Original title: How Pumpfun is killing the trenches
Original author: Sparsity @sparsitydata
Editor's note: In the third quarter of 2026, 3.05 million meme coins were issued on Pump.fun, but only 24 ever reached a market cap of $10 million. According to data from the on-chain research team Sparsity, issuance in the fourth quarter of 2024 was similar, at 3.16 million, yet 207 coins reached the same market cap threshold. The number of coins issued has returned to around the previous cycle's peak, but for every 100,000 coins launched, the number that can rise to a $10 million market cap has fallen from about 6.5 to 0.8.
When the meme coin market weakens, the community often points the finger at launch snipers, teams that mass-launch tokens, and KOLs who shill a coin and then leave. Sparsity traced the problem one level up, looking at where Pump.fun’s revenue actually comes from. According to its analysis, about 95% of the platform’s protocol revenue comes from tokens still on the bonding curve or with a market cap below $250,000. A bonding curve is a pricing mechanism used during the early stages of a new token launch, where buying and selling move the price according to preset rules. At this stage, every buy and sell incurs a 1.25% fee: 0.95% goes to the platform and 0.30% to the creator. Even if the price doesn’t move, one round of buying and selling costs about 2.5%. By token age, roughly three-quarters of the platform’s revenue comes from tokens launched less than an hour ago.
This fee structure puts the platform, creators, and buyers at odds. For the platform and creators, a token generates fees as long as people trade it soon after launch. Buyers, on the other hand, still need someone else to buy in at a higher price later to make a profit. Creators can copy a popular token or launch dozens at once; as long as a few generate trading volume at launch, they earn money. Sparsity found that the 25 wallets that launched the most tokens launched 761,000 in total. Fewer than 1% reached the threshold to “graduate” from the bonding curve, yet these wallets collectively earned $128 million.
Shilling rewards widened this divide, too. Pump.fun used to reward shillers based on the trading volume they generated, and concentrated buying can more easily move low-market-cap tokens. On October 10, the platform said it would factor in whether copy traders made money and reduce rewards for shilling low-market-cap tokens. But over the following three weeks, according to Sparsity, the share of rewards going to the top 20 shillers who mainly promoted low-market-cap tokens rose from 68% to 94%. Among the 14 shillers the authors tracked separately, the median price change of promoted tokens was up 6% one minute after a post, down 48% after an hour, and down 74% after a day. For each buyer, selling early makes sense. But as more people rush to sell, fewer are willing to buy and keep holding, making it harder for a new token to survive beyond its initial hype.
The article ends with a proposal for fixed-price fundraising. Buyers subscribe at the same price; if the minimum raise isn’t met, they automatically get refunded. Once the target is reached, a contract creates a liquidity pool, and the platform charges a fee only when the raise succeeds. This could weaken the advantage of sniping the launch and give the platform one less way to earn from failed mass launches, but splitting funds across wallets to bypass limits and selling immediately after launch would remain problems. What this proposal changes is when the platform gets paid. Today, a token can last just one hour and still make the platform money, as long as it trades frequently enough. Delaying fees until a raise is complete would give the platform a more direct reason to care whether a launch attracts enough buyers.
Here is the original article:
“The market is bad on-chain.” “All the KOLs are dumping on us.” “We’re still idolizing people who snipe launches with bundled trades.”
These problems are real, but it makes little sense to look for their cause at the very bottom of the system. Holding culture has collapsed at the front lines of meme coin trading. No one wants to hold anymore. People copy popular tokens to ride their attention, launch dozens of tokens at once, use bundled trades to snipe the launch, and shill tokens to collect rewards because these things make money. The problem lies further up, with the platform that controls this market. The biggest meme coin launchpad is paying for the very behaviors everyone complains about, turning meme coins into a market of infinite supply and disappearing demand.
Today, everyone wants KOLs to be accountable for their shills, and that’s a good thing. But holding shillers accountable isn’t enough if the platform paying them never has to answer for anything.
pump.fun makes money from tokens that eventually go to zero
Over the past 30 days, pump.fun accounted for 92% of launches and 84% of trading volume across meme coin launchpads on all chains. It determines what makes money in this market.
Ninety-five percent of its protocol revenue comes from trading tokens still on the bonding curve or with a market cap below $250,000. On the bonding curve, every buy and sell incurs a 1.25% fee: 0.95% goes to pump.fun and 0.30% to the creator. That means a round of buying and selling costs 2.5%, even if the price doesn’t move at all.
Protocol revenue from bonding-curve and low-market-cap tokens
The money is made when a token’s market cap is only a few hundred thousand dollars; what happens to its price after that doesn’t matter.
The same pattern holds when we look at token age. Three-quarters of this revenue comes from tokens launched less than an hour ago, and 87% from tokens launched less than a day ago.
Revenue by token launch age
pump.fun makes its money in the first hour after a token launches.
The tokens that surge the most contribute almost none of the revenue. Fifteen of the biggest winners reached a combined market cap of $1.4 billion at their respective peaks, yet brought pump.fun only about $1.5 million in revenue. By contrast, tokens still trading on the bonding curve or with a market cap below $250,000 brought in $506 million. Put simply, 1,000 tokens that stall at a $20,000 market cap make it more money than one token that reaches a $100 million market cap.
Fee revenue from top-rising tokens versus small tokens
Those 15 big winners contributed less than 0.33% of the revenue brought in by small tokens.
So the platform’s product is designed around increasing token launches. Every incentive on the platform pushes up the number of launches, because the more tokens are launched, the more the platform earns.
Four incentives
There’s a reason every behavior criticized on social media exists on this platform. All four drive more trading, and every trade pays pump.fun a fee.
How four behaviors generate platform fees
Bundled sniping, mass token launches, shilling, and PvP all ultimately generate the same fee for the platform.
Bundlers snipe the launch block because it’s cheap
Bundlers exist because tokens start out with tiny market caps and little liquidity. Spread a few SOL across dozens of wallets and buy together in the block where a token launches, and you can grab a large share of the supply—then sell it to you, the later buyer. The lower the initial market cap, the cheaper this is, and pump.fun sets that initial market cap.
Creators mass-launch tokens because creator fees are paid based on trading volume
People launch tokens one after another because creator fee rewards incentivize launching. Whether a token lasts an hour or a year, creators receive 0.30% of every trade. So the most profitable strategy is to launch as many tokens as possible, collect fees from early trading volume, and move on to the next one. The top 25 launchers by token count launched 761,000 tokens. Fewer than 1% “graduated” from the bonding curve, yet they earned $128 million.
Token launches and revenue from top-launching wallets
The top wallet launched 171,195 tokens, earning an average of about $200 per token.
Mass launches occasionally produce one or two big winners. The best-performing token from each of two wallets reached market caps of $122 million and $327 million. But big winners are a byproduct of all that trading; trading volume itself is what earns the money.
Shillers promote low-market-cap tokens because rewards are paid based on trading volume
Shilling rewards gave people a new reason to promote tokens. The platform turned shilling into a source of income. Since August, it has paid $29.6 million in shilling rewards to 51,115 wallets, with the rewards heavily concentrated. Ten wallets took 16%, while the median amount received by a winner was $13.
Distribution of shilling rewards
Those 10 wallets received an average of about $485,000 each.
Rewards are paid based on the trading volume generated by each shill, and the smallest tokens are the easiest to drive trading volume in. In August, 40 of the 50 top-earning shillers promoted tokens with an average market cap below $100,000.
Average market cap of tokens promoted by top shillers
Twelve of the top 50 shillers promoted tokens with an average market cap below $10,000.
pump.fun has heard the feedback. On October 10, the platform said future rewards would depend on whether copy traders made money, and that rewards for shilling low-market-cap tokens would be reduced. Over the following three weeks, the share of rewards among the top 20 earners going to shillers whose promoted tokens had a median market cap below $100,000 rose from 68% to 94%. Weekly rewards also increased by 53%.
Shilling rewards and the share of low-market-cap promotions before and after the rule change
After the rules changed, the total rewards increased, as did the share going to shillers promoting low-market-cap tokens.
Most of the cost falls on copy traders. We tracked 14 shillers, 10 of whom promoted tokens with a median market cap below $100,000. Their copy traders lost a combined $926,000 one day after the shills. Based on the median price change of the promoted tokens, they were up 6% after one minute, down 48% after one hour, and down 74% after one day. Shillers get paid whether token prices rise or fall.
Copy-trading profits and losses for 14 shillers
Only one of the 14 had copy traders who made money overall, earning $10,000.
Performance of tokens promoted by each shiller one day after the shill
For every shiller, the median price change of the tokens they promoted was negative one day later.
People ask why KOLs keep promoting tokens that lose their followers money. This is a big part of the reason: they’re being paid to do it.
PvP makes each user trade more
PvP (player versus player) works because, without new money coming in, experienced traders make money from newcomers. These formats are designed around that dynamic. Trades execute faster, each user makes more trades, and every trade incurs a fee. On June 30, pump.fun itself removed one of these formats: the Tokenized Agent mode. The platform said at the time that too many ways to launch tokens were creating “predatory player-versus-player behavior.”
Infinite supply, disappearing demand
These incentives have been around long enough to split the market in two.
On the demand side, you rush in to buy tokens and, on average, lose money. Only a tiny number of wallets make any meaningful profit. On the supply side, you launch tokens and get paid with almost no risk.
There’s no barrier to becoming a supplier. Launching a token takes one click, and you already know how to do it. Make a copy of a popular token in seconds, wash-trade it to make it look active, then buy aged wallets by the hundreds or thousands. This playbook gets easier every month.
So we think people who might otherwise have held tokens for the long term will increasingly move to the supply side. Launching tokens pays; holding them doesn’t. Every new player who figures this out becomes another launcher competing for fewer and fewer buyers. That’s how holding culture breaks down. No one collectively decided to stop holding. The platform made holding a losing trade.
The performance of these new tokens reflects the result. pump.fun is now launching about as many tokens as it did at the market’s peak in late 2024. By our count, 3.16 million tokens launched in Q4 2024, and 207 reached a market cap of $10 million. In Q3 2026, 3.05 million launched, but only 24 reached that market cap.
Weekly token launches and the rate of big winners
Weekly launches have climbed back above 300,000. For every 100,000 tokens launched, the number of big winners fell from 6.5 to 0.8.
Some will point out that trading volume on the bonding curve has been rising since June. That’s true, but combined volume on the bonding curve and PumpSwap has fallen from about $25 billion to $30 billion per month in Q1 2026 to about $13.8 billion in October (bonding curve and PumpSwap). And pump.fun makes money from trading volume. Buyers, meanwhile, make money from the big winners, and there are now only one-ninth as many.
Survival rates tell the same story. pump.fun has launched 24.4 million tokens, but only 312 currently have a market cap above $100,000. That’s about one in 78,000 tokens, or 0.0013%.
Token survival rates across launchpads
Most launchpads issue only a fraction of the tokens pump.fun does, but their token survival rates are higher.
This is the first market we know of where supply has grown this quickly while demand has shrunk. If the platform weren’t paying for this behavior, it would never have reached this scale.
We’re responsible, too
pump.fun built this system, but we supply the trading volume that keeps it running. Every incentive described in this article requires someone on the other side of a trade to generate revenue. Most of the time, that someone is us.
We reward the behavior we complain about
We say bundlers are the problem, then follow them to see what they’ll do next. A developer copies a popular token and walks away with the fees, and someone posts a thread explaining how clever the move was. Someone shills 20 tokens a day, and people follow them because they’re always first to call them out—whether copy traders make money becomes secondary. Of the 14 shillers tracked above, only one had copy traders who made money overall. All copy traders combined lost $926,000 one day after the shills.
We paid for all this twice. Once with trading volume, which pump.fun and creators profit from. And again with attention: the next copycat token and the next shill are both vying for ours.
Holding is no longer a strategy
A year ago, the play was to spot a token early, weather the noise, and hold on. Now, the median token promoted by a shiller is down 74% a day later, so the people who make it out are the ones who sell first. That’s rational for each of us, but when everyone does it, nobody is left to catch the big winners. If no one is willing to hold a token past its first hour, it can’t reach a $10 million market cap. And that one hour is where three-quarters of pump.fun’s revenue comes from.
What would happen if we changed our behavior?
You don’t need pump.fun’s permission to do these things. Hold tokens you genuinely believe in; buy and sell less often just because you see another chart. Judge shillers by whether their followers make money, not how often they post. Stop spreading copycat launches and bundled sniping as clever plays. Give your trading volume to launchpads that make money from tokens that survive. None of these actions alone will change the incentive structure. But every trade we stop contributing to this system is one less trade it earns money from.
How to change the incentives
We’re offering these ideas at a high level; engineers would need to design the specific solutions. Each one aims to shift revenue toward the period after a token’s first hour.
Raise initial market caps, deepen initial liquidity, and make liquidity grow over time, so the pool expands as trading volume increases.
Share fees with holders and liquidity providers (LPs), and have creator fees unlock gradually over the token’s lifetime.
Make shilling rewards depend on whether copy traders hold the token. Pay shillers based on whether their followers still hold the token and are in profit days later.
Approaches like these could keep this market alive, and pump.fun has the ability to try them. It holds about $2.84 billion in on-chain assets, and controls the entire process. In the long run, a market that keeps producing big winners would make it more money than one that keeps running out of buyers. It chooses extraction because that earns more today, but it isn’t sustainable.
Even so, this might not be enough. These changes would raise the costs for token issuers. But suppliers would still have stronger tools than demand-side participants, and that wouldn’t change.
What a token launch without a bonding curve could look like
All of the approaches above still rely on the bonding curve. The next step would be to try removing the bonding curve entirely.
In this model, everyone pays the same price. Without a bonding curve, there’s no benefit to getting the first transaction into a block, so sniping loses its advantage. The creator sets a minimum and maximum raise. If the minimum isn’t met within a set time—for example, 24 or 48 hours—the smart contract automatically refunds everyone, so buyers lose no money if the launch fails. Once the fundraising cap is reached, the contract creates a liquidity pool and burns or locks the LP tokens, so the creator can never access the funds raised. Each wallet has a strict contribution limit, such as 1% of the total raise, preventing a single whale or bot from taking half the token supply.
The platform earns revenue only when a raise is fully subscribed. It makes money from successfully launched tokens; even if 1,000 launches fail to meet their targets, it earns nothing.
The platform switches from charging per trade to charging only when a raise is fully subscribed
The launchpad’s fee trigger shifts from each trade to each fully subscribed raise.
This would eliminate bundled sniping and launch-block sniping at the start of a bonding curve, as well as draining the liquidity raised. It would also remove the incentive to launch 50 tokens in a row, none of which reaches its target, and still profit.
This model has flaws worth spelling out. Per-wallet limits only work if wallets are hard to fake, and aged wallets are cheap right now. Without some kind of anti-Sybil mechanism to stop one person from participating with multiple wallets, bundlers can simply spread their buys across 100 wallets. A fixed price also won’t stop people from selling on day one. It can only make the starting line fairer.
What happens next
Meme coins in their current form have run their course. We think they can return only if the launch model itself changes—for example, by adopting an approach like the one above, where the platform makes money when a token launches successfully.
As long as a launchpad can profit from every token that eventually goes to zero, meme coin trading will remain dormant, and holding will still be a losing trade. Something new will emerge, but it won’t be meme coins, and it won’t happen soon.
Nothing in this article constitutes investment advice. We do not hold $PUMP or any other tokens mentioned here. Data is approximate and current as of November 9, 2026. Retrieve and verify the underlying data before use.