Open in interactive viewer → charts, metric popovers & call review

Bit Digital's Buyback Ran Into Its Own Articles of Association

A strategic-asset story tells four levers, delivers three, and defers the fourth to Q1 — while a 59% drawdown does the talking
BTBT · Earnings Call · 2026-11-12

A Quarter That Was Mostly Marks, Not Operations

Bit Digital reported third-quarter revenue of $27.1 million, down from $32.1 million in the second quarter, on nine-month revenue of $87.1 million, up 7% year-over-year. Consolidated net income was $0.6 million — but $28.5 million landed in shareholders' laps, or $0.08 per diluted share, after allocating $27.9 million of losses to non-controlling interests. Getting to that number required $84 million of gains on digital assets, $10.7 million of gains from remeasuring the convertible-note derivative, $17.1 million of unrealized gains on digital assets held through a third-party fund, and a $80.8 million induced-conversion expense at WhiteFiber that the company reported as a discrete item. The market, for its part, was not grading the marks: the stock is down 54.4% over the last 90 trading days and 58.8% from its September 22 high of $1.86, against a $276 million market cap. That gap between accounting income and share price is the entire tension of this call, and management knows it.

The Mix Is Changing, And That Part Is Real

There is genuine operating news. Cloud, colocation, and staking were about 91% of consolidated revenue this quarter versus 74% a year earlier. That is a real remix, driven by AI infrastructure and by the deliberate sunset of the mining business — mining revenue was just $2.0 million on 28.4 Bitcoin, down from $2.4 million on 32.3 coins. The exit announced quarters ago is finally visible in the P&L rather than just in the rhetoric. “Some of that is deliberate, particularly our continued exit from Bitcoin mining. Some reflect the timing and variability of cloud services revenue.” — Samir Tabar, Chief Executive Officer · 2026-11-12 The caveat matters. Cloud came in at $18.1 million versus $23.8 million, a decline that included $3.4 million of termination-fee revenue from the company's initial customer — meaning recurring cloud revenue fell further than the headline did. Colocation grew to $5.4 million from $4.7 million, and the NC-1 first phase contributed roughly $1.1 million of net income. Native staking is now the company's number-one ranked theme, at $1.0 million on 474 ETH. But CFO Erke Huang quietly repriced that engine: “native staking rewards had been decreased in the past few quarters. Now it's about 2.2% native staking” — Erke Huang, Chief Financial Officer · 2026-11-12 — the 3.4% average yield reported for the quarter is flattered by the liquid-staking and collateral arrangements layered on top.

Four Levers, One Charter Problem

Here is where the story turns. On the August call, the buyback was framed as a near-certainty of intent: “We think the 40% or sometimes even 43% discount to NAV is unacceptable and makes no sense” — Samir Tabar, Chief Executive Officer · 2026-08-13, with the program sitting "on our menu" as "a very attractive dish." Twelve weeks later, the dish has been sent back to the kitchen for a legal opinion.

The board cannot authorize a program on its own. Our articles of association require shareholder approval before we can implement a repurchase program, including approval of how the purchases would be made. We recognize the appeal of buying back shares at the current discount to NAV, which is currently about 39%.

Samir Tabar, Chief Executive Officer · 2026-11-12
The articles of association is a new keyword for this company, and a revealing one — a strategy question collapsing into a corporate-charter procedural. Pressed on what discount would justify selling ETH to fund buybacks, the answer was: no formula, still evaluating. So the repurchase program remains a puzzle where the pieces are individually described but never assembled: shareholder authorization, source of funds, and the trade-off of giving up native staking rewards to retire shares. The WhiteFiber lever has shifted too. In August, management committed not to sell: “we have already committed to the markets that we will not be selling down our white fiber shares this year.” — Samir Tabar, Chief Executive Officer · 2026-08-13 Now the door is ajar for 2027 — “potentially next year, we may sell down the shares and use that cash to buy Bit Digital shares” — Samir Tabar, Chief Executive Officer · 2026-11-12 — with explicit emphasis that any sale would be measured to avoid pressuring WYFI's stock, and that there is "no math around that" target stake. The fourth lever, an in-house "alpha-oriented business," was teased by two different analysts and declined twice. “I would refrain from commenting on that into details... will provide more color, I believe, in Q1 next year.” — Erke Huang, Chief Financial Officer · 2026-11-12 Investors are being asked to underwrite a lever that does not yet have a description.

The Leverage Behind the Treasury

What is disclosed is the mechanics. The company borrows dollars against liquid-staked ETH at 5.5%, lends to WhiteFiber at 9.5% with a 1.1x MOIC over 180 days, and now carries $100 million outstanding. That is a genuine spread trade and it is also leverage on a volatile collateral — collateralized liquid ETH stood at 64,000 ETH-equivalent as of October 21, up from 19,000 in September after the Galaxy repayment. The balance sheet reflects the strain: liabilities were 56.9% of assets in the latest filing, up more than 44 percentage points year-over-year, while capital expenditure reached $176 million, up 114% year-over-year and free cash flow less stock-based compensation was negative $132 million. Reported operating income was negative $113 million in the prior quarter, a reminder that the $28.5 million of attributable net income is a mark-to-market artifact, not an operating result.

What Would Actually Re-Rate This

Two things, and neither is within management's unilateral control. First, WhiteFiber executing: all 40 megawatts at NC-1 are now operational and billing against an $865 million ten-year contract, a $300 million secured facility is expected to fund this month, NC-2 and NC-3 in Yadkin County have closed, and remaining performance obligations stand at $729 million. If WYFI's market cap re-rates, the sell-down-to-buyback loop becomes arithmetic rather than aspiration. Second, the tape. That is the uncomfortable part. The global keyword record shows the entire AI-datacenter complex de-rating over the last 90 days — data center capacity and high bandwidth memory among the biggest decliners, with MARA, IREN, CORZ, HUT, WULF and CIFR all appearing in the negative-ticker columns. price of bitcoin fell roughly a quarter over 30 days. Bit Digital's controlling asset is levered to precisely the theme the market is currently marking down, which is why the discount to NAV persists even as the operating story improves. The honest read of this quarter: Bit Digital is no longer a Bitcoin miner and not yet an ETH treasury with a functioning capital-return mechanism. It is a company with a credible second asset, a levered first asset, and a governance bottleneck standing between a 39% discount and the action shareholders keep asking for. Revenue at 5.2x is not the question. The question is whether the board can convert conviction into a share count reduction. On that, the call offered a process description and a promise to look harder.