Author: @PineAnalytics
Compiled by: AididiaoJP, Foresight News
Summary
Collector Crypt is in the business of locking graded trading cards in a vault, issuing a token for each card on Solana, and selling them through gacha machines. A gacha machine is a fixed-price pack that randomly dispenses one vaulted card. Buyers can hold it, redeem it for the physical card, or sell it back to the company for cash. The company's revenue comes almost entirely from pack sales, with the marketplace, redemption fees, and eBay auction tools serving as minor add-ons.
In Q3 2026, the company generated $433.5 million in revenue and $25.4 million in gross profit, both record highs, up 8% and 12% quarter-over-quarter respectively. But despite the impressive totals, the trajectory is not promising: June revenue was $212 million, the best month in the company's history; July was $157.3 million, August $141.7 million, and September $134.5 million, declining month by month. Q3 only surpassed Q2 on the strength of June.
Who the buyers are and where they buy from has also changed. Jupiter began selling packs on July 13, and 30-day active wallets grew from 7,051 to 9,127. But the money didn't follow: daily revenue per wallet dropped 39%, sales of the $1,000 Pokemon gacha machine fell 73%, while packs priced at $2,500 and above accounted for 39% of all sales. In June, the top 100 buyers were pouring money into the newly launched $2,500 gacha machine; by July, their purchases had fallen to just a quarter of June's level, and most of September's top 100 buyers had never bought before July.
Courtyard is the second-largest on-chain card platform, with Q3 revenue down 57% from Q2; Collector Crypt's revenue over the same period was 3.3 times its own, despite trailing in Q1. Tokenized inventory reached $55.1 million. In September, two first-edition Charizards sold via peer-to-peer transactions for $600,000 and $825,000 respectively; Coinbase teased its own packs on September 28. CARDS fell 14% to close at $0.195; internal token unlocks were more than 50 times buybacks.
Record Quarter, But Declining Month by Month
Q3 revenue was $433.5 million, of which $433.1 million came from pack sales. Buyback and redemption expenses—cash paid to players who sell back cards they've pulled, plus the cost of redeeming or buying back cards on the secondary market—totaled $408.1 million, or 94.1% of revenue. This ratio has been stuck between 93.9% and 94.3% for four consecutive quarters, keeping gross margin steady at around 6%: 5.9% in Q3, 5.7% in Q2, and 6.1% in Q1. Simply put, this is a business that relies on scale with a fixed take rate.
Sales peaked at $13.5 million on June 15; the full month of June reached $212.2 million, 2.1 times May's figure. July was $157.3 million, August $141.7 million, and September $134.5 million, down 37% from June. Gross margin moved in the opposite direction, rising from 5.4% in July to 6.7% in September, the highest monthly figure in a year.
The company's two previous record months, October 2025 and January 2026, were each followed by a pullback and then a new record within two months. But June's record is now three months old. The first nine days of October averaged $5.4 million per day versus $4.5 million in September—the first signal that the model may still hold.
New Launches and Channels: 16 Gacha Machines, Three Sales Entry Points
July added 12 sports gacha machines and a $1,000 One Piece gacha machine; the $5,000 Celestial gacha machine announced in July didn't actually go on sale until August 2; watches launched in August; Riftbound and Dragon Ball launched in September. Channels expanded in tandem: the Jupiter web app went live on July 13; deployment to Base on August 30 brought almost no business; Jupiter Mobile launched on September 7; integration with Robinhood Chain on September 25, where one contract accepts USDG for pack sales and burns the card NFT when players sell back. Cumulative revenue crossed $1 billion on August 13. No enforcement actions, lawsuits, or service disruptions were identified this quarter.
The $1,000 Tier Falls Out of Favor, the $5,000 Tier Becomes the Big Spenders' New Favorite
In Q3, packs priced at $2,500 and above sold $169.6 million, accounting for 39% of pack sales; in Q2 that was $82.9 million, or 21%, and nearly zero in Q1. The $1,000 tier fell from $132.1 million to $53.4 million; packs at $100 and below actually rose 23% to $101.1 million. All the growth came from the top of the price table, while the middle tier was hollowed out.
The shift came down to two gacha machines. The $2,500 Mythic sold $83.4 million, flat versus Q2; the $5,000 Celestial sold $81.3 million. The $1,000 Grail fell from $129.5 million to $35.3 million. The money moved up, not out: of the 1,137 wallets that bought $1,000 packs in Q2, 662 returned in Q3 and directed 49% of their spending to $2,500 and $5,000 packs, totaling $102.8 million, or 61% of sales at those two price points.
Beyond the high end, the number of gacha machines grew from 24 to 40, mainly in new categories: by quarter's end, watches had sold $17 million and the $1,000 One Piece gacha machine had sold $17.9 million. Non-Pokemon gacha machines accounted for 24% of pack sales, up from 19%.
Jupiter Brings in New Users, Big Spenders Provide the Capital
30-day active wallets grew from 7,051 on June 30 to 9,127 on September 30, of which 5,581 were new within the window. Wallet active days rose 76%, from 43,400 to 76,600; but daily revenue per wallet fell from $9,100 to $5,600. Average spending per active wallet—the line in the chart—dropped from $9,300 in June to $4,800 in September.
Every pack payment carries a memo indicating which entry point made the sale. Jupiter captured 35% of sales in its first week, when it offered up to $100,000 in rewards; 30% in the second week, then stabilizing at 6% to 9% in most weeks thereafter. Jupiter Mobile, launched September 7, did not change this pattern. For the full quarter, Jupiter sold 11% of packs by value, Collector Crypt's official website accounted for 80%, and card payments accounted for 1.6%.
What Jupiter really changed was the number of wallets. Sample data shows 2,043 wallets bought via Jupiter and 2,387 via the official website; among Jupiter buyers, 92% had not bought in Q2, versus only 49% of official website buyers. The two groups bought packs at similar price points, with average order values of $161 and $174 respectively, but Jupiter buyers retained for shorter periods: of the 935 wallets that made their first purchase via Jupiter in July, 34% bought again in August or September, spending 42% of July's total; among the 204 wallets that made their first purchase via the official website, those figures were 48% and 137%.
The money is still concentrated in a few hundred wallets, but no longer the same few hundred. In Q3, 21,876 wallets bought packs on Solana, double Q2's 10,978; the 548 wallets that spent over $100,000 contributed 83% of the money, of which 85 spent over $1 million, and the top 100 accounted for 54% of the quarter, down from 60%.Look deeper, and the top of the leaderboard has already changed hands, and the chart shows why. In June, the top 100 wallets spent $136 million that month, 69% of June's total, and their purchases almost all followed one product: the $2,500 Mythic gacha machine. That gacha machine only really went on sale on June 10, and on several peak sales days it moved $6 million to $8 million a day.
June was also "June Gacha Games" month, a mission-plus-leaderboard promotion with about $500,000 worth of free card packs. But that explanation does not hold up: this group's spending fell from $9 million a day in mid-June to $2 million a day on June 27, while the event did not end until three days later; there was no cliff on July 1 either ($1.8 million on June 30, $1.9 million on July 1).
By September, 54 of the top 100 were still buying, spending $32 million in total. September's own top 100 spent $85 million, 65% of the month; only 18 of them had been in June's top 100, and 39 had never bought before July.
One gacha machine brings in one cohort of big spenders, the next brings in the next. During the handoff, monthly sales can still hold steady around $134 million.
Inventory doubled this year, with One Piece accounting for a quarter
Tokenized inventory means every card that enters the vault, whether it belongs to the company or the user, corresponds to a token, and as of September 30 it was worth $55.1 million at current insured value, up 47% from June and 2.3 times the level at the start of the year. Pokemon was $31.4 million, 57%, up 21%; One Piece was $13.6 million, 25%, up 62%; sports cards were $3.5 million; other categories were $6.5 million. The Q1 report said the figure was $18.6 million on March 31; the same holdings now read as $25.7 million because appraisal values rose.
The gacha business's own treasury, including cards used in the gacha machines, stablecoins, and SOL, was $30 million, versus just $17.1 million on June 30: of that, cards were $26 million and USDC was $3.8 million. What cannot be seen on-chain is where the quarter's $25.4 million in gross profit went: the visible stablecoin balance at quarter-end was that $3.8 million, plus $300,000 in the company's multisig wallet, while the rest of that multisig wallet's assets were CARDS. Cash was swept somewhere the public ledger cannot follow.
Two "Charizards": the marketplace is only 2%, but the high-end market has torn open a gap
In Q3, the marketplace did $8.6 million in volume, while card pack sales were $433 million, only 2% of the business. Of that, $2.8 million was still the company buying graded cards for its own vault.
The real change came in September, when the remaining transaction mix looked different. In the week of September 16, a first-edition Charizard sold for $600,000 through the platform's swap tool, avoiding the 10%-plus fees typically charged by auction houses or eBay; on September 23, a second sold for $825,000, which the company said was the largest tokenized card transaction ever.
We traced the second one: a PSA 10 1999 first-edition Base Set Charizard, held in the vault at a $1 million insured value, was transferred to the buyer's wallet, while $808,500 went to the seller and $16,500, exactly 2%, went to the company's wallet. Those two days accounted for 36% of September's $4.3 million in volume, and September itself was twice August and 18% of the full quarter.
Competitor Beezie completed its own million-dollar Charizard trade the next day, a long-distance response.
The opportunity is right there. The high-end physical card market runs through auction houses and eBay, where selling from PSA Vault costs 7% to 16%; a vaulted card on Solana changing hands carries a royalty of about 2%, and the vault holds $55 million in cards, versus just $24 million at the start of the year. Collector Crypt has no foothold there yet, and two trades do not prove anything, but every card that comes in through a pack can later be resold without leaving the platform. The number worth watching is marketplace volume as a share of card pack sales: it is 2% now, and any quarter above 5% would mean inventory has begun to monetize a second time.
Courtyard shrinks, Coinbase looms
Courtyard is a platform on Polygon that led the category in Q1, with Q3 revenue of $7.6 million, down from $17.8 million in Q2. Its monthly revenue peaked at $6.7 million in May, a month earlier than Collector Crypt; by September it was only $2.2 million, down 67%, while Collector Crypt fell 37%. Among the five on-chain gacha platforms tracked by Messari, total activity in July was $284 million, 17.5% below the June record, with Collector Crypt accounting for about 55%. The decline was category-wide, and Collector Crypt fell the least.
Other players are either small or new: Beezie moved to Solana in May; NFW.fun launched on July 30 and uses Collector Crypt's graded cards as deposits. The truly important new entrant has only been announced, not launched. On September 28, Coinbase teased physical-backed Pokemon card packs in its app, vaultable and shippable, but with no date, no chain, and no pricing. Coinbase Wallet has partnered with Courtyard since 2024; whether this product is built on the existing platform or competes against them is the biggest question of Q4.
The physical market has not shrunk: Q3 grading volume rose 11%, and U.S. search interest for "pokemon cards" is near its April all-time high, so Collector Crypt's 8% growth underperformed the broader hobby market for the first time; in Q1 it had been growing at triple speed.
CARDS: a quarter circulating, unlocks are more than 50 times buybacks
CARDS is a fixed-supply token with no fee switch and no voting rights. The supply cap is 2 billion, with no further issuance, and it can only shrink if the company burns tokens. Holders get monthly snapshots, receive free card packs based on holdings, are weighted in airdrops and Gacha Games points, and can enter Card Club; but they do not share in the $25.4 million the company earned this quarter. The token's only claim on the company is buybacks, nothing else.
Who holds the 2 billion matters more than the price. According to the company, 475 million, or 24%, is circulating. The foundation holds 752 million in unlocked treasury, and the community rewards pool has another 290 million; on-chain, the company's Squads multisig holds 1.08 billion CARDS in one account, 55% of supply. Team, advisor, and seed round allocations total 445 million, locked for the first year and now unlocking on the 29th of each month through July 29, 2027. The next unlock is October 29, 59 million: 32.5 million for the team, 7 million for advisors, 5 million for the seed round, and 15 million for the community pool. CoinGecko counts 895 million as circulating, so its market cap is nearly twice the company's figure; the difference lies in the community pool and vesting wallets.
Buybacks are real, but small. A bot funded by the gacha pool has been buying CARDS in the Raydium pool since April 19, and the company then burns them, while also burning CARDS from the fees the treasury earns as a pool liquidity provider. The bot spent $567,000 in Q3, buying 3.3 million CARDS at an average price of $0.17; in Q2 it was $218,000, or 2.2% of quarterly gross profit. On August 29, 22.5 million CARDS were burned, 5.4% of circulating supply, the largest single burn, bringing total burns to 25.8 million, or 1.3% of the cap. Put the two flows side by side: the bot retires about 1.1 million CARDS a month, while the vesting schedule releases 59 million a month, roughly 55 times the former. At quarter-end prices, that is $11.6 million of new supply against $190,000 of buying.Price follows the sales curve, not buybacks. CARDS hit a peak of $0.319 on June 19, four days after the sales peak; on June 24 it fell below the $4 target price given by Arthur Hayes' family office (reports we have not verified said the fund sold $1.92 million the next day), and on June 29 another 28.8 million tokens unlocked; on July 27 it fell to a low of $0.119. It closed September at $0.195, down 14%, then rebounded to $0.32 on October 11. The main trading venue is Raydium, followed by KuCoin and OKX.
Annualizing the quarter, gross profit is about $100 million per year. On September 30, market cap was $93 million based on company circulating supply and $174 million based on CoinGecko; that is, the token price is equivalent to one to two years of gross profit in which it has no share, while more insider unlocks are coming over the next year. What will determine holders' next four quarters is not revenue, but when burn volume can approach the 59 million tokens per month that the vesting schedule dumps onto the market.
Data Sources and Statistical Methodology
Financial Statements
Revenue, costs, pack sales, user counts, inventory, and treasury data come from Collector Crypt's on-chain financial statements, compiled by Blockworks, read on October 10, 2026. Revenue refers to funds paid to purchase packs: USDC on Solana and card payment rails, and USDG on Robinhood Chain starting September 25. Cost items include buybacks paid to players, cards redeemed at 85% of insured value, and graded cards the company bought in the secondary market. Gross profit is revenue minus the above costs, using the same cash basis as the Q1 report. This data source restated earlier quarters: Q1 2026 revenue here is $146.3 million, versus $146.9 million at the time. Inventory is recorded at current insured value, so past quarter-end figures change with appraisal prices.
On-Chain Accounting
We recalculated USDC inflows to the three receiving wallets, which were $432.3 million in Q3 versus $432.4 million in the statements; June 25 was not included in our records due to a query timeout. The share of each sales entry point comes from the memo on each purchase, with sampling covering every day of the quarter, 205 hours in total, involving $40.9 million of purchases. Wallet grouping counts the payment wallet in each purchase, excluding card payment wallets; repeat purchase rate tracks wallets whose first purchase in July went through a given entry point. Active wallets refer to wallets that bought packs in the past 30 days; spend per wallet-day is daily sales divided by the number of buyers that day. We verified the August 29 burn (22,485,690 CARDS, a single transaction) and directly read supply, treasury, and vesting accounts. The $825,000 Charizard trade has been traced to an on-chain transaction on September 23; the $600,000 trade is corroborated by the marketplace series' single-day total on September 16.
Third-Party Data
Courtyard: uses the DefiLlama revenue series, the same source used in the Q1 report. Category totals: Messari. Grading volume: GemRate monthly review, excluding Beckett. Search interest: Google Trends United States. CARDS price and CoinGecko-reported supply: CoinGecko. Token allocation and vesting: the company's tokenomics page and documentation. Launches, promotions, and plans: company briefings and posts. Robinhood Chain mechanics: DefiLlama adapter issue. Coinbase teaser, Beezie trading, and the Maelstrom incident: media reports. Unlock quantities come from company announcements, as relayed by CoinGecko and CoinMarketCap.