Crypto projects have poured a record $638 million into buying back their own tokens so far in 2026, a 17% jump from the $545 million spent over the same period last year, with Hyperliquid and Pump.fun alone accounting for close to 90% of that spending, according to Allium Labs data cited by the Financial Times.
The surge is not, however, spread evenly across the industry. Hyperliquid and Pump.fun account for close to 90% of the tracked spending, making their revenue models the primary engines behind this year’s increase.
That concentration is important. Rather than signaling that every decentralized finance project has suddenly embraced repurchases, the numbers point to a smaller group of high-revenue protocols using cash generated from users to create recurring demand for their native assets.
Hyperliquid turns trading fees into a powerful HYPE engine
Hyperliquid has emerged as the clearest example of a protocol linking network activity directly to token demand.
Its Assistance Fund receives the overwhelming majority of eligible trading fees, with approximately 99% directed toward automated purchases of HYPE. The acquired tokens are subsequently retired, creating a permanent reduction in supply.
Hyperliquid’s model has attracted significant attention because the purchasing mechanism is tied directly to protocol activity rather than relying solely on discretionary treasury decisions. More trading activity can therefore translate into greater fee generation and, consequently, larger HYPE purchases.
The scale is substantial. Hyperliquid Strategies said in a recent regulatory filing that approximately $1.2 billion worth of HYPE had been bought back and retired since the program began, with 46.4 million HYPE removed from supply. The company also said roughly 99% of net protocol fees were being directed toward the mechanism.
Bitwise Chief Investment Officer Matt Hougan has described the model in similarly direct terms, arguing that “99% of trading fees” go toward buying HYPE and that the relationship creates a straightforward connection between platform activity and token value accrual.
The distinction between cumulative and annual figures remains critical. Hyperliquid’s roughly $1.2 billion-plus cumulative spending since launch should not simply be added to the $638 million industry total because the two figures cover different periods and datasets.
Pump.fun adds another major source of buying pressure
Pump.fun has become the other major contributor to this year’s surge.
The Solana-based token launchpad says half of every dollar generated by the platform is allocated toward buying PUMP on the open market and subsequently burning the tokens. Its official dashboard currently reports more than $442 million in cumulative PUMP purchases and 162.95 billion tokens burned.
The mechanism draws revenue from several parts of the Pump ecosystem, including its bonding curve, PumpSwap and Terminal products. Pump.fun says its revenue methodology includes those activities across Solana, Base, Ethereum and BNB, after specified deductions such as referral fees and cashbacks.
The strategy creates an important counterweight to token issuance. When tokens are purchased and permanently destroyed, the number remaining in circulation falls. But that effect can be complicated by scheduled unlocks.
Pump.fun has also faced significant token distributions to team and investor wallets as vested PUMP became transferable. That creates a fundamental tension: buy-and-burn activity removes tokens, while unlocks can increase the amount of supply available to the market.
The result is a reminder that a large repurchase program does not automatically translate into higher prices.
Sky and Lido take more conditional approaches
Sky Protocol represents a different model, using protocol surplus to purchase SKY through its Smart Burn Engine.
Its approach links repurchases to the financial health of the protocol rather than committing an unlimited amount of capital regardless of revenue conditions. Sky’s mechanism has therefore become another prominent example of a decentralized protocol attempting to connect economic activity with token supply management.
The broader trend has also reached Lido, although its NEST mechanism is more tightly bounded.
Lido’s governance-approved framework establishes a $40 million annual revenue baseline, with 50% of eligible surplus allocated toward LDO purchases. The mechanism includes a $50,000 daily ceiling and a $10 million rolling annual limit. The ETH price floor that appeared in earlier versions of the proposal was ultimately disabled at launch.
Lido’s NEST system was fully activated in August following governance approval, marking a significant step toward a rule-based connection between protocol revenue and LDO demand.
That structure illustrates an increasingly important principle in decentralized finance: protocols are becoming more selective about when they deploy capital rather than treating buybacks as an unconditional promise.
Record Token Buyback Spending Does Not Guarantee Higher Prices
The rise in repurchases is significant, but investors should be careful about interpreting the $638 million figure as evidence that buybacks automatically create sustainable token value.
A repurchase can establish a recurring source of demand and, when tokens are permanently burned, reduce supply. But price performance still depends on revenue growth, market liquidity, investor demand, token emissions, unlock schedules and broader market conditions.
Crypto trader Ansem has made precisely that argument. In July, he questioned whether recurring repurchases could overcome weak community alignment, pointing to the dramatically different valuations of HYPE and PUMP despite both benefiting from revenue-funded buying programs.
Research from Coinbase Institutional has also highlighted the other side of the equation. Its analysis of Hyperliquid identified token unlocks and the relationship between fee generation and buyback conversion as important risks to the HYPE investment thesis.
That makes the next phase of the market particularly important. If trading activity remains strong, revenue-funded repurchases could continue expanding. If volumes fall sharply, however, the buying pressure generated by these mechanisms could decline at the same time that investors become more cautious.
For that reason, the real significance of the $638 million milestone may not be the size of the spending alone. It is the emergence of a new crypto economic model in which protocol revenue, token demand and supply reduction are increasingly connected.
Whether that model produces durable value will depend on what happens when the revenue cycle turns lower — not simply how aggressively protocols buy their tokens during a strong period.