As reported on November 11, 2026, a study published the previous day by on-chain data researcher Sparsity estimated that pump.fun accounted for 92% of meme-token launches across launch platforms over the preceding 30 days, and 84% of trading volume on those platforms. Those figures measure launch-platform activity—not how often a launched token reached a higher market cap.
Where fees were generated
Sparsity estimated that 95% of pump.fun’s protocol revenue came from trades while tokens were still on the bonding curve or had a market cap below $250,000. Each buy and sell on the bonding curve carried a 1.25% fee: 0.95% went to pump.fun and 0.30% to the creator. A buy followed by a sell therefore cost 2.5% in fees before any price change.
According to the study, roughly three-quarters of that revenue came from tokens launched less than an hour earlier, and 87% from tokens launched less than a day earlier. It said 15 leading tokens with a combined peak market cap of $1.4 billion paid pump.fun about $1.5 million, while tokens on the bonding curve or below a $250,000 market cap paid $506 million.
Launch volume versus outcomes
Sparsity said the 25 most prolific deployers launched 761,000 tokens in total, with fewer than 1% graduating, and received $128 million. The wallet with the most launches created 171,195 tokens, averaging about $200 per token, according to the study.
The study also reported that, since August, promotional rewards had paid $29.6 million to 51,115 wallets. Ten wallets received 16% of those payments; the median payment was $13. The report did not specify the year for that August starting point.
For a comparison across periods, Sparsity counted 207 tokens reaching a $10 million market cap among 3.16 million launches in the fourth quarter of 2024, versus 24 among 3.05 million launches in the third quarter of 2026. Of 24.4 million cumulative pump.fun launches, it said 312 tokens currently had a market cap above $100,000—approximately one in 78,000. That last figure is a snapshot of tokens above the threshold at the time of the study, not a count of every token that ever reached it.