The biggest bitcoin miner earnings print of the season landed Thursday night, and it was ugly in exactly the way this site warned it would be — and shrugged off in exactly the way this market keeps shrugging. MARA Holdings reported a second-quarter net loss of $611 million, or $1.60 per share, against expectations that ranged from a $0.06-per-share loss (FactSet) to a $0.35 profit on some compiled estimates, per MarketScreener and Stocktwits. Revenue came in at $174.9 million, down 27% year over year from $238.5 million and roughly 16% below the ~$208–209 million consensus, per Investing.com. The stock’s reaction, as of the August 6 after-hours session: up 0.75% to $10.73, after a 5.25% slide to $10.65 in the regular session before the print.
That non-reaction is the story. A company just reported a loss larger than its entire quarterly revenue — three and a half times larger — and the market treated it as old news. It was old news, for one reason: about $343 million of the loss was an unrealized mark-to-market charge on digital assets under the FASB fair-value rules, driven by bitcoin marking at roughly $63–64K on June 30 versus roughly $80K on March 31. Readers of our miner-earnings guide published Wednesday saw this exact trap coming: fair-value whiplash makes headline losses season-wide, mechanical, and — crucially — non-cash.
Grading the four questions we set before the print
Wednesday’s preview framed MARA as the sector’s tone-setter and put four questions on the table. Here is how each resolved. One: the headline versus consensus. A hard miss on both lines — revenue 16% light, EPS a $1.60 loss against modeled profit or small loss. Two: treasury policy, on trial after Strategy’s sale. The answer was louder than expected: bitcoin holdings fell 29% year over year, from 49,951 BTC to 35,577 BTC, per compiled filings data — the company describes this as strategic deployment of assets. The week Strategy disclosed selling coins at a loss, the largest listed miner confirmed its own stack is a funding source, not a museum. The treasury-model inversion we flagged Tuesday — coins funding paper because paper can no longer fund coins — now has a second data point. Three: do AI dollars show up? They are the entire forward story: the Exaion majority acquisition closed, a definitive agreement to acquire Long Ridge Energy & Power was announced, and guidance projects Q3 EPS of $0.43, Q4 at break-even, full-year 2026 revenue of $810 million and fiscal 2027 revenue of $947 million, per Investing.com’s slide coverage. Four: the size of the mark. $343 million — the fair-value trap printed at almost exactly the scale the mechanics implied.
“No longer simply a Bitcoin miner”
MARA is no longer simply a Bitcoin miner. We’re building an energy-dominant digital infrastructure platform. — Fred Thiel, MARA chairman and CEO, Q2 2026 earnings call, August 6, 2026
Thiel went further on the call: “The question is no longer who can fund the next wave of compute, it is who has the power.” Strip out the framing and the operating quarter underneath was actually competent by mining standards: energized hashrate rose 22% year over year to 70.3 EH/s, bitcoin production rose 3% to 2,422 BTC, and cost per petahash per day improved 4% to $27.70. The problem is arithmetic, not execution: production rose 3% while bitcoin’s average price fell roughly 28% year over year. When your commodity drops by a quarter, a 22% capacity increase does not save the revenue line. Hence the pivot: power, land and compute, sovereign cloud via Exaion, AI-ready data centers — the same lane TeraWulf, Galaxy and Hut 8 are racing into, as of this week’s parallel earnings prints.
| Metric (Q2 2026) | Actual | Comparison |
|---|---|---|
| Revenue | $174.9M | vs ~$208–209M consensus; $238.5M Q2 2025 (−27% YoY) |
| Net income | −$611M (−$1.60/sh) | vs FactSet −$0.06 / other est. +$0.35 |
| Unrealized fair-value mark | ~−$343M | non-cash, BTC ~$63–64K Jun 30 vs ~$80K Mar 31 |
| BTC holdings | 35,577 | vs 49,951 a year ago (−29%) |
| Energized hashrate | 70.3 EH/s | vs 57.4 EH/s (+22% YoY) |
| BTC produced | 2,422 | +3% YoY |
| Cost per PH/day | $27.70 | vs $28.70 (−4%) |
| Guidance | Q3 EPS $0.43; Q4 break-even | FY26 rev $810M; FY27 $947M |
What it means for the sector — and what to watch next
MARA’s print lands in a week where TeraWulf reported 71% of its revenue from HPC leases rather than mining, and Galaxy narrowed its loss on data-center buildout — we cover both in today’s companion update. The sector-wide pattern as of August 7: mining revenue shrinks, fair-value marks swing the headline, and every conference call leads with megawatts instead of exahash. Riot — which postponed its own call this week without explanation — still has not named a new date; our standing marker treats an announcement by August 13 as the routine outcome. Meanwhile bitcoin itself sat near $64,400 early Friday, per Fortune’s tracking, with the July jobs report due at 8:30 a.m. ET — the single biggest scheduled input of the week for the Fed path that every miner balance sheet now leans on. As of publication, MARA shares are worth roughly what they were a week ago; the $611 million headline moved the tape less than one macro print is about to.
Why did MARA lose $611 million?
Mostly accounting: roughly $343M was an unrealized fair-value mark on its bitcoin under FASB rules, because BTC ended June near $63–64K versus ~$80K at the end of March. The rest reflects a 27% revenue decline driven by bitcoin’s lower average price.
Did MARA sell bitcoin?
Holdings fell 29% year over year to 35,577 BTC, which the company frames as strategic deployment of assets to fund its infrastructure buildout — a clear break from the pure-HODL treasury era.
Why did the stock not crash?
The loss was mostly non-cash and well-telegraphed, and the market is valuing MARA on its AI/energy infrastructure guidance (Q3 EPS $0.43 projected) rather than on quarterly mining economics.
Is the AI pivot real revenue yet?
Partly: Exaion is closed, Long Ridge is under definitive agreement, and guidance calls for $810M FY26 revenue — but Q2 itself was still overwhelmingly a mining quarter. Execution risk on the pivot is the bull-bear battleground.
What should bitcoin investors watch next?
Riot’s rescheduled earnings date, the ~August 11 difficulty retarget, and whether miner treasuries keep shrinking — miner selling is a structural supply flow for bitcoin itself.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrencies and crypto-linked equities are volatile and you can lose money. Do your own research and consult a licensed financial advisor before making investment decisions.