Fee switch in DeFi: which token actually distributes in 2026

The most generous headline in DeFi promises 80% of revenue to token holders. Translated into real yield on that token's price, today it is worth almost zero. Over the last twelve months, four protocols activated revenue-sharing mechanisms with radically different styles: Uniswap turned on its fee switch (diverting a portion of protocol fees to holders), Optimism buys back its token with sequencer revenue, Jito (a liquid staking and MEV protocol on Solana) launched a trading terminal that distributes 80%, and Arbitrum generates revenue but distributes nothing. The problem is that "80%" and "17%" are not comparable: they depend on the base they are calculated against. This article normalizes all four to a single metric — the annual yield effectively reaching the holder for every $1 billion in market cap — and compares it against a Treasury bill (short-term public debt, the benchmark risk-free asset). The result dismantles the headlines.

Why can "80%" be worth less than "17%"?

A distribution percentage means nothing without its denominator. When Jito announces it allocates 80% of revenue from its JTX terminal to buy back JTO, that 80% applies to fees from a product launched on June 26, 2026 — just a few weeks of operation. When Uniswap says its fee switch captures 17% of swap fees, that 17% applies to the largest decentralized exchange market on the planet, which moves billions per day.

17% of a giant figure comfortably beats 80% of a tiny figure. The headline optimizes perception; real yield is what optimizes the wallet. The only way to compare them is to bring them down to the same unit: how many dollars per year reach the holder, and what proportion they represent relative to what the token is worth on the market today. That is an earnings yield — the crypto equivalent of a stock's earnings yield: annual profit divided by market cap.

Which protocols activated revenue sharing in the last twelve months?

The wave of fee switches and buybacks in 2025-2026 has specific dates. Reconstructed in order:

• December 28, 2025 — Uniswap activates the fee switch on Ethereum following the UNIfication proposal; protocol fees begin accumulating in a contract (TokenJar) and are used to burn UNI.
• January 28, 2026 — Optimism governance approves with 84.4% of the vote to allocate 50% of net sequencer revenue from the Superchain to buy back OP, in a twelve-month pilot.
• February 2026 — The OP buyback pilot begins with monthly open-market purchases.
• February 26, 2026 — Uniswap votes to expand the fee switch to eight additional L2 chains; UNI rises 15% on the day of the announcement.
• June 26, 2026 — Jito launches JTX, a self-custodial trading terminal on Solana that directs 80% of its fees to buy back JTO.
• No date — Arbitrum continues without any buyback, burn, or dividend mechanism, despite generating revenue with Timeboost since 2024.

Each data point lives in a different source — Talos, CoinDesk, SolanaFloor, Crypto Economy.

How do you convert a headline percentage into a comparable yield?

The recipe has four steps, and any reader can replicate it with public data:

• One: set a single time window. It is not useful to mix Uniswap's "cumulative since activation" with Jito's "Q1." Here, everything is annualized at a yearly rate (recent revenue projected over twelve months), measured as of July 2026.
• Two: isolate the money reaching the holder. Only the portion converted into buybacks or burns of the governance token counts here, excluding the rest of the protocol's revenue. In Uniswap, it is 100% of the captured protocol fee (all is burned); in Optimism, 50% of the sequencer's net revenue; in Jito, 80% of JTX revenue; in Arbitrum, zero.
• Three: divide by the token's current capitalization. Annual buyback or burn divided by the token's market value gives the earnings yield.
• Four: compare against the risk-free asset. A three-month U.S. Treasury bill yields 3,79% annually (July 16, 2026). That is the benchmark: any governance token with smart contract and market risk should, in theory, exceed it to compensate for the extra risk.

What did each protocol promise when activating its revenue sharing?

This is how the four mechanisms are communicated, exactly as they appear in press releases and governance tweets. The column that matters is the last one: the base upon which that flashy percentage is actually applied.

Uniswap is the cleanest mechanism of the four. Its fee switch takes protocol fees from v2, v3, and Unichain pools to a contract called TokenJar; value only leaves there by burning UNI through a second contract, Firepit. Since its activation (December 28, 2025), it has accumulated nearly $23 million, and its recent pace is around $4.9 million per month according to CryptoBriefing and DeFiLlama — about $50 million annualized — compared to practically zero in previous periods. The proposal to expand it to eight additional L2s could add an extra $27 million annualized on top of the nearly $50 million already fueling the burns. It is real distribution, although — as will be seen — modest relative to the token's value.

The Optimism case deserves two nuances. Its full story is covered in how the Base exit breaks the Optimism Superchain economy, and the revenue landscape framing all four protocols in the real DeFi revenue ranking 2026. First: its 50% sounds substantial, but net sequencer revenue is modest — the official Optimism blog estimates the buyback base at about 5,868 ETH annualized, of which 50% equals about $8.75 million annually if activity holds. Second: bought-back tokens return to the treasury and can be burned or distributed as staking rewards, so the distribution goes beyond Uniswap's pure burn.

How much reaches the holder for every $1 billion in market cap?

This is where everything changes. By normalizing the annual money reaching the token against its market cap, the headlines are reordered inversely to how they sound. The last column translates the earnings yield into concrete dollars for every $1 billion of market value — the figure that makes all four comparable to each other and against the Treasury bill.

The ranking is reversed. Jito, the loudest headline with its 80%, falls to the bottom alongside Arbitrum: JTX has been trading for less than a month and its verifiable buyback rate is still indistinguishable from zero. Uniswap, the most celebrated fee switch, yields ~2.2% —little more than half of a Treasury bill, compared to ~4.0% for Optimism— while the holder bears all the smart contract and price risk. Optimism appears at the top with ~4.0%, but it is worth reading the fine print.

The Jito case encapsulates the lesson of the entire article. Its visible revenue —the $78 million in MEV fees (value extracted by validators from transaction ordering) reported by the network— is distributed between validators and stakers, remaining out of reach for JTO holders. The first mechanism that actually impacts JTO is the 80% buyback of JTX, and JTX launched on June 26. Measuring an earnings yield over just three weeks of operation produces an expectation, not consolidated data. This is why its box shows a verified zero: the distribution exists on paper, but there is no history to annualize yet. The design, however, was strengthened this very week: JIP-38, approved on July 14, 2026, commits 100% of the DAO portion —that 80% of JTX fees— to JTO buyback and burn at least until the fourth quarter of 2027. The machine is now permanent; what remains to be proven is the revenue fuel.

OP's apparent leadership is largely a mirage of the denominator: its market cap has plummeted, and a fixed buyback on a cheap token buys proportionally more and raises the arithmetic yield. A high yield can reflect protocol generosity or price desperation; here, the latter carries more weight. Furthermore, the ~$8.75 million annual figure is an estimate based on volatile sequencer revenue: net sequencer revenue for the Superchain was around $14 million in the first quarter, and half of the buyback base — about $8.75 million annualized — is what effectively goes toward purchasing OP, according to the official Optimism blog.

Why does Arbitrum distribute zero despite generating revenue with Timeboost?

Arbitrum is the critical position in this table. Timeboost — its transaction priority auction, which sells the right to execute before others — generated $406,000 in gross revenue in the first quarter of 2026, and the network as a whole produced about $23.49 million in gross profit in 2025 via fees, Timeboost, and the expansion program. None of that reaches ARB.

ARB is a pure governance token: it gives a vote over the treasury but grants no direct economic claim on protocol revenue. No proposal has advanced to allocate sequencer revenue to holders, nor is there a buyback or burn schedule. The design worked in 2023, when the L2 narrative (second-layer networks that make Ethereum cheaper) rewarded decentralization over returns. In 2026, with the competition sharing cash, this absence is read as a decision — conscious or due to governance paralysis — to leave the token without a value engine.

The only recent move in the opposite direction is Robinhood Chain (Robinhood's L2 on Arbitrum Orbit), which routes 8% of its net protocol revenue to the Arbitrum DAO treasury. But the treasury is a common fund: the money swells the DAO's coffers, far from the token price. For the earnings yield calculation, ARB remains at zero.

Does any of these tokens beat a Treasury bill?

None of them beat it clearly. The three-month Treasury bill yields 3.79% without credit risk, without exposure to a smart contract, and without price volatility of the principal. Against that benchmark:

• Uniswap (~2.2%) yields about two-thirds of that benchmark, and its holder assumes market, contract, and regulatory risk in exchange for that burn.
• Optimism (~4.0%) nears the benchmark only due to the effect of a depressed capitalization, with an income figure still to be confirmed.
• Jito (≈0% today) bets on JTX growth; if the terminal were to reach the protocol income rate that Jito recorded in the first quarter —2.33 million dollars, about 9.3 million annualized— 80% would be around 7.5 million per year, a yield in the range of 2.6% on the current JTO capitalization. It would still remain below the bill, and that figure is a hypothesis still lacking a track record to support it.
• Arbitrum (0%) is not even competing.

The conclusion is uncomfortable for DeFi marketing: in July 2026, the best revenue sharing among these four governance tokens does not convincingly beat cash. The headline percentage measures public relations; the earnings yield measures what reaches the pocket, and that number is, at best, tied with a Treasury bill that cannot suffer an exploit.

How to distinguish a material fee-switch from a cosmetic one?

Applied to the four cases —~2.2% Uniswap, ~4.0% Optimism, ≈0% Jito, 0% Arbitrum—, the test boils down to three questions that anyone can re-examine before believing the announcement:

• On what base is the percentage applied? 80% of a newborn product carries less weight than 15% of a mature business. Demand the absolute annualized figure over the proportion.
• Does the money burn/buy back the token, or enter the treasury? Burning and buybacks that reduce circulating supply touch the price; the treasury is a fund that governance can spend on anything. Arbitrum illustrates the difference.
• How much does that yield relative to market cap, and how does it stand against a Treasury bill? If a token with smart contract risk yields less than public debt, the distribution is a story, not an investment thesis.

The four protocols have begun to respond to the old criticism that governance tokens do not capture value. But starting to distribute is far from distributing something material. As of July 2026, three of these four tokens yield less than cash, and the fourth stays at zero. The headline sells you the percentage; the earnings yield tells you if it deserves your capital.

Sources and links: Talos — Uniswap Flips the Fee Switch · CryptoBriefing — Uniswap revenue · CoinDesk — Optimism OP buyback · SolanaFloor — Jito JTX 80% · Crypto Economy — Arbitrum revenue void · DeFiLlama — Arbitrum Timeboost · Trading Economics — 3-month Treasury bill