A token buyback works when recurring revenue pays for it and the purchases outpace the new supply the project is releasing. Anything less slows dilution at best. The practice is spreading but concentrated: buyback spending reached a record $638 million in 2026 through August 25, according to Allium Labs data reported by the Financial Times and ForkLog, and two protocols accounted for nearly 90% of it.
Who decides this
The CFO and the board own the decision. The CFO is accountable for the funding source, the runway left after the program and the execution mandate given to any trading desk. The board is accountable for whether the program is the best use of that money, for what is promised to holders and for the legal sign-off. A buyback is a capital allocation decision with market-conduct risk attached. It should not be approved as a marketing announcement.
Should the buyback be funded from revenue or from the treasury?
Fund a recurring buyback from a fixed share of protocol revenue, and use treasury funds only for a one-off, capped purchase. Revenue funding ties the program to usage, so it shrinks when activity falls instead of draining runway. Treasury funding spends capital raised for building, which holders and investors can fairly question.
Live programs show three patterns. The table compares them on the dimensions a CFO has to defend.
| Funding model | Published example | Predictability for holders | Strain on runway | Main risk |
|---|---|---|---|---|
| Fixed share of fees, automatic | Hyperliquid routes about 99% of trading fees to its Assistance Fund, per CoinShares (June 2026) | High: rule-based, visible on-chain | Low: no treasury draw | Spending falls with volume; hard to pause without a governance change |
| Annual budget from revenue, adjustable | Aave’s $50 million annual program, executed at $250,000 to $1.75 million a week | Medium: budget known, weekly pace flexible | Medium: committed even if revenue dips | A fixed budget can outrun revenue in a downturn |
| One-off or pilot purchase | Jito DAO’s $1 million pilot buyback, run in four installments over ten days | Low: no ongoing commitment | Depends on size | Reads as price support, not value returned |
The Aave proposal, approved by Snapshot vote in November 2025, is a useful model of the middle path. It names protocol revenue as the source, sets an annual budget and lets a finance committee move weekly volume within a 75% range. The factors listed for that discretion include liquidity, volatility and available revenue. That is the kind of written mandate a board should require before any purchase.
Our view: a project without durable revenue should not run a recurring program at all. If revenue cannot fund it, the honest question is whether the treasury money belongs in runway or product instead. Our guide to crypto treasury management after the raise covers that runway test.
Is the buyback larger than the supply you are releasing?
Size the program against unlocks and emissions, not against market cap. The most useful single test is a coverage ratio: repurchase dollars divided by the value of newly unlocked and emitted supply over the same period. CryptoSlate’s January 2026 analysis argues this is the number that decides whether a program works. Above 1, net supply contracts. Below 1, the buyback only slows dilution.
Most programs fall well short. The same analysis estimated that Optimism’s program, which directs half of Superchain revenue to monthly buybacks, would come to about $9.1 million a year. That is roughly 0.7% of the token’s fully diluted valuation at the time. Set against the unlock calendar, that is a signal of intent, not a supply change.
The supply side is large. Tokenomist counted about $97.4 billion in token unlocks across 2025, of which roughly 19% went to insiders. A CFO should model the coverage ratio month by month for at least the next 24 months. Months where a large cliff lands are the months a buyback will be judged.
Should repurchased tokens be burned, held or paid to stakers?
Burning is the cleanest choice for a token whose case rests on scarcity. Holding in treasury keeps optionality but invites the question of when the tokens return to market. Distributing to stakers turns the buyback into a yield source, which changes what the token is.
Each choice has a live precedent. CoinShares reports that Hyperliquid sends repurchased tokens to an inaccessible address, and that validators formally recognised this as a burn by vote in December 2025. CryptoSlate notes that Optimism’s repurchases go to treasury rather than being burned. Uniswap’s fee switch, approved in late 2025, uses redirected fees to buy and burn the token.
The destination should follow the token’s stated purpose. If the token’s reason to exist is governance, burning shrinks the voting base. If it is a claim on activity, paying stakers is honest about that. Our guide on value accrual through fee switches and staking sets out how to choose. Whichever destination the board picks, it belongs in the published allocation and the circulating supply calculation.
How should the program execute, and what does it disclose?
Execute by rule, in small slices, and publish the rule before the first purchase. Gradual execution limits market impact. Jito ran its first program in four time-weighted installments. Hyperliquid’s fund places daily limit orders below the market price, according to CoinShares.
Disclosure matters as much as execution. Holders should be able to see the funding rule, the wallet that buys, the destination of purchased tokens and a regular report of spend and average price. Undisclosed discretion is where a buyback starts to look like price management.
As of October 2026, we know of no adopted token-specific buyback safe harbour of the kind US equities have. For listed shares, SEC Rule 10b-18 gives issuers a safe harbour from manipulation liability if each day’s repurchases meet four conditions on manner, timing, price and volume. The volume condition caps daily purchases at 25% of average daily trading volume. That rule is written for share repurchases, not tokens. We still use its four dimensions as a design checklist for a token program.
In the EU, MiCA’s Article 91 prohibits transactions that give false or misleading signals about a crypto-asset’s supply, demand or price, unless carried out for legitimate reasons. The article itself does not mention buybacks. Whether a given program is legitimate depends on its facts, and that analysis sits with counsel, as of October 2026.
When should the board say no?
Say no when revenue is cyclical and the program is sized on a good quarter, when the coverage ratio is far below 1, or when runway is under the board’s minimum. The evidence that markets reward buybacks is mixed. As reported in October 2025, Messari found in March of that year no clear evidence the market rewards these programs.
Aave’s own pilot shows the exposure. Per TokenLogic data cited in the same report, the DAO spent about $24.7 million on roughly 106,000 AAVE at an average of $234. At the time of reporting those tokens were worth about $23 million, a paper loss of about 7%. The pilot was not a failure. It shows that a buyback is a position, and positions can lose money.
A buyback also cannot substitute for usage. If the product is not growing, the money is usually better spent on the product.
Before approving any program, we would ask the CFO for three numbers in one page: monthly revenue over the last 24 months, the unlock and emission calendar for the next 24, and runway with and without the program. If the board cannot see all three side by side, it is not ready to decide. The same page becomes the baseline for every quarterly review that follows.
What goes wrong: a launch plan that assumed liquidity
The board approves a buyback because the token is weak. The plan assumes the purchases will be absorbed by a normal market. In practice the order book is thin, and the program becomes the main bid.
From there the problems compound. Each purchase moves the price more than the model expected, so the budget buys fewer tokens. Holders learn the schedule and sell into it. When revenue falls in a downturn, the program either drains the treasury or stops, and stopping is read as a signal. CryptoSlate notes that, per CoinGecko, Pump.fun’s purchases were underwater after the October 2025 crash.
The defence is to test liquidity before announcing anything. Measure depth on the venues where purchases will execute. Size weekly spend as a share of real volume, with a ceiling. Write down in advance the revenue level at which the program scales down, so a pause is a rule being followed, not a decision under pressure.
How Alcyone approaches it
Alcyone treats a buyback as one line in a token and treasury plan, not a standalone announcement. Within Alcyone’s fundraising and tokenomics advisory, an engagement starts with a fixed-scope, fixed-fee assessment; for a buyback, that is the point to test revenue, the unlock calendar and market depth before any program is designed. The plan that follows names an owner for every part, from the funding rule to the disclosure cadence. After launch, the same partners report quarterly, which is where the program’s coverage ratio belongs. Clients are under no obligation to use TrustSwap products.