is misunderstood because people apply financial logic to a system governed by physics, thermodynamics, and sunk capital.
This memo argues four things:
Blast furnaces (BF/BOF) are fundamentally different from other industrial assets - once built, they must run.
That physical constraint creates forced demand for metallurgical coal, regardless of price, for decades.
Met coal pricing is therefore set by marginal supply destruction, not demand growth.
Equity returns in this space come from capital recovery, not terminal valuation.
We apply this framework to two U.S. producers:
HCC: a structurally durable, low-cost harvester
AMR: a high-torque, path-dependent option on price spikes
Part I - Why Blast Furnaces Are NOT “Factories”
A blast furnace is a chemical reactor, not a plant
A blast furnace is a continuous, high-temperature reduction vessel where iron ore is chemically reduced into molten iron using coke derived from metallurgical coal.
Key facts:
Internal temperatures exceed 2,000°F
Iron is molten inside the furnace at all times
The process is continuous - not batch
If a blast furnace stops:
Molten iron solidifies
Internal linings crack
The furnace is physically destroyed
Restarting requires:
Complete teardown
Refractory relining
Months (often years) of downtime
Hundreds of millions in capital
This is not an economic choice.
It is a binary physical outcome.
Part II - The Cost and Irreversibility of BF/BOF Capital
What blast furnaces cost
Modern BF/BOF complexes cost:
$1.0–1.5 billion to build new
$300–600 million to reline (every ~15–20 years)
These assets have:
40–50 year design lives
Massive foundations
Coke ovens and sinter plants integrated on site
Rail, port, and power infrastructure built specifically for them
Once built, the capital is irrecoverable.
The global blast furnace fleet is young
Using global tracking data (e.g., Global Energy Monitor Blast Furnace Tool):
China built most of its BF capacity post-2000
India’s fleet is even newer
Europe and Japan are older, but relining instead of retiring
New BF capacity is still being commissioned globally
This matters because:
A young BF fleet means decades of forced utilization ahead.
Steelmakers do not shut down billion-dollar assets because prices are weak for a year or two. They feed them.
Part III - Why BF/BOF Cannot Be Replaced Quickly by EAF
The EAF narrative is incomplete
Electric Arc Furnaces (EAFs) are real:
Cheaper to build ($200–500M)
Flexible
Lower emissions
But they are constrained by inputs, not capital.
Scrap steel is the binding constraint
EAFs require scrap. Scrap supply depends on:
Steel made 30–50 years ago
End-of-life buildings, bridges, autos
Reality:
The world does not have enough high-quality scrap to replace BF/BOF
Scrap quality matters (auto steel, electrical steel, bearings)
Contamination ruins downstream products
Even aggressive forecasts cap EAF share at 50–60%.
That leaves 40–50% of steel demand structurally tied to BF/BOF.
Electricity is the second constraint
EAF cost structure:
15–30% of total cost = electricity
Requires stable, cheap baseload power
Competes with AI data centers, grids, and households
In a world of rising electricity demand:
EAF economics are not guaranteed to dominate.
Part IV - Why Metallurgical Coal Demand Is Forced
Putting it together:
Blast furnaces must run once built
Blast furnaces require coke
Coke requires metallurgical coal
BF/BOF assets last 40–50 years
The global fleet is young
Therefore:
Metallurgical coal demand is structurally forced for decades, even if volumes decline slowly.
Demand does not disappear.
It erodes linearly, not exponentially.
Part V - The Cost Curve and the Marginal Ton
Why price behaves violently
Because demand is forced, price adjusts via supply destruction.
Low-cost producers always run
High-cost producers shut when prices fall
When enough high-cost supply exits, price snaps back
Through-cycle (my conservative underwriting):
Marginal cost ≈ $170 FOB
Normalized price ≈ $180–220
Below $170 → supply destruction
Above $245 → stress pricing
This is why:
Prices spike fast
Prices cannot stay low forever
Duration matters more than price level
Part VI - Applying This to the Companies
Warrior Met Coal (HCC)
Structural position
Concentrated Alabama basin
Longwall mining
Blue Creek adds scale and lowers cost
World-class geology
Through-cycle economics (conservative)
Cash cost (FOB): ~$105–115
All-in sustaining: ~$135–150
Normalized after-tax FCF: $350–450M
Upside mid-cycle: $500–600M
Key point
HCC survives below the marginal ton.
Time works for the equity.
Mental model
A self-liquidating annuity that should retire shares steadily.
Alpha Metallurgical Resources (AMR)
Structural position
Appalachian portfolio
Mix of underground/surface mines
Higher operating leverage
Greater flexibility, higher volatility
Through-cycle economics
Through-cycle volume: ~12–12.5M tons (Peak Cycle ~14M+)
Realized price: ~$155–160
All-in economics: ~$150–170
Normalized after-tax FCF: $200–250M
Peak-cycle behavior
At $300–350 FOB:
Idle mines return
Margins explode
After-tax FCF can exceed $1B (and has already done so in the past)
Key point
AMR out-earns HCC only at peaks.
Time works against the equity.
Mental model
A loaded spring - incredible torque, fragile timing.
Part VII - Why EPS and Terminal Prices Mislead
A crucial mistake:
“If a company generates tens of billions of dollars in cumulative free cash flow and retires most of its shares, then earnings per share must explode - and the stock price should follow.”
False.
Why?
Remaining reserves shrink
Future cash flows are uncertain
Terminal multiples compress (often 5–7× or less)
Markets price future, not past cash
Buybacks are distributions, not magic.
The return is:
How much capital you recover before decline matters.
Part VIII – Capital Recovery Framework (The Only One That Works)
HCC (Durability Asset)
HCC is a structurally advantaged, low-cost producer that can operate profitably below the marginal ton and generate meaningful free cash flow across most of the cycle. Its cost position and operational stability allow buybacks and dividends to persist even when pricing softens.
Because of this durability, I require:
12% capital recovery
Conservative normalized after-tax FCF: $350–450M
Under those assumptions:
Buy zone: approximately <$75–85
More attractive below $80
High-conviction below $70
HCC is not a timing trade. It is a business I am willing to own through time - but only at a price that allows meaningful capital recovery before terminal decline becomes relevant.
Watchlist asset.
AMR (Torque Asset)
AMR is fundamentally different.
While it has demonstrated exceptional cost discipline, its portfolio is higher on the cost curve, more path-dependent, and more exposed to prolonged weak pricing environments. When prices are strong, AMR behaves like a coiled spring. When prices remain suppressed, time works against it.
Because of this structural fragility, I require:
15% capital recovery
Normalized after-tax FCF: $200–250M
Under those assumptions:
Buy zone: approximately $110–135
Attractive only during dislocation
Requires visible cycle stress
AMR is not built to be owned through long stretches of weak pricing. It is a volatility instrument that must be purchased only when the market prices in failure.
Pass unless dislocated.
Part IX – Peak Cycle Overlay (Why AMR Exists)
At true peak-cycle conditions - roughly $350+ per ton FOB, a level the market has reached before - the difference between Warrior Met Coal and Alpha Metallurgical Resources becomes extreme.
In that environment:
HCC is capable of generating approximately $1.2 billion of after-tax free cash flow on annual volumes of roughly 10 million tons, translating to roughly $20–25 of free cash flow per share given its larger share count.
AMR, by contrast, can generate closer to $1.4 billion of after-tax free cash flow on annual volumes approaching 15 million tons. With only ~12.9 million shares outstanding, that equates to well over $100 of free cash flow per share.
This stark divergence explains why AMR exists as an equity at all.
It is not built to be owned through time, nor is it designed to survive prolonged weak pricing without stress. Instead, it functions as a high-torque expression of the cycle itself - an asset that looks mediocre or fragile in normalized environments, but becomes extraordinarily powerful when marginal supply is forced offline and prices spike.
HCC, by contrast, remains a cash-generating business even at lower prices. Its upside is intentionally muted by durability and discipline.
Peak-cycle economics do not define value.
They define optionality.
This is optionality - not value.
Final Takeaway
Metallurgical coal is governed by:
Physics
Sunk capital
Time
Not ESG slogans.
Not quarterly earnings.
The discipline is not in owning these businesses.
The discipline is in waiting for price to reflect their reality.
HCC: buy durability on volatility
AMR: buy torque only when the market panics
As Warren Buffett would say:
“The market is there to serve you, not instruct you.”
In met coal, the market regularly panics.
The job is to be ready when it does.
Disclaimer
This memo reflects my personal investment framework and opinions. It is not investment advice. I may be wrong, and circumstances can change. I reserve the right to change my mind as new facts emerge.
 | A guest post by
|