Gas fees on Ethereum: what you're actually paying for and how L2s change the math

The first time most users send a transaction, they're surprised it costs anything. The second time, they're surprised it costs that much. The third time, they want to know what they're actually paying for.

Gas fees are the most-complained-about feature of Ethereum and one of the least understood. This piece walks through what gas is, why fees move, and how Layer 2s (L2s) have reshaped the cost-benefit calculation in 2026.

What gas actually is

Every operation on the Ethereum Virtual Machine costs a fixed amount of gas. A simple ETH transfer costs 21,000 gas. A token transfer costs around 50,000–80,000. A complex DeFi interaction (deposit into a vault, swap through three pools, mint an LP token) can cost 200,000–500,000.

The total ETH cost of a transaction is:

fee = gas_used × gas_price

The gas_used is determined by the work the transaction does. The gas_price is set by the market — specifically, the user offers a price, and the network includes their transaction if the offered price is high enough relative to other transactions waiting in the mempool.

Two pieces, multiplied. The fee is large when either the work is large or the network is congested.

Why fees move

Three forces.

1. Block space is scarce

Each Ethereum block has a target gas usage and a maximum. If demand exceeds the target, the base fee rises. If demand is below the target, the base fee falls. This is the EIP-1559 mechanism, live since 2021.

The base fee adjusts every block. In quiet hours it can fall to 1–3 gwei. During peak congestion (a popular NFT mint, an airdrop claim deadline, a major liquidation cascade) it can spike to 100+ gwei. The same transaction can cost $2 or $80 depending on when you send it.

2. ETH price affects USD-denominated fees

Gas is paid in ETH. Even if the gwei price of gas is flat, a doubling of ETH's USD price doubles the USD-denominated cost of every transaction. This is why "Ethereum is expensive" tends to track ETH bull markets — the gas mechanism didn't change, but the unit of account did.

3. The complexity of what you're doing

A single-token transfer is cheap. A deposit into a vault that internally routes capital across four protocols is many times more expensive. The complexity is paid for by the user submitting the transaction. There's no subsidy.

This is one reason aggregator vaults (like vildX) are more gas-efficient for users than DIY allocations. The vault rebalances across four protocols as a batched operation; a user doing the same allocation manually would pay four times the gas.

What L2s do

A Layer 2 is a separate execution environment that periodically posts compressed proofs of its activity to Ethereum mainnet. Users transact on the L2, paying L2-native gas; the L2 batches transactions and settles them to L1 in aggregate.

Examples: Arbitrum, Optimism, Base, zkSync Era, Linea, Scroll, Polygon zkEVM.

The economic effect is that L2 transactions are roughly 10–100x cheaper than L1 transactions, depending on the L2 design and the L1 base fee.

Why? Because the cost of the L1 footprint (calldata posting, state commitments) is amortized across thousands of L2 transactions. If the L1 batch costs $500 and represents 5,000 L2 transactions, each L2 user pays a tiny share of the L1 cost on top of a minimal L2 execution cost.

Where L2s help:

• High-frequency activity. Day-trading, frequent rebalancing, NFT minting.
• Small transactions. A $50 swap on L1 might cost $10 in gas; on an L2, the same swap might cost $0.05.
• Onboarding new users. Cheap transactions make experimentation viable.

Where L1 still wins:

• Highest-value, lowest-frequency transactions. A $1M deposit doesn't care about a $20 gas fee, and the L1 has stronger settlement guarantees.
• Maximum decentralization. L2s have varying degrees of training-wheels (sequencers, fraud-proof maturity); L1 has none.
• Protocols that haven't deployed to your preferred L2 yet. Ecosystem coverage matters.

The 2026 L2 landscape

A quick summary of where things stand.

• Optimistic rollups (Arbitrum, Optimism, Base) have the deepest liquidity and the broadest dApp coverage. Settlement to L1 takes 7 days for trust-minimized withdrawals, faster via bridges that front the liquidity.
• ZK rollups (zkSync Era, Linea, Scroll, Polygon zkEVM) settle faster and have stronger cryptographic guarantees, but the dApp ecosystem is still catching up.
• Base is the fastest-growing L2 by user count in 2026, driven heavily by consumer applications and Coinbase's onboarding flow.
• Arbitrum remains the largest by TVL and the default for serious DeFi.

For DeFi-yield use cases specifically, Arbitrum and Optimism have the most mature lending and DEX ecosystems. Base is catching up quickly.

How vildX thinks about chain selection

The vildX strategy currently operates on Ethereum mainnet because that's where the deepest stablecoin liquidity, the longest-tested lending protocols, and the most-audited contracts live. For a strategy whose entire value proposition is don't take more risk than necessary, mainnet is the right starting point.

We're actively planning multi-chain expansion to L2s — specifically Base and Polygon — to lower the gas friction for smaller deposits. The math is straightforward: a $500 deposit on mainnet pays meaningful gas to enter and exit. The same deposit on Base pays cents. Lowering that floor opens the product to a category of users who currently get priced out.

The trade-off is that L2 deployments inherit some of the L2's own risks — sequencer downtime, fraud-proof maturity, bridge dependencies. We're being deliberate about which L2s to add and on what timeline. The goal is broader access, not chasing chains.

How to choose what chain to use as a user

Three quick rules.

1. Match transaction size to chain. Small experiments → L2. Significant deposits → mainnet if available.
2. Stay where your assets are. Bridging assets between chains is its own cost and its own risk. Pick a chain and concentrate activity there.
3. Don't optimize for the lowest fee. A 5-cent fee on a chain with poor security guarantees is a worse trade than a $5 fee on mainnet. Security has a price; that price is sometimes worth paying.

Gas is the friction of using a public, decentralized network. L2s have made it dramatically less frictional than it used to be, but the underlying cost — block space is scarce, computation isn't free — hasn't gone away and won't. The right reaction isn't to complain about fees; it's to pick the layer that fits the transaction you're doing.