BTDR — Q2 2026 Financial Report Analysis

Bitdeer mined almost five times as much bitcoin as a year earlier, 2,694 BTC against 565, because it has now installed its own SEALMINER machines at scale. Revenue rose 47% to $228.8 million. The company still lost more money, though. Depreciation on the new machines, $31 million of net interest on the debt that paid for them, and a lower bitcoin price meant every bitcoin cost slightly more to produce than it earned. The net loss widened to $92.3 million from $62.9 million. Compared with Q1 2026 the direction is better: the loss shrank from $159.5 million and adjusted EBITDA more than doubled. AI Cloud, the business Bitdeer is moving toward, brought in $14.0 million, about 6% of revenue, and has not yet made a gross profit.

At a glance

• 2,694 BTC mined (+377% YoY): Bitdeer's own machines now run 73.0 EH/s of mining power, up from 16.5 EH/s. Bitcoin mining provides 85% of revenue ($193.4M of $228.8M).
• Gross margin −3.7% (vs +7.7%): after electricity and depreciation, the core business lost money on every dollar of sales. A year ago the gross profit came mostly from selling SEALMINER machines to outside customers, and those sales have almost stopped ($0.4M vs $69.5M).
• $456.7M raised from new shares in one quarter: shares sold through the at-the-market (ATM) program, net of costs, paid for the build-out. The diluted share count rose 27% year on year, so the loss per share widened much less than the loss itself.

The numbers

Prior-year figures are recast to U.S. GAAP. The company switched from IFRS on 1 January 2026. "Adjusted EBITDA" is the company's own measure. It is earnings before interest, tax, depreciation and amortization, and it also leaves out bitcoin price swings, share-based pay and several one-off items.

Hashrate is the total computing power a miner points at the bitcoin network. One EH/s (exahash per second) is a billion billion guesses a second. A miner's share of new bitcoin roughly tracks its share of the network's total hashrate. J/TH (joules per terahash) measures how much electricity a machine uses for each unit of mining work. Lower means cheaper bitcoin.

Where the revenue came from

Bitdeer explains the jump in self-mining revenue by "the increase in the average self-mining hashrate for the quarter by 389.4% to 69.5EH/s from 14.2 EH/s," which was "partially offset by a lower average Bitcoin price." The business has changed shape. A year ago Bitdeer was largely a mining-chip and machine maker that also hosted other people's rigs. Now it uses its own machines itself. That is why SEALMINER sales went from $69.5 million to almost nothing and hosting revenue fell. Hosting hashrate dropped to 8.2 EH/s from 13.9 EH/s.

Mining economics: each bitcoin cost a little more than it sold for

Self-mining and co-mining revenue together was $193.4 million for 2,694 BTC, about $71,800 per bitcoin. This is our arithmetic from the release's figures. The same calculation for Q2 2025 gives about $105,000 per bitcoin, in line with the release's note that bitcoin prices were lower this year. On the cost side:

• Electricity: $98.0M across the two mining lines, about $36,400 per BTC. Bitdeer paid an average of $44/MWh, against $43 a year earlier. Total power use more than doubled, to 2.54 million MWh.
• Depreciation and share-based pay: $92.5M, about $34,300 per BTC. A year ago self-mining depreciation was only $12.0M. The difference is the new SEALMINER fleet: $613 million of machines and parts moved from inventory onto the balance sheet as equipment during the quarter, and they are now written off over their useful lives.
• All-in cost of revenue: $199.8M, about $74,200 per BTC. That is slightly more than the ~$71,800 received, so the two mining lines made a combined gross loss of $6.6M.

This is the key point for a reader. On electricity alone, each bitcoin cost roughly half what it sold for, so the machines bring in cash day to day. But once the cost of the machines is spread over their lives, Bitdeer roughly broke even on mining at Q2's bitcoin price. After overhead and interest it lost money. The 38% gain in fleet efficiency helped, but it did not offset the lower bitcoin price.

Takeaway: Bitdeer now produces bitcoin at a cash electricity cost of about $36,000 per coin. But after depreciation on its new fleet, each coin cost about $74,000 against roughly $72,000 received, before $31M of quarterly net interest. The company has turned into a large miner with high fixed costs. Its profit now depends heavily on the bitcoin price, and the quarter's losses are being funded by issuing new shares and borrowing.

Below gross profit: overhead and interest

• Operating expenses (selling, admin, R&D) were $72.6M, up from $42.2M. Admin costs rose $14.3M, driven by $6.8M more staff cost and $4.8M more consulting fees for "general corporate management and compliance activities." R&D rose to $36.1M. The release attributes this partly to "a one-off incremental development expense" but does not say how large it was.
• Net interest expense tripled to $31.1M from $9.6M "primarily due to increased borrowing through the convertible senior notes and borrowing from a related party." Total borrowings stand at about $1.8 billion.
• The operating loss of $101.7M compares with a small operating profit of $4.6M a year earlier. But that year-ago profit relied on a $40.7M gain in the value of bitcoin Bitdeer held. Without it, Q2 2025 would have shown an operating loss of about $36M.

What the headline numbers hide

• Cash conversion: operating cash flow looks worse than it is. Operating cash flow was −$158.5M, against a net loss of $92.3M. The gap comes from how miners are accounted for. Mined bitcoin counts as revenue when earned, but the cash from selling it shows up under investing activities. Bitdeer received $195.5M from selling digital assets in Q2. Adding that to operating cash flow gives about +$37M from day-to-day operations. That surplus was far smaller than the $266.0M of capital spending: $150.0M on SEALMINER production and $116.0M on data centres, GPUs, tariffs and freight. The shortfall was covered by $463.8M of new shares (before $7.0M of costs) and new related-party loans.
• Bitcoin reserves have mostly gone. Bitdeer held 150 BTC at quarter end, against 1,502 a year earlier. Digital assets on the balance sheet were $34.8M. A further $162.2M is bitcoin "receivables from a related party," tied to bitcoin borrowed from and pledged to that party. Bitdeer is selling what it mines to pay for growth, not holding it.
• Adjusted figures run in both directions this quarter. Adjusted EBITDA of $31.1M adds back $107.7M of depreciation, which is real given how quickly mining machines age. Adjusted loss was $96.0M, slightly worse than the GAAP loss of $92.3M. The non-cash items the company strips out were net gains this time: a $14.8M gain on a bitcoin-denominated loan (the loan shrinks in dollar terms when bitcoin falls) and a $13.0M gain on Norwegian power contracts.
• One-offs in both years. Q2 2026 includes a $17.2M loss on disposing of equipment, which the release does not explain; $4.3M of insurance money for the fire at the Massillon, Ohio site; and an $11.7M tax benefit. Q2 2025's loss was inflated by a $39.7M derivatives loss and a $16.2M loss on retiring convertible notes, and flattered by the $40.7M bitcoin gain. Those prior-year items make the 46.6% wider net loss a rough comparison. The comparison that does not depend on them, gross margin falling from +7.7% to −3.7%, points the same way.
• Dilution softens the per-share figure. The net loss grew 46.6%, but the loss per share grew only 15.6%, because the weighted share count rose 27% to 246.3M. Shares outstanding (Class A plus Class V) rose about 15% in six months, from 235.6M to 271.8M. On 8 October Bitdeer approved a new $100M share buyback, after using up its earlier $40M program. That buyback is small next to the ~$464M of stock it sold in Q2 alone.
• Near-term debt from a related party. $491.0M of related-party borrowings is due within twelve months, slightly more than the $456.8M of unrestricted cash at 30 June. The release does not name the lender or say whether the loans will be rolled over.

The AI pivot: what it adds up to so far

AI Cloud revenue rose to $14.0M from $1.3M. It is still only about 6% of revenue, and it lost $2.3M at the gross level after $12.3M of "other costs," which the MD&A attributes partly to AI cloud service fees. The financial case for the pivot rests on contracts that begin later:

• Tydal, Norway: a $4.7 billion, 16-year lease with Volta for an AI/HPC data centre. Here Bitdeer supplies power, space and cooling, and the customer brings its own chips. This arrangement is called colocation. 121 IT MW will run NVIDIA GPUs for "a leading AI lab." The agreement was signed after the quarter ended. Phase 1 (66.5 MW) is due to be ready in Q4 2026 and phase 2 in Q1 2027. Spread evenly, the contract would average roughly $290M a year (our arithmetic; the release does not give the payment schedule). That would be more than all of Q2's quarterly revenue on an annualised basis.
• Johor Bahru, Malaysia: a 10-year lease for a 21.7 IT MW facility, handed over in Q1 2027, planned for 128 NVIDIA GB300 NVL72 systems.
• Delays elsewhere: Knoxville, Tennessee was redesigned for larger customers and is now due in Q3 2027. The 570 MW Clarington, Ohio site's timing is "to be updated" because of litigation from a neighbouring company. The 563 MW Rockdale, Texas site is "in active evaluation" for AI conversion.

Outlook

Bitdeer gives no revenue or earnings guidance. Its plans are stated as data-centre ready-for-service dates. The quarter-on-quarter trend is improving: revenue rose 21% from Q1's $188.9M, the gross loss narrowed from $39.0M to $8.5M, and adjusted EBITDA rose from $14.4M to $31.1M as more of the fleet came online. Over the next two to three quarters, mining results will depend mostly on the bitcoin price, because the fleet's costs are now largely fixed. Each $10,000 move in bitcoin changes Q2-scale mining revenue by roughly $27M a quarter (our arithmetic from 2,694 BTC, ignoring changes in network difficulty). The AI side will not affect results meaningfully until Tydal phase 1 starts earning, in Q4 2026 at the earliest. What to watch in Q3: whether Tydal stays on schedule, how the $491M of related-party debt due within a year is handled, and how much further the share count rises.

Source: Bitdeer's Q2 2026 earnings release (6-K Exhibit 99.1, filed 10 August 2026). The full interim financial statements for the six months to June were filed on 29 September 2026.