Block Demand Reprices DeFi Liquidity Before Loan Rates Do

When Ethereum activated Dencun on March 13, 2024, it demonstrated how fee markets respond to block demand: scarce capacity is repriced upward, while new capacity stays cheap until users fill it. For DeFi, that changes financing conditions first through transaction costs, not through a protocol’s quoted loan rate. Borrowers may face the same annual percentage rate yet pay more to refinance, add collateral or exit; suppliers may find a withdrawal uneconomic; liquidators must decide whether a bonus still covers execution.

How does block demand change transaction fees?

Ethereum’s execution market uses EIP-1559. When the previous block consumes more than its gas target, the next block’s base fee rises; below target, it falls. A block can expand to twice the target for a burst of demand, and the base fee can move by as much as 12.5% from one block to the next. Users also offer a priority fee for ordering and faster inclusion, while the base fee is burned.

This is congestion pricing, not a promise of immediate access. Transactions with fee caps below the new base fee wait. Competing liquidations, swaps and collateral top-ups can therefore bid against one another at precisely the moment prices are moving fastest.

The loan rate and the execution price are separate

A lending pool usually moves its borrowing rate with utilization: as more supplied assets are borrowed, the rate curve steepens to attract deposits and encourage repayment. Block demand does not directly alter that curve. It changes the cost and timing of the actions around it.

• Suppliers provide stablecoins or cryptoassets and earn interest, but retain exposure to illiquidity and bad debt.
• Borrowers pay utilization-driven interest plus the network cost of borrowing, repaying or changing collateral.
• Liquidators front gas and capital to close unsafe positions in exchange for a protocol-defined incentive.
• Suppliers, reserves or backstops absorb losses if collateral sales fail to cover debt, depending on the protocol’s rules.

A high displayed supply yield can therefore coexist with poor withdrawal conditions. If utilization is near its ceiling, cheaper gas does not create assets for withdrawal; it only makes the attempt less expensive. Vault shares add another layer because a manager may need several swaps, repayments and redemptions before returning the requested asset.

Why did Dencun create a second fee market?

Dencun implemented EIP-4844, separating temporary rollup data, called blobs, from ordinary execution gas. At launch, the specification targeted three blobs and allowed six per block—about 0.375 MB and 0.75 MB respectively. Blob prices rise when usage stays above target and fall when it stays below, independently of the execution base fee.

That separation matters because a surge in rollup data no longer has to price every Ethereum contract call through the same scarce resource. It does not make all activity free: users still pay rollup execution costs, and settling or bridging can involve another chain’s fee market. Operational references such as Manta Bridge are useful precisely because a low headline fee does not remove the steps and timing involved in recovering liquidity.

Congestion is a solvency test, not just a UX problem

The market significance is clearest during unwinds. Rising base fees ration blockspace toward users willing to pay most, which often favors large positions and professional keepers over small borrowers. Observed utilization and confirmed withdrawals show available liquidity; projected fee savings do not.

Dencun’s implemented split reduced competition between rollup data and execution, but it did not eliminate congestion risk inside lending markets. The verdict is straightforward: fee markets allocate blockspace efficiently over time, yet during a sharp demand shock they can delay the very transactions that protect collateral. DeFi risk models should treat execution cost and inclusion time as part of liquidation and withdrawal capacity, not as a cosmetic wallet charge.