Gemini's Narrower Q3 Loss Leans on Crypto Marks and Cost Cuts

The headline numbers

Gemini's third-quarter net loss fell 63% to $58.6 million from $159.5 million a year earlier. Spending dropped faster than revenue, and that's the part that deserves your attention. But the headline number owes a lot to accounting choices that don't reflect the core business. If you stop reading there, you'll come away with a rosier view than the rest of the release supports.

Revenue slid 19% to $40.9 million, according to an SEC filing made alongside its earnings release. The loss came to $0.46 a share, compared with $6.67 in the same quarter of 2025. Per-share figures only mean something if you know the share count underneath them, and that's where a careful reader should check first. Operating expenses fell 30% to $119.9 million, the third straight quarterly decline, driven by lower pay and a 75% cut in sales and marketing. Cutting marketing that sharply looks good on a spreadsheet. The cost usually shows up later, in fewer new users rather than in a bad quarter.

Trading volume came back, with a caveat

Platform trading reached $4.5 billion. That's a 73% drop from $16.4 billion a year ago, but an 18% rise from $3.8 billion in the second quarter. Gemini credited the gain to more institutional trading in Zcash and expanded market-making by an affiliate. Year-over-year volume comparisons are always sensitive to what the market was doing at the time, and the release doesn't break out how much of the decline reflects market conditions versus the company's own decisions. I'd want that split before calling this a turnaround.

Here's the thing: about $1.2 billion of spot volume, or 26%, came from trades between outside customers and Gemini's own market maker. That's a sizable slice of activity that doesn't reflect two unrelated parties meeting in the market. Market makers do useful work by keeping quotes available, so those trades aren't fake. But when a platform's own market-making arm sits on one side of about a quarter of spot volume, the number tells you how busy the house is as much as how busy the customers are. Exchange revenue fell 47% to $13.4 million. (This is the part most people overlook.) That's the line that shows whether volume is actually turning into money.

What the adjusted numbers are doing

Adjusted EBITDA, which adds back interest, taxes, depreciation, amortization and stock-based pay, showed a loss of $11.6 million, against $74.0 million in the second quarter. Gemini pointed to smaller credit card loss provisions and lower compensation. It also cited its marks, the gains and losses on the value of its crypto holdings and of crypto loans from related parties. Marks added $25.7 million this quarter after subtracting $25.0 million in the second, a $50.8 million swing. Companies that hold volatile assets often see those swings dominate a quarter's results without reflecting anything the business actually did, and that's why the marks line deserves a close read.

Strip those marks out of both quarters and the picture looks less rosy. The adjusted loss would have been about $37.3 million, compared with roughly $49.0 million in the second quarter. That's still an improvement, but a smaller one than the headline suggests. The operating story is improving more slowly than the EBITDA line implies, and that gap matters if you're tracking the company over several quarters. Sustained improvement looks different from a single favorable mark.

Under standard accounting rules, the operating loss actually widened to $79.0 million from $76.9 million. Lower revenue and a $12.4 million jump in stock-based pay, which Gemini tied to annual equity bonuses recognized in July, outweighed a $14.9 million cut in other operating expenses. Stock-based pay is a real cost even though no cash leaves the building. When annual grants land in one quarter, they can make that quarter look worse than the trend. The bonus timing explains the jump, but it doesn't make the expense go away.

Cards, cash and the European retreat

Services revenue and interest income made up 59% of net revenue, up from 47% a year earlier. My read is that this is the more durable part of the mix. Revenue that doesn't depend on the mood of the trading floor tends to be steadier, and that's a better foundation than volume spikes. Credit card revenue rose 82% to $15.5 million, while card charge-offs climbed to $11.5 million. Charge-offs are debts a lender gives up on collecting, and card losses like these often arrive with a lag after a book of accounts grows. Gemini linked most of that to accounts tied to a fraud event it disclosed in the first quarter. Cash and cash equivalents stood at $194.1 million on Sept. 30, down from $252.2 million at the end of 2025. That's a drop of about $58 million since year-end, and it's the number that shows how much room the company has if the next quarter disappoints.

The release has little to say about Europe. It mentions only the February plan "to wind down operations in the United Kingdom, European Union, other European jurisdictions, and Australia," a retreat that came with a cut of roughly a quarter of the company's staff. It doesn't mention the sale of the UK and EU business to stablecoin payments firm KAST for under $10 million, which CoinDesk reported on Thursday. A sale at that price is a small exit number, and it suggests the business wasn't worth much to a buyer. If you're working in this space, that gap is worth noticing. Companies pulling back from whole regions usually want investors to see the mechanics, and here the details arrive through a press report instead.

What to watch next

The cost cuts are real, and nobody should wave them away. But cutting staff, marketing and pay can only carry a business so far before it needs revenue to grow on its own. Cheaper to run, but not yet growing. A reset, not a rebound.

That said, the picture is more mixed than the cost story suggests. The most useful signals are the ones that don't depend on accounting judgment: how much volume comes from outside customers, whether exchange revenue grows along with volume, and whether card losses settle once the fraud-linked accounts are behind the company. Those are the lines I'd check first. Watch the marks too, because a quarter that looks better only because the value of its holdings moved in the right direction isn't evidence of a better business.

If you're using Gemini as a benchmark for the sector, the lesson is less about this one company than about the gap between reported results and operating results. Similar platforms typically lean on adjusted metrics when the underlying numbers are weak, and the spread between the two is where skepticism pays off. What this means for you depends on how closely you read the footnotes. The next release will show whether the operating numbers catch up to the adjusted ones, and that's the test I'd apply.