Between January 2024 and June 2026, pump.fun produced more than 18.6 million tokens that traded at least once. Nearly seven in ten of them made their last trade on the same day they were born. Eight in ten were gone within 48 hours. Fewer than one in twenty were still trading three months later (CoinGecko, via CryptoPotato).
The odds of "making it" are worse. A study of 832,941 launches between May and June 2026 found a graduation rate of 0.198%: about two tokens in every thousand reached the point where they leave Pump.fun's bonding curve for an open market (Kamat, arXiv 2607.02823).
Over the same period, the platform collected more than $1.2 billion in revenue (DefiLlama).
Those three facts are the whole story. Almost every token dies. Most traders lose. The house gets paid either way. I build smart contracts for a living, and I've come to believe pump.fun is the single most damaging product this industry has shipped. And if things don't change, it's time to walk away from it.
How the machine works
On pump.fun, anyone can create a token in seconds by filling in a short form. There is no team to vet, no code to audit, and no liquidity to provide. Each new token starts on a bonding curve: an automated price formula where every buy pushes the price up, and every sell pushes it down.
If enough money flows in, the token "graduates" at roughly $80,000 to $100,000 in market cap and moves to a regular trading pool. Almost none get there. The rest simply stop trading, usually within hours.
The key detail is where the money goes. pump.fun charges fees on trades along the curve and on its own exchange, PumpSwap. It does not need any token to succeed. It needs tokens to be launched and traded, as many as possible, as fast as possible. As one analysis of CoinGecko's data put it, the platform is paid whether a token lasts six minutes or six months (Crypto Briefing).
That is not a flaw in the design. It is the design.
Who wins
The platform does, spectacularly. pump.fun has generated about $1.9 billion in fees and $1.2 billion in revenue since launch (DefiLlama). Its best month, January 2025, brought in $137 million (Unchained). In a 2026 job listing for a Chief Legal Officer, offering up to $5 million a year, the company itself said it made over $500 million in profit the previous year with fewer than 100 people (Mitrade).
Much of that money has left the chain. Onchain trackers estimate the platform has sold about 4.47 million SOL, roughly $780 million, mostly through Kraken (Pluang). In late 2025, wallets linked to pump.fun also moved 436.5 million USDC to Kraken; co-founder Noah Tweedale called reports of a cash-out misinformation and said it was treasury management (ForkLog).
Then there was the token. In February 2025, co-founder Alon Cohen dismissed reports of a pump.fun token as wrong (The Defiant). Five months later, the PUMP sale raised hundreds of millions of dollars and sold out in 12 minutes (Cointelegraph). Half of the platform's net revenue now goes to buying back and burning PUMP (DefiLlama). Put plainly: trader losses are recycled into support for the platform's own token.
The three founders, Noah Tweedale, Alon Cohen and Dylan Kerler, say they built pump.fun out of frustration with rug pulls (Wikipedia). That origin story sits awkwardly next to a 2025 WIRED investigation. It linked a developer using Kerler's name to eight tokens launched in 2017, two of which collapsed amid rug-pull accusations, earning that developer up to an estimated $75,000. Neither pump.fun nor Kerler responded to WIRED's requests for comment (Bitcoinist, summarizing WIRED).
Who loses
Everyone else, mostly. In the first half of 2024, Dune data showed 62.5% of pump.fun traders had lost money, while just 311 wallets, under 0.01% of the total, had made more than $1 million (crypto.news). By June 2025, only 30.1% of active wallets finished the month in profit (CoinGecko).
The pattern held into 2026. In March, about half of the roughly 1.4 million wallets trading pump.fun tokens lost money, and around 96% either lost or made less than $500 (crypto.news, via MEXC). Over the six months before that, the top 250 token deployers pulled an estimated $79 million out of traders while launching 194,000 tokens between them (Protos, via MEXC).
Those are wallet counts, and one person can run many wallets. But the shape is unmistakable: a thin layer of insiders and "professionals" at the top, and a large base of ordinary people paying for the game.
An earlier study by Solidus Labs found that of more than 7 million pump.fun tokens with at least five trades, 98.6% had collapsed below $1,000 in liquidity (The Defiant). Not every one of those was a deliberate scam. But a market where 98.6% of products collapse is not a market. It's a lottery with a fee on every ticket.
The first block
If the odds were merely bad, you could call it gambling and move on. The deeper problem is that the game is often decided before a normal person can even click "buy."
On Solana, a token's creator, or a bot watching for new launches, can buy in the very same block the token is created. Pine Analytics found that this kind of same-block sniping is no longer an edge case but the dominant launch pattern on pump.fun. Looking only at the most obvious form of collusion, snipers funded directly by the token's own deployer, it accounted for about 1.75% of all launches. That is a floor: insiders who route funds through a few extra wallets slip past the method (Pine Analytics).
An entire industry has grown up to serve this. "Bundler" tools, advertised openly, let a launcher buy from up to 20 wallets in the creation block, and many launches use between 20 and 50 (The Coin Republic). The supply is spread across fresh wallets so that it looks organic. Then it is sold to whoever arrives next.
By the time a token appears in your feed, the cheapest supply has usually been claimed by software. The retail buyer is not a participant. The retail buyer is the exit.
What it did to crypto
The damage doesn't stay on pump.fun. For two years, it set the tone for what a "crypto app" is. Attention, liquidity, and talent flowed toward launching and flipping tokens rather than building things people use. New users arrived through a casino, lost money, and left believing that this is what crypto is.
The platform's culture made it worse. In late 2024, pump.fun had to shut down livestreaming after creators broadcast animal cruelty, threats of violence, and self-harm to pump their tokens (Decrypt). It was also reportedly geo-blocked in the UK after action by the Financial Conduct Authority (Mitrade).
And the activity it generated proved hollow. When the memecoin tide went out in 2026, daily Solana network fees fell about 84% between January and June, and pump.fun's own graduation rate dropped 80% in three months (DEXTools). Growth built on extraction doesn't compound. It just ends.
The lawsuit
Pump.fun's operator, Baton Corporation, and its three founders are defendants in a federal class action in New York. The plaintiffs estimate retail losses at $4 billion to $5.5 billion, a figure that has not been tested in court (Decrypt).
On August 31, 2026, Judge Colleen McMahon split her ruling. She dismissed the claims that pump.fun tokens were unregistered securities, rejected the theory that trading memecoins counts as illegal gambling under New York law, and cleared Solana Labs, the Solana Foundation and their executives (SolanaFloor). But she let the racketeering (RICO) claims proceed against Baton and its founders personally (Burwick Law).
Nothing has been proven yet. But a federal judge has found the racketeering allegations against the people running pump.fun detailed enough to go forward.
Their defense, answered
Pump.fun's defenders make four arguments. Each deserves a straight answer.
"It's permissionless. Adults choose to play." Choice requires information. A buyer who doesn't know that about 2 in 1,000 tokens graduate, that most die on day one, and that the cheapest supply was often taken by software in the first block is not making an informed choice. Permissionless shouldn't mean rigged.
"They're reforming." To their credit pump.fun has shipped changes, including cashback for traders and BOOST, a launch mechanism that briefly lifted the daily graduation rate to 6.7% in July 2026 (The Block). Better odds are welcome.
"It brought millions of users to Solana." It did. Most of them left with losses, and the activity evaporated once the hype did. Onboarding people into a machine that takes their money isn't adoption. It's burning the users the industry will need later.
"PUMP holders benefit from buybacks." They do, but it's not sustainable. The token's support comes from fees paid by traders who mostly lose. The platform itself states that PUMP carries no right to its revenue (DefiLlama).
What you can do
To be honest, I've barely used pump.fun. This isn't the story of a burned trader. I'm joining the boycott because a company earning hundreds of millions of dollars responds to one kind of pressure above all: people choosing not to use it.
I also want to give credit where it's due. The bonding curve is genuinely a good idea. It gives every token open price discovery from the first trade, with no insider-arranged liquidity. With real protections, such as limits on first-block buys, vesting for creators, and a visible history for every deployer, it could be a strong tool for ICOs and community funding.
That isn't how it's used. People come to pump.fun to find the next WIF, or to ride the pump and get out before the dump. The platform is built around that behavior. It takes a fee on all of it, and gives remarkably little back to the communities that generate it.
Competition won't fix this on its own, because pump.fun is the template. It showed every launchpad that came after it how to make hundreds of millions while not caring what happens to users. It has given a home to serial deployers who pulled tens of millions of dollars out of traders. That lesson is spreading, and it has tainted the whole industry.
I believe enough pressure can change the product. I'm far less convinced it can change the people running it, and that matters most. Products follow incentives, and a boycott changes incentives. But culture comes from leadership, which is why accountability for the people in charge, including in court, matters as much as any new feature.
Crypto's promise was a fairer financial system. Pump.fun took a good mechanism and built a casino on top of it. I'd like to see the bonding curve outlive the casino, and the only way that happens is if we stop feeding it.
Top comments (0)