BGIN — H1 2026 Financial Report Analysis

Revenue fell 85% as Kaspa mining stopped paying, and the loss narrowed only because last year's write-downs were bigger

BGIN Blockchain, a Singapore-based maker of crypto-mining machines (the ICERIVER brand) that also mines coins itself, reported revenue of $7.4 million for the six months to June 30, 2026, down 84.5% from $47.7 million a year earlier. Its net loss attributable to shareholders was $25.6 million, smaller than the $60.4 million loss in the first half of 2025. The narrower loss does not mean the business recovered. Every revenue line shrank, the company now spends more producing what it sells than it earns from selling it, and the loss got smaller mainly because last year's half carried far larger write-downs and crypto price losses. The figures come from the unaudited interim statements BGIN filed on Form 6-K on September 25, 2026. That filing contains no management commentary, so the explanations of what caused the moves come from the company's FY2025 annual report (Form 20-F, filed April 2026).

At a glance

• Revenue $7.4M, down 84.5%. All four businesses (mining, machine sales, hosting, mining pool) fell between 81% and 96%. This continues a slide from $302M of revenue in 2024 and $67M in 2025.
• Gross loss $9.9M on $7.4M of revenue. The direct cost of producing what BGIN sold was $17.3M, more than twice what it sold for. It lost money on each mining dollar and on each machine sold.
• Cash plus crypto down $11.8M in six months, to $36.9M. That is about the same size as the $34.1M of taxes payable still on the balance sheet, which is the main pressure point in this filing.

The numbers

Gross margin is the share of revenue left after the direct costs of producing what was sold. Operating margin is the share left after all operating costs, before interest and tax. When either is negative, the company spent more than it took in.

What drove the collapse: one coin and the machines built for it

BGIN's business has rested on Kaspa (KAS), a smaller cryptocurrency that its "KS" series machines are built to mine. The 20-F lays out how concentrated this was. In FY2025, KAS made up 97.6% of the coins BGIN mined itself, and all of its mining-pool revenue came from KAS. The average KAS price fell from $0.1440 in 2024 to $0.0816 in 2025. In the company's words, that made "KAS mining uneconomical relative to prevailing electricity costs." By December 31, 2025, only 4,655 of its 74,743 mining machines (6.2%) were running, compared with 37,864 of 46,989 a year earlier. Sales of its KS machines fell from 84,544 units in 2024 to 1,377 in 2025.

The first half of 2026 shows the result:

• Mining: revenue of $5.81M against $8.99M of mining costs. Each $1 of coins mined cost about $1.55 to produce, up from about $1.30 a year earlier. The fleet ran at a loss even after most of it had been switched off.
• Machine sales: $1.02M of revenue against $7.72M of cost. The cash-flow statement shows $6.88M of inventory provisions and write-offs in the half (a provision is a write-down of unsold stock to what it can still be sold for). That is the most likely reason cost of machine sales was roughly 7.5 times revenue. Inventory on the balance sheet fell from $7.85M to $6.45M even though BGIN spent $5.5M of operating cash building inventory.
• Hosting and mining pool: both now roughly break even on tiny revenue ($0.35M and $0.19M). Fewer outside miners are running machines with BGIN or pooling hash power through it.

What the headline numbers hide

The narrower loss comes from smaller write-downs, not a better business. The $34.8M improvement in net loss came almost entirely from items that do not reflect day-to-day trading:

• Impairment of property and equipment (writing down the book value of mining machines) fell from $19.12M to $3.95M.
• Change in fair value of crypto holdings, a loss from coin prices falling while BGIN held them, fell from $14.37M to $4.03M.
• Depreciation fell from $12.29M to $4.15M, because most machines had already been written down or switched off. Much of this sits inside cost of revenue, so it helped gross profit, but not enough to stop it getting worse.
• General and administrative costs fell from $13.48M to $4.51M (-66.5%), and R&D fell from $7.50M to $3.57M (-52.3%).

Meanwhile the trading result, gross profit, got worse: a $9.87M loss versus a $6.31M loss. Cost cutting and smaller write-downs offset a shrinking, loss-making core.

Non-cash charges explain most of the loss, but cash is still leaving. Depreciation, impairment, inventory write-downs and crypto fair-value losses add up to $19.0M of the $25.8M net loss. Operating cash flow was -$14.8M. That figure needs care at a miner, because coins it mines count as revenue but are not cash until sold. BGIN raised cash by selling $19.0M of crypto (an investing inflow), which is why the bank balance rose $5.1M to $31.4M. The clearer measure is cash plus crypto together: $48.7M at December 31, 2025 and $36.9M at June 30, 2026, a drop of $11.8M in six months. Crypto holdings themselves went from $22.5M to $5.5M, so most of that buffer has already been sold.

The tax liability now exceeds the cash. "Taxes payable" stayed at $34.06M, almost unchanged from December. The 20-F says this is "primarily related to income taxes," left over from the profitable years. BGIN paid just $8,594 of income tax in the half, compared with $16.7M in H1 2025. That liability is now larger than the $31.4M of cash and close to the $36.9M of cash plus crypto. Total liabilities of $38.9M are mostly this one item. The 6-K does not say when the balance falls due, which is the most important thing it leaves out.

Equity is being consumed. Shareholders' equity attributable to BGIN fell from $54.5M to $29.1M in six months. Retained earnings dropped from $28.1M to $2.5M, so at this rate of loss they turn negative in the second half.

EPS isn't flattered by share count. No shares were issued or bought back in the half (113.1M Class A plus B outstanding throughout). The weighted average share count was 113.1M compared with 108.1M in H1 2025, which was before the October 2025 Nasdaq IPO. The narrower loss per share comes from the smaller loss, not from share-count changes. Two September 2026 filings matter for the future, though. BGIN registered up to 9.0M new Class A shares (about 8% of shares outstanding) for its equity incentive plan on Form S-8, and on September 15 said it will follow Cayman Islands practice rather than Nasdaq's rule requiring shareholder approval for equity plans.

One comparability note: H1 2025 included $0.56M of realized futures losses, $0.25M of crypto-lending rewards, and a $4.05M cash dividend paid before the IPO. None of these recurred in 2026.

Takeaway: BGIN's improvement is an accounting one. The loss narrowed because there was less left to write down, while its two main businesses, mining and machine sales, still lost money on each dollar they sold. The number to watch is cash plus crypto ($36.9M, falling about $12M per half) against a $34.1M tax bill whose due date the filing doesn't give.

What comes next

BGIN gave no financial guidance in the 6-K. The 20-F said management believed cash on hand, together with "proceeds received from sale of cryptocurrencies," would cover working capital and capital spending for at least 12 months from April 2026. Since then crypto holdings have fallen to $5.5M, so most of that second source has been used up.

The company's plan is to move away from Kaspa:

• BT1 Bitcoin chip: BGIN announced the tape-out of its 4nm BT1 chip on March 17, 2026 (tape-out is the point where a chip design is sent for physical manufacture). As of the 20-F, a prototype miner existed but had not been released commercially.
• BL1 Dogecoin chip: in development. The 20-F estimates combined BL1 and BT1 R&D at about $10M, to be funded from cash and operating cash flow.
• Custom chip work for outside customers: launched May 26, 2026, with a first batch of 3,000 custom mining machines (2,400 delivered to a single customer, 600 to be sold through ICERIVER). H1 revenue does not show it yet: machine sales were $1.0M and "Others" was $2,118.

Our view: H2 2026 depends on whether BT1 machines reach customers and whether the custom-chip business brings in meaningful revenue. Bitcoin-mining hardware is a much bigger market than Kaspa, but larger, established manufacturers already dominate it. Without new revenue, the H1 rate of decline (about $12M of cash plus crypto per half) and the $34M tax liability leave little room. A capital raise or a disclosed payment schedule for the tax balance would be the next things to look for. The next full set of financials is the FY2026 Form 20-F, which on last year's timing (filed April 28) is due around late April 2027.