Pump.fun earns $18.6 million a week from memecoins that Talos finds rarely recover

Pump.fun's protocol kept about 35.5 cents of each fee dollar in the week to Oct. 7 and about 32.9 cents over 30 days ^[18][19]. The week was a strong one. Spread evenly, the 30-day protocol total comes to about $14.2 million a week, and the latest seven days ran roughly 32% above that ^[20][21].

Talos's exchange-listed sample is the survivors' end of the memecoin universe ^[5]. The survivors still did badly. The median coin peaked about 17 days after exchange trading began ^[6] and took a median of about 370 days to hit Talos's collapse line, a 95% fall from that peak ^[7]. In median terms the climb lasts under three weeks and the fall about a year. Measured against first-day price, the survival rate in the return sample is 3.3% ^[23]. On major Solana memecoins, wallets holding at least $1 dropped to 7% of their peaks or less ^[9], and about two-thirds of the Solana-era coins Talos examined never staged a meaningful second rally ^[10].

Pump's revenue needs a transaction somewhere on the platform and nothing more ^[11]. A trader who sells a fading coin and moves into another generates another fee-producing transaction, CryptoSlate notes ^[11], and roughly a third of each fee dollar stays with the protocol ^[19]. I think the two data sets fit that description. Holder counts on the big Solana coins sit at single-digit percentages of their peaks ^[9] while protocol revenue runs above its monthly pace ^[21].

Where the protocol's share goes is Pump's choice. PUMP burns took about 45% of protocol revenue in both the week and the month ^[22], under a pledge to buy and burn the token with part of designated revenue for a year starting in April ^[14]. The other $33.9 million or so of the week's fees was split among creators and liquidity-related recipients ^[26][12]. The burn ties PUMP to activity across the whole platform ^[12]. According to CryptoSlate, it does little directly for someone holding a separate memecoin, whose recovery still needs demand, enough liquidity to sell, and distributions large enough to offset the loss ^[16].

Alon Cohen, Pump.fun's co-founder, makes payouts the centre of the defence. "In time, Pumpfun will vastly outperform the social media industry in user payouts & rewards," he said ^[28]. In the 24-hour figures he gave, creator fees were about 76% of the $4.46 million paid out and Holder Rewards about 16%, with Callout Rewards the remainder ^[15][25]. Spread across more than 140,000 recipients, that day paid at most about $32 a head ^[24]. Holder Rewards of $730,000 were also below the roughly $1.2 million a day that burns averaged in the week to Oct. 7, though the windows differ ^[15][27].

If new launches keep bringing in fee payers, Pump's revenue holds whatever happens to older coins. If attention and liquidity leave first, the platform's income falls with total activity; CryptoSlate's unresolved test is whether rewards can offset token losses before that happens ^[17]. Cohen is right only if payouts grow large enough to cover holder losses. The evidence so far supports the first case, in which the platform earns on turnover and holders of any single coin carry the price risk alone. That view is wrong if weekly protocol revenue falls below the 30-day average of about $14.2 million and keeps sliding while launches continue ^[20].