Pump.fun Buys Back Half Its Revenue. The Price Is Still at Its Low

Pump.fun does what most token models only promise. Since April 2026, 50% of the revenue from Bonding Curve, PumpSwap, and Terminal goes into buying PUMP off the open market. Of everything bought back, 60% leaves the network permanently. 360B tokens, 36% of the maximum supply, are already gone.

PUMP trades at $0.001293. It launched at $0.004. The all-time low is $0.001133.

That gap is the entire lesson. A buyback moves revenue into the order book. It does not create a reason to hold. In our public audit the model scored 51/100, rating BB, and the block that pulled the score down is the one carrying the heaviest weight in our methodology: Token Product Linkage, 50 out of 100 at 40% of the total.

This isn't a story about a failing business. Pump.fun collects $63.6M in fees a month. Lifetime, the platform has earned $1.126B, with another $652.6M on PumpSwap. The revenue is real, the buyback is real, and the token still sits at its floor.

For teams building from the UAE, the framing carries past the price chart. Both the VARA rulebooks in Dubai and the ADGM guidance in Abu Dhabi expect an issuer to describe what a token does and what its economics depend on, not what the treasury intends to buy. A fee-funded buyback is a distribution policy; the question underneath it is still whether anything in the product requires the token.

Measure the buyback against turnover, not against supply

Teams present buybacks as a share of revenue, because that framing flatters the design. Half of all revenue sounds like a commitment nobody can argue with.

The market doesn't price commitments. It prices the size of the bid relative to the size of the flow it has to absorb.

Half of ecosystem revenue produces roughly $1.7M of daily demand for PUMP. Daily trading volume in the token is $112.2M. The buyback covers about 2% of turnover.

Two percent is a signal, not a floor. It tells holders that management is aligned. It does nothing to the price when the other 98% decides to leave. Run the same division on your own model before you write the buyback into a deck: daily buyback bid divided by daily traded volume. If the answer has one digit, the mechanism is a communication tool, and you should stop calling it value capture.

Burning 36% of supply changed nothing about demand

360B of 1T PUMP have been removed from the network. More than a third of the maximum supply is gone, and the price sits at its all-time low.

Scarcity multiplies demand. It doesn't manufacture it. Cut the supply of something nobody needs and you get a smaller quantity of something nobody needs. Our audit puts it plainly in the risk section: it is impossible to achieve scarcity through burning if the token lacks baseline demand.

The corollary matters more than the observation. Burn is a good way to distribute surplus to holders who already hold for a reason. It is a poor way to create the reason. We took that myth apart separately, in why burning tokens doesn't create value.

A revenue-funded buyback weakens exactly when you need it

Here is the design flaw most teams copy without noticing.

Pump.fun's fees depend entirely on demand for memecoins, with no diversification. Demand for PUMP also depends on demand for memecoins, because speculation is the only channel through which the token attracts buyers. Both sides of the model sit on the same cycle.

When that cycle turned, both moved together. Platform trading volume fell from $11.6B in January to $3.65B in June in the half year before the token launched. Over the past year, daily DEX volume fell fourfold. Fees shrank, the buyback shrank with them, and the buyback got smaller precisely in the months when the price needed support.

A buyback funded by revenue that correlates with token demand is procyclical by construction. It is strongest when it is least needed and weakest when it matters. If your fee base and your token narrative share a driver, assume the buyback will disappear at the bottom.

Optional utility is not utility

PUMP has a utility list. Governance votes on platform development and fee structure. Fee discounts in some scenarios. Access to promotions, giveaways, tips for creators, prizes during live streams. Earn programs on Kraken, Binance, BingX, KuCoin, Gate.io.

Not one of them is required to launch or trade a memecoin, which is the reason anyone opens the product.

The staking loop makes the circularity visible. Rewards are paid in PUMP, and those tokens are not minted or held in a reserve. They are bought from the market out of fees. A holder locks the token to receive more of the token, funded by users who never had to touch it. No demand enters the system from outside, and the token is temporarily locked rather than used.

This is the distinction we spend most of our audit time on. A list of things a holder may do is not utility. Utility is an action inside the product that cannot be completed without the token. Pump.fun has the first and not the second, and platform growth therefore doesn't convert into token demand. The gap between a growing product and a flat token is its own story: when the product grows but the token doesn't.

Your buyback competes with your own unlock schedule

Market cap is $577.4M against an FDV of $1.42B. The difference is a schedule, and the schedule is about to start.

From July 2026, the team's 200B PUMP and 130B for existing investors begin unlocking. Against the current float of 345.6B, that is 31% and 20% of circulating supply arriving into a market where the buyback removes about 1.2B tokens a day. Insiders hold 3% of the current effective float and 35% of the final one.

Net float is the number that decides whether a buyback does anything. Take twelve months of scheduled unlocks, subtract twelve months of buyback at the current fee run rate, and see which way the balance points. If the schedule wins, the buyback isn't tightening supply. It is subsidizing an exit.

A mechanism nobody can verify won't be priced

Pump.fun publishes no tokenomics documentation. Staking, farming, and buyback parameters change often, and there is no single place where the current version lives. Our own audit carries a note that the section may go out of date at any moment.

That has a price. Governance and control risk scored 32 out of 100, the weakest block in the profile. An analyst who cannot reproduce your mechanic from public sources will not underwrite it, and a mechanic that isn't underwritten doesn't show up in the price no matter how much money runs through it.

Publishing the model is the cheapest fix available to any team reading this.

Five checks before you put a buyback at the center of a model

• Coverage. Divide the daily buyback bid by daily traded volume. Single digits mean a signal, not a floor.
• Correlation. If your fee base and your token demand share a driver, your value capture is procyclical.
• Obligation. Name the action in your product that cannot be completed without the token. If you can't, the buyback is carrying the whole model.
• Net float. Twelve months of unlocks minus twelve months of buyback at the current run rate. The balance is quick to run in our tokenomics calculator.
• Verifiability. An outsider should be able to reproduce the mechanic from your public docs.

When a buyback earns its place

None of this makes buybacks a mistake. It makes them a second-order mechanism.

A buyback works when the token already carries obligated demand and the buyback returns surplus to the people holding it for that reason. It works when the fee base is diversified enough that revenue doesn't collapse in step with the token. It works when the unlock schedule is behind you rather than ahead. In those conditions, converting revenue into supply reduction is a clean way to share the upside of a growing business.

A structurally close case is how STON.fi routes DEX fees into its token: the path from fee to token exists, and it still runs into the same optionality.

It fails when it is asked to substitute for a use case. Pump.fun shows the failure at full scale, with real revenue and an honest mechanism behind it.

The recommendations we gave in the audit follow from that. Put the token inside the product, for example by requiring it for access to the creative studio. Integrate it into new ecosystem services as they launch. Publish the tokenomics. Build the model around what the token does, not around what the treasury buys.

The demand exists. $220M worth of tokens sit locked in protocols, and the platform has a large user base, with 18.6M tokens created on it between January 2024 and June 2026. What's missing is a single required action that turns a user into a holder.

Read the full model, block by block, in our public tokenomics audit of Pump.fun. If you're designing value capture for a fee-generating product right now, that is the stage where this is cheap to fix.

We run these reviews out of Dubai, and this is the most common gap we see in token models built here: a fully funded buyback in the deck and no obligated demand in the product. It is cheap to correct while the model is still on paper, and expensive once the unlock schedule and the listings are fixed.