Pump.fun Pulled $18.6M in a Week While 81% of Memecoins Fell 90%

Pump.fun booked roughly $18.6 million in protocol revenue in the seven days through Oct. 7. Over the same stretch, 81% of a tracked set of memecoins sat at least 90% below their all-time highs.

That's the whole story in two numbers. One platform printing cash. An entire asset class bleeding out. And the space between those two facts is where the memecoin economy gets uncomfortable.

The 17-Day Peak

Rewind to early October. Talos ran a survival analysis on 150 memecoins and a return comparison on 151. Every asset in the sample needed pricing on at least one centralized exchange. That bar already filters for relative winners, since most launchpad coins never get listed anywhere.

Even inside that stronger cohort, the pattern was brutal.

The median token peaked about 17 days after exchange trading began. Seventeen days to the top, then the fade. Talos defined a collapse as a 95% drop from the eventual peak and put the median time from high to that threshold at roughly 370 days.

Here's the part that stings. Only five of the 151 coins in the return sample were still above their first-day price. Five. Out of 151.

Users vanished with the price. Active addresses holding at least $1 fell to no more than 7% of their respective peaks across major Solana memecoins. About two-thirds of the Solana-era coins in the study never staged a meaningful second rally.

Meanwhile Pump's machine kept turning. Traders paid about $52.5 million in fees over the seven days through Oct. 7. Roughly $18.64 million of that accrued to the protocol. Zoom out to 30 days and fees hit about $184.5 million, with protocol revenue around $60.7 million.

Western media missed this. Here's what happened overnight. The platform doesn't need any single token to recover. It just needs transactions happening somewhere across its system.

Who Gets Paid

So who actually captures that flow?

Pump's fee structure splits trading income across the protocol, creators, and liquidity-related recipients. The PUMP token has its own route through buybacks and burns. DefiLlama recorded about $8.45 million of PUMP burns over seven days and $27.29 million over 30 days. Pump committed part of designated revenue to buying and burning PUMP for a year starting in April.

That supports PUMP's price. It does nothing for the person holding some other token that already gave back 90% of its value.

Co-founder Alon Cohen pointed to the payout side. More than 140,000 users collectively received about $4.46 million in a recent 24-hour window. That breaks into $730,000 in Holder Rewards, $330,000 in Callout Rewards, and $3.4 million in creator fees.

"In time, Pumpfun will vastly outperform the social media industry in user payouts & rewards," he said.

Maybe. But look at where the money lands. Creator fees go to the people behind tokens. Callout Rewards go to promoters and contributors. Holder Rewards apply to participating coins, not every Pump-launched asset. Three buckets, three different groups. The buyer who entered at the peak of a dead coin isn't automatically in any of them.

That's the fault line. A trader can dump one fading coin, rotate into a new launch, and generate another fee on the way through. Churn is the product. Pump earns whether or not any individual token recovers. The trader still holding the original bag earns nothing until demand comes back, and demand usually doesn't.

My read: Pump.fun isn't really a launchpad. It's a fee engine that happens to host launches. That's not illegal. It's just not the deal most buyers think they're signing up for.

The three reward programs are smart engineering. They spread the appearance of distribution without touching the core math. A holder can collect distributions and still be deep underwater if the underlying coin falls faster than the rewards pile up.

The 370-Day Clock

Watch the unlock schedule.

PUMP buybacks create demand and burns cut supply. But the token carries its own market risk and grants no contractual claim on Pump.fun revenue. Scheduled insider unlocks can add supply even while burns pull it out of circulation. Two forces shoving in opposite directions on the same chart.

The real question is whether rewards scale fast enough to matter. If distributions grow large enough to offset declining token values, staying invested after the first speculative rush could start to pencil out. That would be new. If trading keeps migrating to fresh launches faster than rewards build up in older ones, Pump keeps converting churn into revenue while the traders supplying that activity stay stuck below break-even.

Asia moves first on a lot of this. The rotation speed, the appetite for new launches while old ones rot, the willingness to keep buying, that behavior shows up on trading desks in Tokyo and Seoul before it shows up anywhere else. What's happening on Solana right now is the cleanest version of it we've seen yet.

The number to watch is 370. That's the median time Talos measured for a memecoin to fall 95% from its peak. If the next cohort of launches compresses that timeline, Pump's revenue holds up just fine. Churn doesn't need survivors.

It needs buyers.