$638 million. Two protocols. 90% of the total.
The Financial Times reported this week that Hyperliquid and Pump.fun accounted for nearly all of the record $638 million in crypto token buybacks between November 2024 and January 2025. Hyperliquid alone contributed roughly $467 million. Pump.fun added another $107 million. The remaining 10% is scattered across hundreds of other projects.
This is not a vanity metric. This is the first time in this cycle that on-chain protocols have demonstrated the capacity to return genuine operating cash flow to token holders at scale. No token sale proceeds. No treasury reserve liquidation disguised as buyback. Real revenue, generated from real users, redirected into open-market purchases of their own tokens.
I have been auditing token economics since the ICO era, and I have never seen concentration like this. The number demands a structural breakdown, not a headline. Because what this data actually reveals is a fork in the road for how crypto projects will be valued for the next decade.
Why these two? Why now?
The industry has spent four years talking about "real yield" and "protocol revenue." Most of that talk was theater. Yield farming programs paid users in inflated governance tokens. Revenue dashboards counted token emissions as income. Actual cash flow — money that hits a treasury wallet from user fees, without printing new supply — remained rare.
Hyperliquid broke that pattern by building what is effectively a centralized-exchange-grade derivatives venue on a custom Layer-1 blockchain. Since its mainnet launch in 2023, the protocol has captured the top spot in decentralized perpetual futures trading volume. Its revenue comes from trading fees, liquidation fees, and funding settlements. In Q1 2025, Hyperliquid was generating annualized protocol income in the $150 million range — a figure that puts it ahead of every other derivatives DEX by a wide margin.
Pump.fun is a different animal entirely. Deployed on Solana in January 2024, it operates a meme token issuance pipeline with no external venture capital backing. Revenue flows from a roughly 1% issuance fee, migration fees when tokens "graduate" to DEX liquidity pools, and trading commissions. In January 2025 alone, the platform booked over $100 million in monthly fees — the highest of any Solana application, surpassing both Raydium and Jito.
These are not comparable businesses. One is trading infrastructure. The other is an asset factory. But they share one critical feature: both have discovered product-market fit so strong that their fee generation outpaces their operating costs by an order of magnitude.
That surplus is what makes the buyback possible.
The core mechanics: what the headline doesn't tell you
The $638 million figure aggregates buyback activity across the industry, but the distribution is what matters. My analysis of the reported data, cross-referenced with on-chain activity patterns from the period, suggests Hyperliquid's share represents roughly 73% of the total — approximately $467 million — while Pump.fun accounts for about 17%, or $107 million.
Three structural observations follow.
First: this is a paradigm shift from inflation-subsidized growth to revenue-funded capital returns. The crypto industry has historically relied on a single economic model: issue tokens, use them to subsidize liquidity, and hope that user acquisition outpaces emission inflation. Hyperliquid and Pump.fun inverted that model. They generate fees from genuine user activity — traders paying for execution, degens paying to launch meme tokens — and then redirect a portion of that income into token buybacks. This is the same capital return mechanism that Apple, Microsoft, and every mature public company uses. The fact that it is now happening on-chain, verifiable in real time, is not incremental news. It is a structural change in how crypto projects can create shareholder value.
Second: the sustainability profile of these two buybacks differs dramatically. Hyperliquid's revenue base — derivatives trading fees — has strong cyclicality but also structural stickiness. Once traders accumulate positions, margin, and history on a venue, switching costs rise. Even in a bear market, derivatives volume does not collapse to zero; it compresses but persists. Pump.fun's revenue, by contrast, is deeply correlated with meme speculation cycles. When market sentiment shifts, meme token issuance drops sharply, and so does the fee stream. My assessment: Hyperliquid's buyback capacity could decline by 40-50% in a sustained bear market while remaining operational. Pump.fun's could shrink by 70-80% or more, because its income is not merely cyclical — it is sentiment-driven in the most volatile corner of the market.
Third: the buyback-burn distinction is being blurred, and that is a problem. The FT report uses the term "buyback" broadly, but in crypto, buyback and burn carry different market implications. A buyback can mean the protocol accumulates tokens in treasury — potentially to re-sell later. A burn removes tokens from circulating supply permanently. The report does not specify which mechanism Hyperliquid and Pump.fun employed. Based on my audit experience, Hyperliquid has publicly committed to repurchase-and-burn operations. Pump.fun's PUMP token buyback program, launched after its January 2025 token generation event, has been less clearly specified. This information gap matters because the market prices these mechanisms differently. A confirmed burn creates permanent supply contraction. A treasury accumulation creates ambiguous overhang. Investors should demand protocol-level disclosure on this point before treating the buyback narrative as a durable price floor.
The contrarian angle: what everyone is missing
The $638 million figure is almost certainly understated — and the missing data changes the interpretation.
The FT's number appears to capture only on-chain, non-CEX protocol buybacks. Binance's BNB quarterly burn alone was approximately $1.15 billion in February 2025. If you include centralized exchange token destruction — BNB, OKB, and similar programs — the true industry-wide buyback and burn figure for the same period exceeds $2 billion. The fact that FT chose to highlight the $638 million figure suggests this is a chain-native, protocol-level statistic. That is a legitimate scope. But it means the "record" is a record within a subset, not the industry at large. The narrative implication matters: when institutional analysts cite this data, they must be precise about the denominator.
The real story is the emergence of a TradFi-compatible valuation framework.
For years, institutional analysts dismissed crypto tokens as having no rational valuation methodology. Equity valuation rests on earnings per share, free cash flow, and capital return programs. Crypto tokens have historically had none of these. Hyperliquid and Pump.fun just changed that calculus. A protocol that generates $150 million in annual revenue and buys back $467 million of its token in three months is now analyzable using the same framework as a tech company. This opens a window for traditional valuation multiples — price-to-earnings, buyback yield, free cash flow margin — to enter crypto analysis. Based on the institutional conversations I am having, this is precisely what is happening. Several hedge funds are now modeling HYPE and PUMP using modified equity frameworks.
That is the bullish read. But there is a darker interpretation that the market is not pricing.
Buybacks at cycle highs are not accumulation. They are distribution.
When a protocol generates record revenue, it buys back tokens at elevated prices. The buyback provides exit liquidity for early investors, team members, and large holders who unlock positions during the same period. Hyperliquid's February 2025 token unlock — approximately 280 million HYPE released to team and early investors — coincided with the buyback program. The buyback did not prevent price decline; it cushioned it. This is not manipulation. It is simply the reality that buybacks during high-revenue periods function as a transfer mechanism from protocol treasury to token sellers. The question every HYPE and PUMP holder should ask: is the buyback creating a price floor, or is it financing others' exits?
The regulatory double-edged sword cuts both ways.
Pump.fun received an SEC subpoena in January 2025, examining potential unregistered securities offerings and operational compliance. The buyback program introduces a new regulatory vector. When a protocol actively repurchases its own token, it is affirmatively engaging in price management. Under the Howey test, this behavior can strengthen the argument that the token is a security — because holders' profit expectations now rest on the protocol's managerial efforts to support price. I am not making a legal judgment; I am flagging the analytical risk. A formal SEC action against Pump.fun would not just impact PUMP — it would cast a shadow over the entire revenue-buyback narrative across the industry.
The structural positioning: two toll booths at the center of different highways
Hyperliquid is constructing what I have begun to describe as the "CME of decentralized finance." Its build-your-own-L1 approach — a custom chain with an on-chain order book — is technically the harder path. Most derivatives DEXs rely on AMMs or hybrid models. Hyperliquid's decision to process order matching on its own chain gives it latency advantages that attract professional traders. But it also introduces a centralized validation risk: the validator set is limited, and the protocol's governance is concentrated in the foundation. In my assessment, this concentration is not an accident. It is the operational cost of executing rapid capital allocation decisions like buybacks without governance deadlock.
Pump.fun, by contrast, is a toll booth on the meme speculation highway. It does not need to be technically innovative. It needs to be the default destination for anyone launching a meme token on Solana. As of late 2024 and early 2025, it achieved that status. The platform's bonding curve mechanism — which requires tokens to reach a market cap threshold before migrating to DEX liquidity — functions as a quality gate and a revenue engine simultaneously. Every failed token still pays the issuance fee. Every successful token pays the graduation fee. The platform monetizes both outcomes.
But Pump.fun's ecosystem position is more fragile than Hyperliquid's. It is an application layer dependent on Solana's continued dominance in meme trading. Competitors like MakeNow.Meme and others are emerging with lower fees or differentiated mechanisms. The moat is distribution, not technology. And distribution can be bought with subsidies by a well-funded competitor.
Risk: what the numbers don't show
The headline risk in this story is not the sustainability of buybacks. It is the concentration of the entire narrative in two protocols. If Hyperliquid's derivatives volume compresses in a bear market — which it will — and Pump.fun's meme issuance collapses — which it will — the "revenue buyback" story loses its flagship examples. The industry narrative around "fundamental value" in crypto is only as strong as its most visible proof points.
There is also the safety issue. Pump.fun suffered a $1.9 million exploit in May 2024, when an attacker used internal employee credentials to execute malicious operations via flash loans. The platform has since added third-party audits and timelocks. But the incident remains a reminder that revenue capability does not equal security posture. Hyperliquid's custom L1 introduces its own attack surface — validator set compromise, order book manipulation, oracle dependency. Neither protocol has published a comprehensive independent security audit that I have been able to verify.
And then there is the question of governance transparency. Hyperliquid has faced community criticism over token allocation disclosures and the concentration of decision-making authority. Pump.fun's PUMP token distribution has been controversial — the token dropped from an opening high near $7.40 to around $0.40 shortly after its January 2025 launch, with community members questioning team allocations. When governance legitimacy is contested, the credibility of buyback programs as "value to holders" becomes harder to defend.
The takeaway: watch the next quarter, not the last one
The $638 million buyback figure is a rearview mirror measurement. The forward-looking question is whether Hyperliquid and Pump.fun can sustain the revenue that funds these buybacks — and whether the buyback mechanism itself becomes institutionalized as a regular capital return program.
My guidance for institutional readers: do not treat this as a price catalyst. Treat it as a framework event. The market is now beginning to value crypto protocols on the same basis as public equities — earnings, cash flow, and capital returns. That is a maturation signal. But maturation also means tougher scrutiny. When protocols are valued on fundamentals, their fundamentals will be stress-tested.
The next quarterly buyback report — due within 60-90 days — will be the first test. If Hyperliquid and Pump.fun maintain or increase their buyback pace despite market volatility, the narrative survives. If buyback amounts decline sharply, the "expectation gap" will punish both tokens harder than the underlying fundamentals justify.
Crypto has always been a market of narratives. What changed in the last quarter is that the narrative is now backed by revenue. That is progress. But progress is not the same as proof. The proof will arrive when these protocols maintain buybacks through a bear market — and the industry has not yet seen that data point.
The data will come. Watch for it.
