Hallador Sold $700 Million of Power at Record Prices. The Market Sold Hallador.
One year ago this week, Hallador Energy printed its all-time high of $23.81. On Monday it reported what is unambiguously the best commercial quarter in its history — a third large Merom contract at record pricing, roughly 95% of accredited capacity sold through 2035, a forward book near $3 billion — and the shares sit 57% below that peak, down 35% in ninety days. At a $479 million market cap, the company is now worth roughly half the $800 million it says it is spending to build Turtle Creek. That gap, not the earnings beat or miss, is the story.
The Contract Machine Is Working
The October deal is the third of the year, following March and May, and each was priced above the last. Brent Bilsland described roughly $700 million of capacity and energy over six years beginning in 2029, with combined pricing north of $80 per megawatt hour — more than 20% above March. Blended realizations step from $46/MWh for the remainder of this year to $59 next year, $70 in 2029 and $75 from 2031 through 2035, with capacity prices rising from $262 per MW-day in 2027 to $478 in 2029. That is the whole thesis in two numbers: capacity is scarce, and Hallador owns some of it in a state that welcomes load.
What is genuinely new this quarter is the architecture of the contract, not just the price. The deal is unit contingent — if a unit is down, Hallador does not have to bid into the day-ahead market — carries a fuel floor with recovery of excess fuel cost, required no regulatory approval, and, as Bilsland emphasized, has no margin call provisions. Roughly $420 million of the $700 million is estimated unit-contingent energy rather than fixed commitment. This is an unusually defensive way to sell an unusually strong market.
I want to be very clear about one point. We sold to an investment-grade utility, not a data center. Data centers may be driving nearly all the growth in utility load, but our counterparty is an investment-grade utility that serves its whole territory, and it pays us whether any particular data center gets built or not.
Note the promise-keeping. On the August call, Bilsland said “we really think that we will add to the contracts that we've already put in place this year before the year is out” — Brent Bilsland, Chairman and CEO · 2026-08-10; the October deal delivered it. Back in March he had forecast “multiple bites at the apple” — Brent Bilsland, President and Chief Executive Officer · 2026-03-12. Three bites, each richer. Management's forward-sales rhetoric has been one of the more reliable things in this name.
Why the Tape Refuses to Pay For It
Hallador's equity is being marked by a theme it does not actually sell to. Across the market, the AI-power complex has been de-rating hard: AI data centers is among the worst-performing baskets over 90 days, down more than 20% with dozens of decliners, data center capacity is off roughly a third, and even the paper that these deals are written on is out of favour — long term agreements shows a lopsided decline across 28 negative tickers versus six positive. Bilsland aimed straight at that fear, invoking the "Neo cloud" companies that buy computing hardware with borrowed money and rent it out, and drew the line: they are not our customers and not our credit risk.
The irony is that the demand side keeps confirming. In the last five days alone, Black Hills talked up large load growth tied to a planned Google data center, DTE flagged data center development opportunities, and NRG described a multi-gigawatt data center project. Utilities are still building load forecasts around hyperscalers. Hallador sells to exactly those utilities, in front of the meter, in a state that is not pausing build-outs.
Turtle Creek: An $800 Million Asset Priced at Zero
Management's central claim is that the equity ascribes little or no value to Turtle Creek. The math offered: roughly $800 million of project cost, targeted at approximately $1.6 billion of value in about 24 months, built from four $200 million steps — equipment acquisition, permits and entitlements, contracts, and commercial operation. Bilsland says the company is “rounding third on a couple of these steps” — Brent Bilsland, Chairman and Chief Executive Officer · 2026-11-09: the turbines have left the Philippines, title has transferred, the air permit is filed, the MISO study is complete and the interconnection agreement is being finalized. The external anchor is TransAlta's roughly $1 billion purchase of 318 MW of fully contracted gas peakers in June — a mark implying far more per kilowatt than Hallador's stated ~$1,700/kW build cost, if it holds.
The sceptic's answer is that no offtake agreement is signed, and the term loan that funds the build is expensive paper. The $600 million construction bridge ($550 million funded) is structured so lenders earn a minimum return on repayment, and interest accrues rather than pays. Todd Telesz was blunt: “the reported losses will likely widen before the plant runs” — Todd Telesz, Chief Financial Officer · 2026-11-09. The stated path to avoid dilution is refinancing once contracted, a partial sale, or both.
The Other Side of the Ledger
None of this is free. The last filed quarter showed operating income of -$16 million on $102 million of revenue, with free cash flow of -$52 million. Capital intensity is the tell: capex doubled to $57.1 million in Q3 from $26 million, and the net cash position of -$10 million is a pre-bridge snapshot; bank debt is now $550 million against $329.8 million of restricted cash and $103.2 million of total liquidity.
Operationally, Q3 was mixed. Generation rose 77% sequentially to 1.4 million MWh and availability hit 80% in August and September versus an industry norm Bilsland put at “something like 67%” — Brent Bilsland, Chairman and Chief Executive Officer · 2026-11-09 for coal. But purchase power costs rose $9 million, third-party coal tons fell roughly 400,000, and Sunrise's cost per ton climbed to $52.17 from $42.74 — the coal fuel chain is inflating just as Merom is being called to run more.
The bet is straightforward, and unusually binary: a company worth $479 million is telling you it will convert $800 million of spend into $1.6 billion of contracted asset value within two years, funded largely with debt and a pledge of no dilution, while its stock trades as if the plant is worth nothing and the AI demand curve has broken. Bilsland's own hedge is the honest one — “the market will either value this at full value or somebody else will” — Brent Bilsland, Chairman and Chief Executive Officer · 2026-11-09. Everything between now and the second half of 2028 is a race between long term contracts and commercial operation, and the market is currently pricing the gap, not the goal.