Jupiter (JUP): buybacks hold the floor, the fee structure caps the ceiling

TL;DR

The gap between narrative and revenue does weigh on JUP, but what mainly holds it down is the perp fee structure plus shrinking fees overall, not a missing perp product. Buybacks hold the floor, the narrative gap caps the ceiling, and in the near term JUP remains a high-beta SOL proxy.

• Of the four new products, only Vaults takes a real fee. Baskets charges a 0% basket fee, Portfolio V3 and Pulse are pure interface; Vaults opened its private beta today.
• Perps are a JLP return product, not a JUP cash-flow product. Perps are 46.2% of gross fees, 75% of which goes to JLP; the protocol keeps $2.04M a month and holders get only $76K. The aggregator is only 9.0% of gross fees but 75.2% of holder revenue.
• Fees are down 84.6%, protocol revenue only 66.6%. Monthly gross fees went from about $115M to $17.68M, protocol revenue from about $15.33M to $5.12M — the take rate rose from 13.3% to 29.0%. The draft merged these two into one sentence.
• The buyback rate is 2.3% to 3.5%, not one number. At "50% of protocol revenue" it is $2.56M a month, 3.5% of market cap; at the holder revenue in the draft's own fee table it is $1.65M a month, 2.3% — and the latter is exactly the Litterbox purchase for the month of May.
• Vaults cannot take over from perps. Matching perp protocol revenue takes $490M of TVL at a 5% net take for Jupiter, and $2.45B at 1%.
• Cards are a distribution play, not a revenue line. $209M is Collector Crypt volume; by the estimate in an earlier piece on this site, Collector Crypt's own true gross margin is only about 5.9%.

This piece is built on an external draft. Prices and fees are as of 2026-11-11, with fees on a 30-day basis. Every computable figure was rechecked; the last row of the section 2 table, the split in section 3, the two buyback bases in section 4, the 1% case in section 6 and the four findings in section 8 are mine.

1. Four new products: extending aggregation from trading to the balance sheet

The four are the same logic copied sideways, not a new revenue engine.

Only Vaults takes a real fee, and it only opened its private beta today. Decrypt · Solana Compass

The market treated it as a one-day event: on November 10 JUP ran from 0.2321 to 0.2558 (+10.2%), yet it is still down 28.6% over 30 days. A narrative can light the fire; revenue has to keep it going.

2. Fee structure: 46% is gross fees, not value capture

The last row is not in the draft; it is the total minus the sum of the four rows, see section 8.

Three points:

1. Perps are 46.2% of gross fees and 39.8% of protocol revenue, but only 4.6% of holder revenue. 75% of perp fees go to JLP as real return and Jupiter keeps 25%: $8.17M × 25% = $2.0425M, matching the table's $2.04M. Eco
2. The aggregator is only 9.0% of gross fees but 75.2% of holder revenue (1.24 / 1.65). What actually funds buybacks is the least glamorous-looking business, routing.
3. Of the $2.04M perps leave the protocol, only $76K (3.7%) reaches holders.

So "missing perps" is not a valuation gap: perps are a JLP return product, not a JUP cash-flow product. What the four new products lack is not perps but a fee layer that improves retention.

3. Fee scale: fees down 85%, revenue down 67%

Phemex/Coinotag · DefiLlama

The draft says "now $5.12M a month, about $212M annualized, down about 85% year over year" — $5.12M is protocol revenue, $212M is gross fees annualized ($17.68M × 12), and 85% is the fall in fees. Split apart, one more fact shows up: revenue fell less than fees, and the take rate more than doubled. That may be the fee mix shifting toward lines where Jupiter keeps more, or partly a difference in basis between the two dates — the draft gives nothing that separates the two.

4. Valuation support: the buyback is real, but its rate has two numbers

One cross-check that holds: market cap over price gives about 3.317B circulating, and 3.46B unlocked minus 130M burned is 3.33B, less than 20M apart.

The buyback policy is 50% of protocol revenue. CryptoSlate But the draft actually carries two bases for the buyback amount:

Holder revenue is 32.2% of protocol revenue, not 50%. The draft cites Litterbox as support: 113.5M JUP / $24.45M accumulated through May 18, 2026, with 8.39M JUP, about $1.65M, added that month, and says this "roughly matches the estimate above" — but what it matches is $1.65M, not $2.56M. Coinness The draft does not explain the gap: the buyback base may not be all protocol revenue, or the data source may classify holder revenue more narrowly. This piece lists both and states conclusions as a range.

Supply discipline is real: 130M JUP burned, net emissions cut to zero, the Jupuary airdrop postponed indefinitely. The remaining 3.4B has been postponed, not removed. BlockBeats · TradingView

5. Price: over 30 days it is SOL beta

"JUP has no bid" holds over 30 days and not over 7. Over 30 days the two move almost together, so the price is mostly SOL beta — the narrative gap explains why it does not rise, not why it falls. It is still 86.9% below the February 2024 high of $2.00, and 93.2% above the February 2026 low of $0.1358.

6. Vaults cannot take over

The fee caps are a management fee of up to 10% a year plus a performance fee of up to 50% of gains, shared between Jupiter and the curators. Perp protocol revenue is about $24.48M a year; for Vaults to match it:

5% is an aggressive assumption: it has Jupiter alone netting 5% of AUM, already half the management-fee cap before the curators take their share. 1% is closer to what yield vaults commonly charge, and then it takes $2.45B.

The draft estimates early beta TVL at $50M to $150M and Jupiter revenue at $1M–$4M a year — figures that imply a net take of 2.0%–2.7%, not the 5% used above, and come to 1.6%–6.5% of current annualized protocol revenue. This is an inference from the product design, not disclosed data.

So the value of Vaults is not on the 2026 income statement but in three things: staking JUP is the ticket into the beta (which creates demand to hold it), keeping capital circulating inside Lend and the aggregator, and staking out the future curator economy. The risks are plain: no loss coverage, standard redemptions take up to 30 days, curators can pause deposits and withdrawals; on October 30 Lend already queued USDC withdrawals over tight liquidity. Solana Compass

The checkpoint is Solana Breakpoint on November 15–17: a public launch schedule, each curator's actual fees, and the first TVL figures.

7. Cards: a distribution play, not a revenue line

The category is growing: onchain card gacha spending reached $324M in June 2026, a fourth straight monthly record; Collector Crypt alone did $209M, or 64.5%; Solana carries 64% of global onchain card volume. Bankless

The product is live: Jupiter Gacha entered beta on July 13 with Collector Crypt, every pull a PSA-graded, slabbed physical card, tokenized and tradable on Jupiter. The Block Offerbook accepts graded cards as collateral at fixed rates and fixed terms with no price liquidation; within hours of launch $120K of cards backed about $75K USDC in loans (62.5% loan-to-value), and loan fees go straight into JUP buybacks. Solana Compass

The business logic has four parts: acquisition (Pokémon and One Piece bring in collectors who are not crypto-native), frequency (opening packs is high-frequency), a closed loop (opening, secondary trading through the aggregator, and collateral in Offerbook, three fees on one asset), and differentiation (rivals attack on trading; Jupiter defends with asset breadth). It is the same distribution playbook as bringing in tokenized stocks through Securitize.

Two layers of froth need taking off, though. The draft already removes the first: $209M is Collector Crypt volume, not Jupiter revenue. The second comes from an October card study on this site: 93.91% of card gacha volume flows back to users as buybacks, and Collector Crypt's own true gross margin is about 5.9%. At that scale, $209M of volume corresponds to roughly $12.33M of gross profit — and that is Collector Crypt's, of which Jupiter takes a distribution cut.

8. Four places the draft does not reconcile

1. The fee table total is not the sum of its rows. The four rows of gross fees sum to $16.68M against a stated total of $17.68M, $1.00M apart; protocol revenue $3.681M against $5.12M, $1.439M apart; holder revenue $1.316M against $1.65M, $334K apart. In the gap, protocol revenue ($1.439M) exceeds gross fees ($1.00M) — a business line cannot earn more revenue than fees, so the total and the rows are not on the same basis. The draft's shares (46.2%, 39.8%, 4.6%, 9.0%, 75.2%) are all divided by the total; the arithmetic is right, but part of the denominator is not listed. Section 2 here shows the gap as its own row.

2. The buyback has two numbers. $2.56M is the policy basis; $1.65M is the holder revenue in the fee table and also the Litterbox pace in May. The draft says Litterbox "roughly matches the estimate above", but what it matches is $1.65M. So "buybacks at about 3.5% of market cap a year" in its conclusion should read 2.3%–3.5%.

3. Revenue and fees share one sentence. "$5.12M a month, about $212M annualized" — the first is protocol revenue, the second gross fees annualized; "down about 85% year over year" is the fall in fees, while revenue fell 66.6%.

4. The two Vaults estimates use different take rates. Matching perps uses 5%; the early beta revenue of $1M–$4M implies 2.0%–2.7%.

9. What this piece cannot compute

• Which products make up the fee table gap. DCA, limit orders, Launchpad, Offerbook and Gacha could all be in it; the draft does not break it out.
• The real buyback base. The draft gives no formula for which revenue the 50% applies to.
• Actual Vaults fees and the Jupiter share. Those wait for Breakpoint.
• Whether Q3 2025 and now share a data source and basis. Part of the doubled take rate may come from a basis difference.
• The Litterbox pace after May. The draft's data ends on May 18.

Conclusion. The valuation floor has support; the ceiling is capped for now. Buybacks at 2.3%–3.5% of market cap a year, zero net emissions and a postponed Jupuary make a floor that does not fall far; but monthly gross fees 84.6% below Q3 2025, perp revenue that depends heavily on SOL volatility, and four new products that add no near-term revenue mean the multiple cannot expand. JUP is priced more like "SOL beta plus a 2.3%–3.5% buyback" than the next Hyperliquid.

Until Vaults launches publicly and reaches more than $300M of TVL (which even at a 5% net take equals only 24.4% of current annualized protocol revenue), or monthly gross fees return to $20M, the gap between narrative and revenue will keep holding back multiple expansion. Four things to watch: monthly gross fees, perp protocol revenue, Vaults TVL and fee disclosures, and the weekly Litterbox purchase pace.

Data: DefiLlama and the links in the text; prices and fees as of 2026-11-11, fees on a 30-day basis. Research notes only, not investment advice. DYOR.