Strategy/
Yield in decentralized protocols: sources, sustainability, and risk
Some on-chain yield is payment for a real service. Some is a marketing budget denominated in tokens. Telling them apart is most of the work.
A yield figure on its own is close to meaningless. The useful question is always what is being paid for, by whom, and out of what.
In decentralized protocols, there are essentially three answers, and they have very different durability.
Fees paid by users
The most durable source is a fee someone is willing to pay for a service they want. Traders pay to swap; the fee accrues to whoever supplied the liquidity that made the swap possible. Borrowers pay interest; the interest accrues to whoever supplied the asset being borrowed.
This yield is sustainable in the same sense that any business's revenue is sustainable: it lasts as long as the demand does. It is also variable in the same way. Lending rates are set by utilization and can move sharply when demand shifts. Liquidity provision returns depend on trading volume, which is not constant.
Payment for securing a network
Staking pays for a service too, but the service is security rather than liquidity, and part of the payment is issuance rather than fees. Restaking extends this: the same staked capital is committed to secure additional protocols in exchange for incremental rewards.
That extension is not free. Each additional commitment adds its own slashing conditions and its own smart contract dependencies on top of the base staking risk. The capital is the same; the number of ways it can be impaired is not.
Incentive emissions
The third category is a protocol distributing its own token to attract capital. This is a customer acquisition cost paid in equity, and it should be read that way. It can be entirely rational for the protocol and entirely real for the recipient, but it decays by design, and a strategy built on it is a strategy with a clock on it.
The failure mode is not that emissions stop. It is that the reported yield never distinguished emissions from fees in the first place, so nobody noticed the composition changing.
The risks that are not about yield at all
Smart contracts can fail or be exploited, and an audit reduces that probability without eliminating it. Liquidity provision carries impermanent loss when paired asset prices diverge, a loss that is unrealised until withdrawal, and then is not. Stablecoins and liquid staking tokens can de-peg, and a de-peg propagates through any position that treated them as equivalent to the asset they track.
Looping, borrowing against supplied collateral and re-supplying it, compounds all of the above. It amplifies the yield and it amplifies everything else, including the sensitivity to a borrowing cost that can rise faster than the yield it funds.
The practical consequence is that a position size should be set by what survives the failure, not by what the yield looks like when nothing fails.