BTBT — Q2 2026 Financial Report Analysis
Q2 · Fiscal year 2026 · Published by Pham Hop
Bit Digital lost $107.2M ($0.31/share) as ETH's price fall flowed through fair-value and LsETH impairment charges; revenue rose 25% to $32.1M only thanks to a $12.3M one-time customer termination fee.
- Revenue
- $32M
- +25.2% YoY
- Net income
- -$107M
- Diluted EPS
- $-0.31
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.
Bit Digital lost $107.2 million attributable to its shareholders in the second quarter of 2026 ($(0.31) per share), against a $14.9 million profit a year earlier, even though revenue rose 25% to $32.1 million. Almost all of the swing came from Ethereum: the price of ETH fell to about $1,569 by June 30, and the accounting rules for crypto turn that drop straight into the income statement. The operating businesses, which are now mainly the AI data centers and GPU cloud run by majority-owned WhiteFiber (Nasdaq: WYFI), were weaker than the headline suggests: a one-time $12.3 million termination fee paid by a departing customer is counted inside the quarter's cloud revenue.
At a glance
- $32.1M revenue, +25% year over year. Without the $12.3M one-time termination fee it would have been about $19.8M, roughly 23% lower than a year ago, mainly because bitcoin mining is being wound down and the largest cloud customer left.
- About $71.6M of the loss came from crypto price accounting. That is the $28.8M loss on ETH and bitcoin held, a $46.0M write-down of liquid-staked ETH, an $8.0M loss on crypto held in an outside fund, partly offset by an $11.3M gain. None of it is cash leaving the company.
- 164,310.5 ETH (about $258M at $1,569) and a ~69.6% stake in WhiteFiber. Those two holdings are what the company now consists of. Bitcoin mining earned only 32.3 BTC in the quarter.
What Bit Digital is now
The 10-Q describes the company as a "strategic asset company" with two pieces:
- An Ethereum treasury. It owns ETH and "stakes" it, meaning it locks the coins up to help run the Ethereum network and is paid small rewards in new ETH (a few percent a year).
- About 69.6% of WhiteFiber, a separately listed company that rents out Nvidia GPU servers for AI work ("cloud services") and rents space, power and cooling in its own data centers ("colocation"). Because Bit Digital controls WhiteFiber, all of WhiteFiber's revenue and costs are included in Bit Digital's figures, and the ~30% owned by outside shareholders is backed out at the bottom of the income statement (the "noncontrolling interest", a $4.5M share of the loss this quarter).
Bitcoin mining, the original business, is being shut down: active hash rate (mining computing power) was 0.4 EH/s on June 30, against 1.2 EH/s a year earlier, and the company says it will not put meaningful new money into it.
The quarter in numbers
| Metric | Q2 2026 | Q2 2025 | YoY Change |
|---|---|---|---|
| Total revenue | $32.1M | $25.7M | +25.2% |
| Cloud services revenue (incl. $12.3M termination fee) | $23.8M | $16.6M | +43.5% |
| Cloud services revenue excl. termination fee | $11.5M | $16.6M | -30.7% |
| Colocation revenue | $4.7M | $1.7M | +173% |
| ETH staking revenue | $0.9M | $0.4M | +147% |
| Bitcoin mining revenue | $2.4M | $6.6M | -64.2% |
| Gross margin (revenue minus direct costs, before depreciation) | 57.9% | 48.6% | +9.3 pts |
| Gains/(losses) on digital assets | $(28.8)M | $27.2M | n/m |
| Operating income/(loss) | $(82.6)M | $11.7M | n/m |
| Net income/(loss) attributable to Bit Digital | $(107.2)M | $14.9M | n/m |
| Diluted EPS | $(0.31) | $0.07 | n/m |
| Weighted diluted shares | 350.0M | 208.8M | +67.6% |
| ETH held (incl. liquid-staked and fund) | 164,310.5 | n/a | n/a |
n/m = not meaningful, because the result went from a profit to a loss. Operating margin isn't shown as a headline figure here: GAAP operating loss was 257% of revenue, and that number reflects ETH's price fall far more than how the data centers performed.
Why the loss was so large
Fair-value accounting. Since 2024, US rules (ASU 2023-08) have required companies to carry crypto at its market price on the last day of each quarter, and every rise or fall goes through profit even if nothing is sold. ETH fell during the quarter, so the ETH and bitcoin Bit Digital held directly produced a $28.8M loss. A year earlier, rising prices produced a $27.2M gain. That alone accounts for $56.0M of the $126.5M year-over-year swing in net income.
The liquid-staking write-down. In April the company handed 73,235 ETH to the Liquid Collective staking protocol and got back 66,192 "LsETH", receipt tokens that can be swapped back for the ETH plus rewards. The accounting for LsETH is different from ETH: it is carried at cost and can only be written down (impaired) when the price falls, never back up. That produced a $46.0M impairment, which the company calls non-cash and unrealized. A separate $11.3M gain was booked when 49,000 LsETH were handed to a lender as collateral. The press release points out, correctly, that the company's economic exposure to ETH did not change: it is the same ETH, now in a different box on the balance sheet.
Other non-operating items. Crypto held in an externally managed fund lost $8.0M in value. The value of the conversion option on the 2030 convertible notes (bonds that can be swapped for shares) rose, which counts as a $14.0M non-cash charge. Interest expense was $8.1M, up from zero a year ago, on about $440M of debt (convertible notes, loans and credit facilities) taken on since mid-2025. The company also wrote off $5.0M of internal software it stopped developing.
Taking out the crypto items ($71.6M), the convertible-option charge ($14.0M) and the software write-off ($5.0M), the pre-tax loss would have been about $22.7M, not $113.1M. The company's own estimate is that non-cash and non-operating items made up "approximately $86 million" of the loss. Even on that more forgiving view, the business still lost money, and that $22.7M already includes the $12.3M termination fee.
Segment by segment
- Cloud services (WhiteFiber GPU rental): $23.8M revenue. The 10-Q says the increase came from "an increase in deployed GPU servers to new and existing customers," and that lost monthly revenue from the departing "Initial Customer" was "substantially offset by $12.3 million of termination fee revenue." Without the fee, cloud revenue was about $11.5M, down from $16.8M in Q1 2026 and $16.6M in Q2 2025. Direct costs rose 56% to $10.0M, partly because of a one-time payment to end a GPU server lease. That customer also left a $2.2M bad-debt charge.
- Colocation (renting data-center space): $4.7M, almost three times last year's figure, because the MTL-3 site near Montreal began billing chip maker Cerebras in November 2025 at about $1.0M a month. Gross margin was about 63%.
- ETH staking: $0.9M, down from $2.3M in Q1. The company earned 440.1 ETH from native staking, against 949.1 ETH in Q1. The fall comes from the accounting. Rewards on liquid-staked ETH are recorded only when the LsETH is redeemed, so most of the treasury's yield since April is not in revenue at all.
- Bitcoin mining: $2.4M on 32.3 BTC, compared with 68.2 BTC a year earlier, at about a 26% gross margin. Hosting deals with Soluna, GreenBlocks, Digihost and Bitdeer ended between February and June, and some of the machines are now in storage.
Takeaway: Bit Digital's results now depend almost entirely on two outside numbers: the ETH price and WhiteFiber's ability to bring new data centers online. This quarter both moved the wrong way for the headline. ETH's fall drove a $107M loss. The flattering +25% revenue growth rests on a $12.3M one-off payment from a customer who left, and without it revenue fell by about a fifth. The figure to watch next quarter is cloud and colocation revenue once NC-1 and the new GPU contracts start billing.
What the headline numbers hide
- Most of the loss is crypto price accounting, not cash. About $71.6M of the pre-tax loss came from ETH, bitcoin and fund marks plus the LsETH write-down. The LsETH rules are lopsided: if ETH recovers, the $46.0M impairment is not reversed while the tokens are held, so future profits will not reflect the recovery in the same way. Each 10% move in ETH changes the value of the direct holdings by roughly $12M, and by about $26M across all 164,310.5 ETH at the quarter-end price.
- The termination fee hides a revenue decline. The press release says cloud revenue rose 42% from Q1 at about a 58% margin. The 10-Q shows that $12.3M of that $23.8M is a one-time fee from the departing customer, and the amount has since been raised to $15.7M, still unpaid at the filing date. Without the fee, total revenue was about $19.8M and gross margin about 32%, not 57.9%. WhiteFiber's largest customer made up about 63% of its first-half revenue, so it depends heavily on a few clients.
- Operating cash flow is mostly prepayments. Operating cash flow for the first half was +$46.8M, but $63.6M of it was customer prepayments (deferred revenue), cash received for services not yet delivered. Without those prepayments, the business used cash. Capital spending was $344.7M in the six months, funded with borrowed money and new shares.
- Debt has gone up quickly, and some of it can be called in at short notice. Convertible notes are carried at $336.2M, net of discounts. They are Bit Digital's $150M of 4% notes due 2030 and WhiteFiber's $230M of 4.5% notes due 2031. The $50M Galaxy Digital loan at 5.5% is secured by 49,000 LsETH and can be "recalled by the lender upon seven days' notice." If ETH falls further, the company may have to post more collateral; it keeps 17,192 LsETH aside as "a buffer against margin requirements." A separate $19.4M B. Riley facility carries a 50.6% effective interest rate. At June 30, cash was $83.6M, of which only $27.5M sat at Bit Digital itself and $56.1M at WhiteFiber. Management says existing cash covers at least 12 months of operating needs.
- The share count rose 68%. Weighted diluted shares rose from 208.8M to 350.0M in a year. In Q2 alone the company sold 23.0M shares through its at-the-market program (selling new shares gradually on the exchange) for $34.5M net, or about $1.50 a share. Equity attributable to Bit Digital shareholders fell from $724.0M at December 31 to $395.2M.
- The year-ago profit was mostly crypto as well. Q2 2025's $14.9M net income included the $27.2M crypto gain and a $3.9M fund gain, so the year-over-year comparison flatters 2025 as much as it hurts 2026.
Outlook
Management gave no revenue or earnings guidance. The signals in the filings:
- NC-1 (North Carolina) starts contributing. The 10-Q says billing on the roughly $865M, 10-year Nscale colocation contract "is expected to commence during the third quarter," and the press release says initial billing has started, with full run-rate billing across 40 MW of contracted capacity expected in late August. WhiteFiber targets about 70 MW of gross data-center capacity by the end of 2026.
- A wave of new GPU contracts. Since its last earnings call WhiteFiber has signed more than $540M of new multi-year cloud contracts, including a $165.2M BaseTen order (start targeted November 2026), an $87.5M Iceland order (December 2026), a 5-year deal of more than $160M near Paris (now expected September 2026, pushed back from July) and a $108.2M Vera Rubin order (Q2 2027). Remaining contracted revenue was about $1.0 billion at quarter end, of which $57.7M is due in the rest of 2026.
- Financing is still the constraint. The 10-Q says funding the buildout "depends on successfully obtaining additional financing." The France project financing was still a term sheet when the 10-Q was filed.
- Strategic review. CEO Sam Tabar said "the Board is currently evaluating our options" because the market values Bit Digital "primarily as a passive digital asset treasury." The company's 27.04M WhiteFiber shares were worth about $1.05 billion at WYFI's $38.85 close. The company says it does not intend to sell WhiteFiber shares in 2026.
Our read: Q3 should show a real increase in recurring infrastructure revenue as NC-1 and MTL-3 bill for a full quarter. It will not repeat the $12.3M fee, though, and several large GPU contracts only start in Q4 or later, so reported Q3 cloud revenue could fall from Q2 before it rises. Net income will keep following ETH's price. A further ETH decline would also put pressure on the Galaxy loan's collateral. The test of the strategy is whether WhiteFiber's contracted backlog turns into revenue on schedule while the borrowing used to fund it stays serviceable.