Gemini Space Station, Inc. (GEMI) Earnings Call Transcript & Summary

Thank you, operator. Good morning, everyone, and I appreciate you being here for Gemini's second quarter 2026 earnings call. I'm Ryan Todd, and I serve as Head of Investor Relations at Gemini. With me today are Gemini's Co-Founders, Cameron and Tyler Winklevoss, along with Danijela Stojanovic, our interim CFO. Our second quarter 2026 financial results were released yesterday. On this call, we might make forward-looking statements that could differ materially from actual results, since they rest on management's present expectations, forecasts, and assumptions. Details about the risks, uncertainties, and other factors that might lead these results to diverge appear in our SEC filings. Today's discussion will likewise cover certain non-GAAP financial measures. You can find reconciliations to the most directly comparable GAAP financial measures in the earnings presentation posted on our investor relations website and on the SEC's website. Non-GAAP financial measures ought to be viewed alongside GAAP measures, not in place of them. We'll begin today's call with prepared remarks and afterward take questions; with that settled, I'll hand things over to our founders, Cameron and Tyler.

Thanks, Ryan. Morning, everybody. As Q2 wraps up, we're about to hit one year since Gemini's public listing. Back in September 2025, Bitcoin's price hit $117,000, and the market we were stepping into looked nothing like the one we're in now, with Bitcoin finishing Q2 under $60,000. Even with those challenges, we've kept our heads down, building toward tomorrow's Gemini. Over the last nine months, the Gemini platform has shifted more than it did across the prior decade. During the previous quarter, we discussed rolling out prediction markets in December 2025, and last month, after Q2 ended, we introduced commission-free U.S. stock trading; now that stocks are included, customers in eligible states across the United States can trade 1,000s of U.S. equities at 0% commission, take part in prediction markets, and trade crypto, all within the Gemini app. Since Q2 wrapped, we also started running our derivatives clearing house and settling our own prediction markets contracts, thanks to the DCO license granted by the CFTC in April. This quarter, we stayed focused on quickly advancing our predictions business. Q1 was mostly about products and features, while this quarter we put money into marketplace liquidity and enhancing the overall trader experience on our prediction marketplace. Starting from Q1, we tripled contracted market makers on the platform, rolled out three new maker and taker incentive programs, and grew trading infrastructure through several upgrades to our predictions API. On the product front, we released personalized insights driven by AI and a rebuilt interface featuring unique category-specific experiences across the contract markets we provide. Going forward, we still view predictions as the biggest near-term growth opportunity on the platform, particularly as we head into the upcoming sports season in the second half of the year. We've lately certified a broad set of new product filings and keep delivering new experiences to directly capture that activity. Now that our clearing house is live and clearing our own contracts, which retains those economics in-house, we're focused on bringing on distribution partners for our DCM and expect further updates on our progress soon. Throughout all of this, the focus remains on steadily improving the health of both sides of the order book. Deeper taker flow draws in more makers, which tightens spreads and allows customers to trade with greater capital more predictably. We think this is the next unlock for pushing prediction activity higher at Gemini. With that, I'll hand it to Tyler to cover our recent business highlights.

Appreciate that, Cameron. This is Tyler. During Q1, we set out the plan for Gemini 2.0, and Q2 marked the first complete quarter with that plan in action. As we move forward into the back half of 2026, the business looks noticeably different in shape. We have more products, a leaner cost structure, fewer dollars leaving the company, and revenue that depends less on spot crypto trading with each passing quarter. Danijela will walk you through the financials thoroughly in a moment. My time will go toward what we built during this quarter and what we plan to do with it. We have been operating through crypto market cycles for more than a decade, and the same rule has proven true each time. Bear markets are for building. The first half of 2026 was intentionally centered on building. We directed our dollars and attention toward shipping new products and securing regulated infrastructure instead of overspending on acquisition in an environment where we believed paid acquisition would not deliver the returns we needed. We still think that sequence was correct. It positioned us to finish and bring into operation our regulated derivatives stack. Our designated contract marketplace at the beginning of this year, our derivatives clearinghouse which went live earlier this month, and our FCM application submitted in June. Hardly any firms in this country possess that mix of licenses, and we developed ours internally instead of spending substantial capital to buy it. The stack was designed for more than predictions. The bulk of price discovery in crypto occurs in perpetual futures, not spot. Right now, the overwhelming majority of that volume is located outside the United States. We already provide perpetual contracts to customers in Singapore, so the product and the technology exist and are operating. We hold the DCM and the DCO and have submitted our FCM application. If and when the approvals we require come through, we are prepared and plan to launch perpetual futures for U.S. customers. On the product front, this year, we have significantly broadened what customers can actually trade at Gemini. With commission-free U.S. equities and ETFs launched in July, the platform now provides over 5,000 tradable markets spanning equities, crypto, and event contracts, compared with fewer than 100 a year ago. We also redesigned our advanced trading experience on mobile, so customers can trade straight from the chart or the order book with positions, open orders, and margin all viewable in a single place. We added more assets available for margin trading, and we keep enhancing our developer platform and API trading capabilities for programmatic and agentic traders. All together, the first half of this year was focused on building the products and licensing the structure required to support a financial super app; we feel confident we have never had a better mousetrap, and it is producing stronger engagement, with multi-product users almost doubling year-over-year. Attention now turns to distribution and growing revenue by bringing more customers and more activity onto the platform while maintaining the cost discipline we put in place this year. Having said that, I'll now hand the call over to Danijela, our interim CFO, so she can walk us through the quarter's financial results in more depth.

Thanks, Cameron and Tyler. Morning, everyone, and I appreciate you being with us today. I'll begin with several main points from the quarter, then take you through the results in more depth, and finish with our revised financial outlook for the year. There are 3 items I'd like to put front and center. To begin, total revenue rose 37% compared with a year earlier, reaching $45.5 million, powered by ongoing expansion in services revenue along with our OTC business. That increase occurred even as exchange revenue fell 38% versus the prior-year period, since crypto market volumes stayed under pressure for the entire quarter. Next, services revenue and interest income came to $26 million, an increase of 117% year-over-year, and they now make up 59% of net revenue, compared with 50% in Q1, which reflects both the continuing expansion of the card and staking businesses and a weaker trading environment. Third, our cost restructuring is paying off. Total operating expenses fell 15% quarter-over-quarter to $122.4 million, and headcount finished the quarter at roughly 402, which is 40% lower than our Q3 2025 peak. The complete benefit of the Q1 restructuring is now moving through the cost structure. Moving to revenue, net revenue came in at $43.7 million, an increase of 33% year-over-year. Transaction revenue was $17.8 million, a decline of 15% year-over-year and 26% sequentially. Inside that number, several significant components shifted. Exchange revenue totaled $12.5 million, down 38% year-over-year and 27% sequentially, which reflects ongoing weakness in crypto trading activity. Total spot trading volume dropped to $3.8 billion from $11.3 billion in Q2 of 2025, a 66% decrease set against a 38% decrease in exchange trading revenue. Institutional volume represented almost 90% of the volume decrease, while fee economics kept getting better in both the retail and institutional trading segments. OTC revenue was $4.7 million, an increase of 671% year-over-year. As in Q1, the quarter featured episodic client demand responding to stretches of heightened volatility in the crypto market, which added to elevated volume. The underlying eOTC API program keeps bringing on institutional clients, and given the character of large institutional trades, we expect OTC to stay variable from quarter to quarter. Prediction markets added $0.5 million to transaction revenue, an 18% rise sequentially. That amount is reported net of rebates. As Cameron noted, we introduced new maker and taker incentive programs in the quarter, and the rebates paid under those programs, approximately 20% of gross fees, are reported as contra revenue. Event contracts traded in Q2 increased 93% quarter-over-quarter, and our emphasis during the period was on developing depth in the order book instead of maximizing short-term fee capture. Although this remains an early-stage product, the growth in customer activity encourages us, and we believe the investments we're making now put the marketplace in position for stronger monetization as liquidity and participation keep scaling. Now to services revenue and interest income, which totaled $26 million, up 117% year-over-year and 6% sequentially. Let me take you through the main components. Credit card revenue was $16.2 million, up 231% year-over-year and 10% sequentially. We finished the quarter with 106,000 card MTUs. That's a 165% increase compared with a year earlier, but a 7% decline from the prior quarter, since we keep moving our acquisition approach toward channels with better returns and lower spending. Roughly 5,000 new sign-ups came in during Q2, which shows the intentional step back we took on acquisition marketing in that period. Card receivables stayed about level from one quarter to the next, at $219.6 million, while pre-provision net revenue for the card rose 44% quarter over quarter to $5.5 million, which speaks to the portfolio's ongoing maturation. Staking revenue came to $4 million, an 88% sequential gain and a 50% increase versus a year ago. Three main factors drove that rise. To begin with, about $1.6 million is tied to revenue that in economic terms was earned in the first quarter, yet it was booked in Q2 once we finished moving onto our in-house validator infrastructure and worked out the related data capture problems. Next, now that this migration is done, we function as a principal within the staking arrangement, meaning validator costs are recorded in operating expenses instead of being offset against revenue, which lifts reported staking revenue on a growth basis. Finally, staking adoption kept expanding over the quarter, a sign of solid underlying customer engagement with the product. Advisory fee revenue totaled $2.7 million, matching the two preceding quarters and stemming from the advisory services agreement we continue to hold with a strategic customer, which was signed in Q3 2025. Because the compensatory warrant arrangement tied to that agreement hits the close of its recognition period in July 2026, we anticipate booking the modest amount of advisory fee revenue that remains in the third quarter, and after that we expect this revenue stream to come to an end. Custodial fee revenue reached $0.6 million, falling 67% against both the year-ago period and the prior quarter, driven by weaker average asset valuations on the platform during the quarter plus net custody asset outflows. Assets on platform closed Q2 at $8.4 billion, versus $18.2 billion in Q2 of 2025, which mirrors the drop in crypto asset prices over the last year along with the custody net asset outflows. Interest income was $2.4 million, essentially unchanged sequentially; let's move on to expenses now. Total operating expenses came to $122.4 million, a 15% decrease from $144.5 million in Q1 on a sequential basis and a 24% increase year-over-year. That sequential gain comes from having the entire quarter's worth of benefit from our restructuring moves. Salaries and compensation were $48.2 million, a 26% sequential decline. Stock-based compensation of $20.3 million is included in that figure. Excluding that $20.3 million in stock-based compensation, cash compensation stood at $27.9 million, down 32% sequentially and down 20% from a year earlier, which mirrors the smaller headcount base we have after the workforce reduction. Severance was zero in Q2, versus $6.5 million in Q1. We closed the quarter with roughly 402 people, a decrease from 442 in Q1 and a drop of about 40% from the peak we saw in Q3 2025. Sales and marketing was $8.8 million, falling 54% sequentially and 45% year-over-year. Inside that, brand and performance marketing came to roughly $0.1 million, since we halted broad-based acquisition spending and put our focus on organic and targeted channels. That amount of spending is a result of the tactical choices we made in the quarter, and it shouldn't be treated as a set operating run rate, since future investment will keep being guided by market opportunities and expected returns. Credit card rewards along with promotional and referral incentives came to $8.7 million, a 24% decline sequentially, which reflects lower cardholder spend activity in the quarter. Transaction losses rose from $3.6 million in the prior year to $20.1 million. This increase is mostly attributable to a higher provision for expected credit losses on the credit card portfolio. As we mentioned last quarter, we had previously identified an identity fraud event in Q1 and set up an initial $4.1 million reserve based on what was known at the time. In the second quarter, as our investigation moved forward, we found additional fraud patterns and affected accounts tied to the same Q1 origination cohort. Those account balances keep maturing through the delinquency cycle, and our June 30th reserve reflects our updated estimate under the CECL methodology. Importantly, according to our analysis, we believe the elevated provision is concentrated in this identified fraud-related cohort and does not indicate broad-based deterioration in the underlying credit portfolio. We have taken steps to strengthen our fraud controls and monitoring, and although provision expense will continue to evolve as the portfolio seasons, we expect future provisioning to be driven mainly by the underlying credit performance of the portfolio rather than one-time fraud events.

Expenses tied to technology came in at $18.8 million, a 15% drop from the prior quarter and a 5% rise versus the same period last year, as operating efficiencies mostly counterbalanced ongoing spending on product development. General and administrative costs totaled $20.6 million, 5% lower quarter over quarter and 7% higher year over year. Taken together, tech plus G&A reached $39.3 million in the quarter, staying inside the guidance range for the full year. Turning to the bottom line, net loss stood at $107.7 million, a 19% year-over-year gain compared with a net loss of $133.2 million in Q2 2025. Adjusted EBITDA came in as a loss of $74 million, versus a loss of $59.9 million in Q1 and a loss of $51.9 million in Q2 of 2025. That worsening stems mainly from non-cash mark-to-market losses on Bitcoin holdings, mirroring the drop in Bitcoin prices during the quarter after the $100 million strategic Bitcoin investment we received in May closed. In our view, operating loss gives a clearer picture of operational performance, which rose 18% sequentially, from $94.2 million in Q1 to $76.9 million in Q2, thanks to a full quarter of benefit from our restructuring moves and ongoing expense discipline throughout the business. Monthly transacting users reached 580,000, an 11% increase year over year, but a 2% decline sequentially, as softer crypto market conditions pressured trading activity. Let me wrap up with our outlook. As in previous quarters, no formal revenue guidance is being provided right now. Even though we have broadened our product portfolio and market infrastructure, a lot of these initiatives remain early in monetization. During the first half of this year, our focus was on building capacity through an expanded regulated footprint, new product launches, and marketplace improvements. As attention shifts toward distribution and customer acquisition in the second half of 2026, we anticipate these investments will pay off more substantially over time. Still, how quickly that ramp occurs is hard to forecast, given the early stage of these businesses along with the wider macro and crypto backdrop. Regarding expenses, we are sharpening how precise our outlook is. Cash compensation should still fall 15%-20% against 2025 levels, and that figure excludes stock-based compensation or restructuring charges. Stock-based compensation should still come to $100 million-$115 million for the entire year. Technology and G&A expenses are now projected at between $155 million and $170 million for the full year, tightening the range we gave previously. For marketing, guidance stays the same at 10%-15% of revenue excluding rewards and promotions. Marketing spend during the second quarter was deliberately far under that range, since we prioritized product development, regulated infrastructure, and marketplace capabilities amid a stretch of weaker customer acquisition economics. Heading into the back half of the year, we plan to boost brand and performance marketing while staying inside the full year guidance range we gave earlier. With close to two quarters of restructuring behind us and the full cost run rate showing up in our results, the first phase of Gemini's transformation is essentially complete. Looking back across the last twelve months, we feel we've essentially rebuilt this company from the ground up — widening what we sell and putting a far stricter operating discipline in place; new products keep coming. Stocks have launched, predictions are growing at scale, and the derivatives plumbing now exists; our sense is that we're running leaner than we have at any moment since the IPO. What comes next is delivering on execution and driving greater uptake across this growing platform, all while holding on to the rigor we've put in place. So to wrap up: this quarter shows steady momentum as we keep turning Gemini into a wider multi-product financial platform; revenue rose 37% versus a year earlier even though crypto trading conditions were among the worst we've seen since listing, our operating loss got better for a third quarter in a row, and we kept pushing past the exchange business that defined our history. Plenty of meaningful effort still lies ahead before profitability improves, yet in our view the hard part of building out products and market infrastructure is now mostly behind us. Attention turns to scaling adoption, lifting revenue, and proving out the operating leverage this platform contains. On that note, I'll pass things to Ryan so we can begin the Q&A.

Thank you, Danijela. At this point we'll move into the Q&A portion. Written questions came in yesterday, following the publication of our results for the second quarter. Rather than jumping straight into going analyst by analyst, we'd like to begin with something several analysts asked more than once — specifically about the provision for credit losses this quarter. Since it kept coming up so consistently, we'll take it up first. Regarding the provision for credit losses this quarter, and recognizing that most of these provisions stem from a fraud incident, what do provisions for credit losses look like across the portfolio once you set this incident aside? When it comes to fraud, could you help us grasp what shifted in your assessment between May and today? And how confident are you that we won't see another meaningful provision coming out of this cohort?

Appreciate that, Ryan. Given how large the Q2 provision came in, the question makes sense to us. Let me start by drawing a line between the operational problem and how the portfolio is actually performing on credit. What our work so far indicates is that this was a concentrated identity fraud event tied to one specific Q1 origination cohort, not a broad deterioration across the portfolio. A lot of those accounts were still current when the first quarter closed, and then they moved through the delinquency cycles in the normal way during Q2. As our investigation moved forward over that period, we found more fraud patterns and more affected accounts tied to that same fraud event. Once those amounts matured and we had more performance data, we revised our CECL estimate upward to capture the wider scope of this identified fraud event. What I'd add is that the headline delinquency metrics for this quarter are driven overwhelmingly by that single concentrated cohort, not by any shift in the broader credit characteristics of the portfolio. Take out the fraud-related cohort, and the underlying portfolio has performed largely as we expected. Delinquency went from 3.8% to 9.4%, but almost all of that rise came from fraud. Fraud-related delinquency, specifically, climbed from 1% to 6.1%. Non-fraud delinquency, by contrast, rose only modestly, from 2.8% to 3.3%, which lines up with normal seasoning in a portfolio that is still relatively young. Since then, we've applied what we learned to tighten our fraud controls, onboarding, and monitoring. On timing, these provision balances will move into charge-off status through the usual credit loss cycle as they age through delinquency and reach charge-off. Finally, on the second half of the year, it's clearly difficult to call, and we aren't giving quarterly guidance on provision expense or charge-offs. That said, as with any lending portfolio, provision expense will keep evolving naturally as the portfolio seasons and new information comes in. What we can say is that, based on what we know today, we expect future provisioning to increasingly reflect the portfolio's underlying credit performance rather than this particular fraud event.

To ask again what I asked before, what exactly was different in the way fraud screening works that will keep this from happening down the road?

Right. Regrettably, stopping fraud is not something that stays the same over time. In this space, the people doing bad things keep changing how they operate. So our safeguards, our monitoring, and our underwriting skills must keep changing right along with them. Drawing on what this incident taught us, we've put a number of improvements and safeguards in place, and going forward we'll keep putting money into those capabilities. Apart from any particular incident, we also roll out fresh improvements and safeguards on a regular basis, which keeps us ready to spot and block fraud schemes as they shift. Past that point, we don't believe it's right to talk about the specifics of our fraud safeguards in public, since those safeguards form part of how we prevent fraud; still, we see this as continuous work, not something you fix once.

James Yaro from Goldman Sachs is up next with a question. Regarding prediction markets, could you take us through what sets you apart competitively there, particularly the fact that you've developed them internally? And with rivals putting in heavy investment and coming in with far bigger customer bases, how are you growing your position in that market?

Cameron speaking. A crucial thing to note is that what we have isn't merely a predictions application. What we're really doing is creating a super app that provides predictions, and those are delivered alongside spot Bitcoin, staking, a credit card, U.S. equities, and beyond; we see significant potential for cross-pollinization, and this represents a wider set of offerings rather than just predictions. Not many companies actually possess that kind of extensive breadth; on top of that, with our emphasis on predictions, we've put money into the full stack. Owning the clearing house and the exchange gives us the ability to shape our own future with greater control, plus we can pursue distribution partnerships that keep the flywheel turning. When you put all of this together, the narrative becomes quite distinct from being only a predictions business, and cross-selling to the customer base we already have strikes me as a fantastic opportunity. To illustrate, 50% of those who have made predictions on Gemini also hold a Gemini credit card. That's a solid illustration of cross-pollinization within our super app.