ANY — Q2 2026 Financial Report Analysis
Q2 · Fiscal year 2026 · Published by Pham Hop
Sphere 3D (now DarkHorse Technologies, DRK) lost $13.8M on $2.5M revenue in Q2 2026 as Bitcoin-driven impairments hit, with $2.8M cash, a going-concern warning and a share count that more than doubled to fund its AI-infrastructure pivot.
- Revenue
- $2.5M
- -18.8% YoY
- Net income
- -$14M
- Diluted EPS
- $-2.68
- Operating margin
- -563.7%
This period vs a year ago
- Same period last year
- This period
Year-ago figures (≈) are worked out from the growth rate the company reported. Each row has its own scale.
Sphere 3D, which now trades as DarkHorse Technologies (Nasdaq: DRK), lost $13.8 million in the second quarter of 2026 on revenue of $2.5 million. A year earlier it reported a $1.7 million profit. Most of the swing comes from a $7.6 million write-down of mining machines and supplier contracts after the Bitcoin price fell, plus the loss of a one-time investment gain in the prior year. The bigger story is the balance sheet. The company ended June with $2.8 million of cash, used $9.0 million in operations in the first half, and says there is "substantial doubt" it can keep operating without new funding. It is paying for its move from Bitcoin mining into AI data-center sites by issuing large numbers of new shares.
At a glance
- Revenue fell 18.8% to $2.5 million. The company mined 29.0 Bitcoin, down 6.1%. Each coin was worth much less than a year ago, and even this figure includes about $0.4 million from the Cathedra business, which was acquired on June 1.
- The loss from operations was $13.8 million, against $2.6 million a year ago. About $7.6 million of that is non-cash impairment (a write-down of an asset to what it is now worth). Even without it, the operating loss was roughly $6.2 million.
- The share count more than doubled in six months. It went from 3.39 million at December 31 (counted after the 1-for-10 reverse split) to 7.64 million at June 30 and 8.70 million by August 11. Another 1.67 million shares, with matching warrants, were sold in a September private placement.
Quarter in numbers
| Metric | Q2 2026 | Q2 2025 | YoY Change |
|---|---|---|---|
| Revenue | $2.45M | $3.02M | -18.8% |
| Bitcoin mined | 29.0 BTC | 30.9 BTC | -6.1% |
| Cost of revenue (excl. depreciation) | $2.29M | $2.30M | -0.6% |
| General & administrative | $4.83M | $2.13M | +127% |
| Impairments (equipment + intangibles) | $7.63M | — | n/m |
| Loss from operations | -$13.82M | -$2.63M | n/m |
| Operating margin | -563.7% | -87.0% | n/m |
| Net income (loss) | -$13.83M | +$1.67M | n/m |
| Diluted EPS | -$2.68 | +$0.60 | n/m |
| Weighted diluted shares | 5.17M | 2.78M | +86% |
| Cash (period-end vs Dec 31, 2025) | $2.85M | $3.71M | -23% |
Operating margin is the operating result divided by revenue. Here it is deeply negative because costs were several times revenue. "n/m" means a percentage change is not meaningful when the result moves from profit to loss.
What drove the quarter
Revenue: fewer coins, each worth less. Management attributes the $0.5 million drop "primarily" to the lower Bitcoin price. Volume also fell, because older machines were being removed and replaced. A rough calculation using the filing's own figures: about $2.05 million of mining revenue across 29.0 coins works out to roughly $70,000 per coin. The same calculation for Q2 2025 ($3.02 million across 30.9 coins) gives about $98,000. Over the first half, the company mined 54.3 Bitcoin, down 11.4%, and revenue fell to $4.4 million from $5.8 million.
Costs: the acquisition added overhead before it added much revenue. Direct cost of revenue was flat at $2.3 million, but that includes $0.8 million of costs tied to Cathedra's business. Cathedra brought in $0.4 million of revenue and an operating loss of $0.6 million in its first month. General and administrative expense more than doubled to $4.8 million. The filing says that is "primarily due to a $1.8 million increase in costs related to our acquisition," along with $0.6 million more in share-based pay (compensation paid in stock), plus higher investor-relations and legal costs.
Write-downs. The company recorded a $7.0 million impairment on mining equipment "primarily due to the change in market conditions and decline in Bitcoin prices," plus $0.6 million on supplier agreements. Property and equipment fell to $10.7 million from $14.6 million at year-end.
The Cathedra deal and the pivot
On June 1 the company closed an all-stock merger with Cathedra Bitcoin. It issued 2.41 million common shares and 1.39 million convertible preferred shares, valued at $7.3 million in total. Former Cathedra holders now hold about 33% of the votes. The deal changes what the company is: it now owns the power and data-center sites under much of its fleet instead of paying third parties to host machines. The company reports about 53 MW of operating power across five data centers in Iowa, Kentucky and Tennessee. In late May it also signed a 30 MW co-mining agreement with Bitdeer across three sites. Bitdeer supplies the machines and the two companies split the net mining proceeds.
Events after the quarter push further in the same direction. In September the company finished a 90-day strategic review and agreed to sell its Iowa site for $1.5 million, expecting about $0.5 million more back in utility deposits. It also agreed to sell its legacy fleet of about 5,500 mining machines for about $3 million. It changed its name to DarkHorse Technologies and its ticker to DRK on September 17. After these sales, the company's own exposure to Bitcoin mining comes mainly through revenue-share hosting deals, and its stated plan is to convert small, power-ready sites in the Tennessee Valley Authority region to AI and high-performance-computing (HPC) use. The filing and the company's own investor materials say none of that AI/HPC capacity is contracted yet.
What the headline numbers hide
- The year-ago profit was not an operating profit. Q2 2025 net income of $1.7 million came entirely from a $4.3 million gain on the company's former stake in Core Scientific. The business itself lost $2.6 million from operations that quarter. So the move from +$0.60 to -$2.68 per share overstates how much the underlying business got worse. The operating loss excluding impairments still more than doubled, to about $6.2 million.
- Revenue is mostly not cash when it is booked. Mined Bitcoin counts as revenue on the day it is mined, but it only becomes cash when the company sells it. In the first half, $4.0 million of revenue was Bitcoin mined, and the company sold $5.3 million of Bitcoin, running its holdings down to 20.5 BTC ($1.2 million). Under the cash flow rules, those sale proceeds count as investing cash, not operating cash. That is why operating cash flow (-$9.0 million for the half) looks much worse than the cash actually spent net of Bitcoin sales.
- Going-concern warning. Management says cash "may not be sufficient" and that there is "substantial doubt about our ability to continue as a going concern within 12 months" without more funding. Working capital (current assets minus current liabilities) was only $0.2 million, down $6.7 million since December. Current liabilities more than tripled to $5.9 million after the Cathedra combination. The company also prepays about $0.9 million a month for power.
- Dilution is how the company pays its bills. In the half it raised $2.5 million by selling 1.15 million shares into the market (an "at-the-market" program), plus $1.7 million more after June 30. It then expanded the program to $10.3 million. In September it sold 1.67 million units at $3.00, each with a five-year warrant at $3.50, for $5.0 million; three directors, including the CEO and chairman, bought about $1.0 million of it. On top of that, 1.39 million convertible preferred shares and the warrants inherited from Cathedra could add more shares later. The 1-for-10 reverse split in February cut the share count on paper; the share count has more than doubled since.
- Possible cash claims from executives. Retention bonuses of $1.6 million (CEO) and $1.1 million (CFO) become payable if operating milestones are reached by June 2028. They can be paid in cash or shares. On July 31 the CFO gave formal notice to preserve his right to the $1.1 million. No liability has been recorded.
- Some legacy problems are fading. Legal fees fell $0.9 million in the half after the dispute with former mining partner Gryphon Digital Mining was resolved. A separate note receivable from Rainmaker, which was supposed to be settled for $0.5 million by February 27, was not paid and is fully reserved (written down to zero on the books).
Takeaway: This quarter is less about Bitcoin mining than about whether the company can fund its pivot. Cash and Bitcoin together came to about $4.0 million at June 30, against first-half operating cash burn of $9.0 million. Planned asset sales (about $5 million) and the $5.0 million private placement buy time. But the AI/HPC plan has no contracted customers yet, the company has offered to fund an $8–10 million substation in Hopkinsville, and every round of funding so far has meant a lot of new shares. Anyone holding the stock is betting that power-ready sites will be worth more per share than the dilution needed to develop them.
Outlook
Management gives no numerical guidance. It says it expects "to expand our revenue mix and utilization going into the second half of 2026" as the Bitdeer sites come online: one site was installed after the quarter, and the remaining two are expected before November 2026. Revenue from hosting and revenue-share should be steadier than mining with its own machines, but the company's share of each coin will also be smaller. Once the Iowa site and the legacy fleet are sold, revenue will shrink and costs should fall. That will make Q3 and Q4 hard to compare with earlier quarters.
What to watch in the Q3 10-Q (due in November):
- Cash runway. Cash after the private placement and the asset sales, measured against monthly burn and the $0.9 million monthly power prepayment.
- Whether the going-concern language softens or stays.
- Any signed AI/HPC customer, or zoning and utility approval, for the proposed 50 MW Hopkinsville data center. Without one, the pivot remains a plan, not revenue.
- Further share issuance, including any conversion of the Series I preferred.
Our read: the business keeps shrinking until AI/HPC contracts exist. The most likely path is more losses and more dilution through the rest of 2026. Two things would change that view: a signed customer contract, or the asset sales closing quickly enough to fund development without another round of share sales.