American Bitcoin's 49% Mining Margin Is Real — and Still Left Shareholders Down 95%

American Bitcoin makes a claim that sounds like a license to print money: it can mint bitcoin for roughly half of what the coin sells for. The pitch belongs to Eric Trump, co-founder of the company (and the president's son), and to Hut 8, the mining firm that supplies and runs the machines. It was put into specific numbers at the venture's launch in March 2025 — up to 13 bitcoin per day in production, at a 49% operational margin, on the way to a targeted fleet of more than 50 exahash. If you can produce a coin for thirty-something thousand dollars and sell it for seventy-something, the logic goes, the factory is a cash machine.

That factory math is genuinely good. That's the part the headline gets right — and the part you can actually verify.

In the quarter that ended June 30, the company says it mined a record 932 bitcoin, roughly ten coins a day on average, at a stated cost of about $36,500 per coin — down from nearly $47,000 at the end of 2025. With bitcoin near $78,000 today, a ~$36,500 cost is just under half of spot, which tracks the line Eric Trump used in a November interview, that the company mints "roughly 50 cents on the dollar to the average to the spot price of BTC." The company reported a mining gross margin near 50–52% in both quarters of 2026, running a fleet of roughly 89,000 machines at about 25 exahash of operational capacity.

Here is where the story needs a careful eye. That margin is measured on a slice of the business, not the whole of it.

American Bitcoin lost $57.2 million in the second quarter and $81.8 million in the first — three consecutive losing quarters. A "mining gross margin" counts revenue from freshly minted coins minus the direct cost of producing them: power, hosting, running the machines. It leaves out the machinery itself, depreciation on it ($28.2 million in one quarter), selling and administrative costs, and the biggest swing item of all — the mark-to-market on the stash of coins the company already holds. When bitcoin fell through the spring, that book adjustment alone produced a $71.2 million loss on digital assets in Q2. So "a 50% mining margin" and "a $57 million loss" are not contradictory; they are measurements of two different things. One describes the mint. The other describes the whole company, whose quarterly result has been decided by the price of its hoard, not by the efficiency of its miners.

The hoard is the second thing the "low-cost miner" framing quietly manages. Roughly 8,000 bitcoin sat on the balance sheet at the end of June, and it is real money — around $620 million at today's prices. But most of those coins were not mined. Forbes' analysis put about 70% of the treasury's growth down to purchases funded by selling new stock, not to the mining fleet. The first 27 days after listing, the company sold 11 million shares for $90 million; between January and late March of this year it sold 84 million more for $111 million, buying about 1,430 bitcoin with the proceeds. Every one of those sales grows the giant pile of coins and thins the claim on them of everyone already in. The company publishes the honest version of this arithmetic — the "satoshis per share" metric — and in Q2 it grew a modest 11%, because the coin count rose faster than the share count, but not by much.

There is a third line the stacking narrative skips: a meaningful share of that celebrated treasury is already pledged away. Of the 8,002 coins held on June 30, about 3,090 — roughly 39% — were pledged as collateral to Bitmain under miner-purchase agreements, sitting against a $371.7 million non-current liability. The company can pay cash before each redemption window closes, roughly 24 months out, and keep the coins; if a window lapses, the coins pay for the machines and drop off the balance sheet. Forbes estimates the biggest tranche was pledged near last year's price peak and could be forfeited if bitcoin has not recovered by around 2027. The headline counts those coins as stacked. The contract counts them as owed.

Now map who holds which role, because the people who run this machine monetized it differently than stockholders did. Hut 8 contributed its mining fleet, took roughly 80% of the entity at launch, and still appoints a majority of the board. It built American Bitcoin expressly as a majority-owned public subsidiary that would "raise its own capital" — in other words, the parent split the miners off so the miners would draw their own funding, while Hut 8 itself pivoted toward AI data centers, a business so well received that Hut 8's own stock is up sharply this year. Eric Trump reportedly grew his personal fortune from around $190 million to $280 million along the way. Public shareholders, by Forbes' estimate, are down about $500 million combined. This is not a mining company whose chief executive happens to carry the family name; it is a Hut-controlled treasury machine with a Trump-branded sales front.

The verdict that follows from the record is uncomfortable for the pitch but clarifying for the investor. The margin claim was substantially true — American Bitcoin really is a cheap producer, and cheap production is the reason it kept operating through an industry downturn while other miners bolted for AI. But the margin never reached the shareholders. The stock peaked in the days after its September 2025 listing, when investors briefly valued the company near $13.2 billion against roughly $270 million of bitcoin on the books; it now trades around a dollar, down roughly 95%, and needed a 1-for-15 reverse split in July just to keep its listing, after which the shares sank again.

So treat the headline as the start of a valuation, not the answer to one. What an American Bitcoin buyer actually owns is a leveraged, heavily diluted, partly-encumbered bet that bitcoin will rise, and rise faster than the share count grows. The metrics that test that bet are not the margin in the marketing deck but the satoshis-per-share figure, the balance of the redemption windows against the price of bitcoin, and how fast new shares are being sold to buy more coins. On the first three quarters of evidence, the factory delivered and the equity did not. The 49% margin told you the machine was good. It never told you the machine was yours.