Warrior Met Coal (HCC)

One Benchmark, No Basket

Warrior Met Coal sells exactly one product: metallurgical coal from its Alabama longwall mines, shipped to blast-furnace steelmakers and priced off the Platts premium low-vol (PLV) FOB Australia index. That makes it the simplest revenue model in the steel value chain. A steelmaker earns a spread, steel price minus a raw-material basket, so two or three marks have to cooperate before the income statement moves. Warrior has no basket. Its costs are labour, equipment and the mines themselves; the PLV mark alone decides the top line.

In FY2025 the company sold 9.6M short tons against production of 10.2M st. The quality mix was 54% premium low-vol and 46% High Vol A, and that blend realised roughly 80% of the PLV index on a gross basis. The quality ladder matters here: premium low-vol tracks the benchmark; High Vol A trades at a discount. What hits the P&L is the weighted blend. Our met coal vs thermal coal guide works through that ladder and why coking coal prices off steelmaking value rather than energy content.

Keep two price lines separate when you model this. The ~80% figure is the gross realisation against the index. The net realised price, the one the income statement sees, was $132.62/st in FY2025 after freight, demurrage and other selling deductions. Taking 80% of the PLV mark and stopping there overstates revenue; the gap between gross realisation and net price is real money on 9.6M tons.

There is also a unit to convert before the two lines can meet. PLV is quoted in US dollars per metric tonne; Warrior, like every US producer, reports short tons, and a short ton is about 91% of a tonne. Run the chain in order: at the $200/t planning benchmark, 80% is $160 a tonne, which is roughly $145 per short ton before the selling deductions come off. Subtract a short-ton cash cost straight from a per-tonne index and the margin is wrong twice, once for the quality mix and once for the unit.

What the Trough Looks Like

At the cycle low Warrior still earned $31/st of margin and $256.5M of adjusted EBITDA. The net realised price was $132.62/st against cash cost of sales of $101.30/st (quoted free-on-board port), a spread of $31.32/st. That spread is not what reaches the EBITDA line: head office and selling costs still come out of it, which is why 9.6M tons at $31.32 gives about $301M of margin over cash cost but $256.5M of adjusted EBITDA, roughly $27 per ton sold.

That is the same per-ton lens our EBITDA per tonne guide applies to the steelmakers, where the FY2025 ladder ran from Nucor at ~$210/ton down to Cleveland-Cliffs at ~$2/ton. Warrior does not belong inside that ladder, because a ton of steel sells for several times a ton of coal and per-ton profit only compares within a product. Measured against its own realisation, $27 on $132.62 is a fifth of revenue, printed at the bottom of the met coal cycle rather than through a structural cost problem.

A thin margin cuts both ways, and 2026 has shown the upside half. A fall of about $31/st in the FY2025 realised price, less than a quarter, would have taken the margin to zero. Equally, every $10/st of price recovery drops almost straight through to EBITDA, because the cost base barely moves with the benchmark: worth roughly $95M on FY2025 volumes and nearer $135M on the 13-14M tons guided for 2026. In the first half of 2026 Warrior sold 6.7M st at $143.04/st against a cash cost of $94.17/st, and adjusted EBITDA of $300.3M in six months already beat the whole of FY2025. Higher volumes did most of that; a better price and a lower unit cost did the rest.

Blue Creek: Capitalise the Ramp, Not the Nameplate

Blue Creek is the growth case, and unusually for a mining growth story it has largely arrived. The longwall began commissioning in October 2025, eight months ahead of plan and on budget; development spending is now complete and the mine has driven four consecutive quarters of record volumes. Warrior sold 3.7M short tons in the second quarter of 2026 alone, against 9.6M in the whole of FY2025, and raised full-year sales guidance to 13.0-14.0M st. Nameplate is 6.0M st/yr for the mine and roughly 14-15M st/yr for the group.

That still leaves the classic mistake available: multiply 14.5M tons by a margin and capitalise the result. Nameplate is capacity, not sales, and the company's own guidance sits below it. FY2025 showed a 0.6M st gap between production and tons sold, and ramp tons arrive at higher unit cost than steady state before settling down. They also ship into whatever PLV is doing at delivery, not today's spot or last year's trough. Price the ramp year by year at the planning price and treat nameplate as the ceiling.

Net Cash Through the Trough

Warrior finished FY2025 with roughly $105M of net cash. Read that alongside the cycle position: it funded a billion-dollar mine through to longwall commissioning while prices sat at the trough, burning $173M of free cash in the year, and still ended it with no net debt. It was still net cash at the half year. The trough did not threaten the company, so shareholders got to own the ramp.

We screen leverage at under 1.0x as conservative, 1.0-2.5x as normal and above 2.5x as stretched. Warrior is not just under the conservative line; it is on the other side of zero. In a business that was earning $31/st at the trough, a bad quarter away from breakeven, the absence of debt service is what converts price risk into a survivable waiting game rather than a solvency question. Cleveland-Cliffs shows the alternative: $7.2B of net debt against $37M of FY2025 EBITDA.

What to Watch in the Financials

Blue Creek tons and their cash cost. The growth case is volume at acceptable cost, nothing else. Commissioning tonnage usually runs expensive, so the test was whether the new mine pulled the group cost figure down or up. So far down: cash cost of sales fell from $101.30/st in FY2025 to $94.17/st across the first half of 2026, and guidance for the year is $95-105/st. Keep watching that line, because it is the half of the margin management can actually influence.

Realisation against PLV. The ~80% blended gross realisation is a mix outcome, not a constant. Blue Creek's coal is High Vol A, so its tons pull the blend away from the premium low-vol end of the ladder and the realisation percentage should drift down as they arrive, with no move in the index at all. The gross-versus-net gap moves separately, with freight and demurrage. Recompute the realisation each quarter rather than carrying 80% forward on autopilot.

The benchmark itself. PLV futures at $246/t (9 June 2026) sit well above the level implied by FY2025's realisations, and our planning price is $200/t. Where the mark settles decides whether the ramp lands into a tailwind or into more trough.

Inventory. Production ran 0.6M st ahead of sales in FY2025. Watch whether that build clears as Blue Creek volumes arrive or keeps growing; unsold premium coal is deferred revenue at best and a demand signal at worst.

Key Risks

Single-benchmark exposure. Everything keys off seaborne premium coking coal. There is no second commodity, no downstream spread, no hedge in the portfolio. A third off the first-half 2026 realisation takes the margin to zero, and at FY2025's thinner spread it took less than a quarter. The company has limited cost levers to pull against either.

The customer is the blast furnace. Met coal demand is BOF steelmaking demand. The global route split moved to 69.4% BOF in CY2025 from 70.4% a year earlier, a slow structural drift toward electric-arc steelmaking rather than a cliff, but it is the terminal-value question every met coal valuation has to answer. Premium low-vol coal is the most defensible part of the demand curve; it is not exempt from it.

Sustaining the ramp, not starting it. Eight months early and on budget was a strong start, and the volumes have since come through. Commissioning a longwall is not the same as running one at 6.0M st/yr for a decade, though. Longwall moves stop production while they happen, and Warrior has three of them scheduled across the second half of 2026, so quarterly volumes will not step up in a straight line.

How much of the recovery is already in the price. This is the harder question now the ramp has delivered. At a market cap near $4.4B in early August 2026, less net cash, the enterprise is capitalised at roughly seven times a first-half 2026 EBITDA run rate, against about seventeen times the FY2025 trough. Neither denominator is the right one to value on. A trough year understates the asset and a strong half-year overstates it, which is why the honest exercise is a mid-cycle benchmark price run through the full volume book. Net cash protects solvency in the next trough; it does not tell you what the ramp is worth.