I checked etherscan on a random Tuesday afternoon in February 2026 — the kind of midweek day where nothing in particular was happening in markets. Base fee was sitting at 6 gwei. Pending transactions in mempool: under 100K. A standard Uniswap swap would have cost about $4 in gas. A wstETH supply-and-borrow on Aave V3 maybe $11. A simple ETH transfer under $1.
That's the new normal. Random midweek Ethereum mainnet activity costs about as much as a coffee. For anyone who lived through the 2021 NFT mint days when minting cost $300+ in gas and you needed to set 200 gwei priority just to get included, this still feels surreal. But it's been the realized environment for most of 2025 and into 2026.
The compression from 2021 peaks (100-300 gwei sustained) to current 4-12 gwei base fee is one of the most significant unappreciated structural shifts in crypto. EIP-4844 plus L2 migration combined to drain demand from L1 calldata. The result: Ethereum mainnet now operates at fee levels closer to its theoretical "blocks not full" baseline than to its boom-time congestion peaks.
Below is a tour of what gas actually looks like in Q1 2026, when spikes still happen, and why I'm increasingly skeptical that "100 gwei is back soon" is a reasonable prediction for the rest of this cycle.
The Gwei Calendar of a Quarter
If you graph Ethereum base fee through Q1 2026, the picture is mostly flat with occasional vertical spikes. The flat baseline runs 4-12 gwei. The spikes hit 30-80 gwei briefly during specific events:
The week ETH crossed a major resistance level in late January saw base fees hit 45-55 gwei for about 18 hours as leveraged trading activity surged. Settlement-related transactions on Aave V3 and Pendle clogged blocks briefly.
A specific NFT collection drop in February (one of the few that still gets meaningful attention) pushed gas to 60-80 gwei for about 3 hours. By the next morning, gas was back to 8.
A March CPI release that came in well above consensus drove a sharp ETH selloff, which triggered cascading liquidations on Aave V3, MakerDAO/Sky vault adjustments, and stablecoin movement. Gas hit 40-60 gwei for most of one trading day, then settled back.
These spikes have a pattern: they're event-driven, short-duration (hours, not days), and revert to baseline quickly. Pre-2024, similar events would have driven gas to 200+ gwei sustained for days. Post-EIP-4844, the L2 capacity absorbs the marginal flow that used to spike L1.
Why Gas Stays Low Most of the Time
The structural reasons gas baseline shifted lower are worth understanding because they bear on whether gas spikes again.
L2 ecosystem migration is the biggest factor. Pre-2024, DEX swaps, lending interactions, NFT mints all happened on L1. Each transaction consumed L1 gas. Aggregate retail activity created sustained L1 congestion. Post-2024, most of that activity moved to Arbitrum, Base, Optimism. L1 retains only operations that genuinely need L1 (large institutional positions, MakerDAO/Sky operations, bridge settlements, MEV-related activity). The volume of L1 transactions dropped substantially.
EIP-4844 blob storage is the second factor. Pre-EIP-4844, L2s posted their data to L1 as expensive calldata. Each L2 batch consumed substantial L1 gas. Post-EIP-4844, L2s post data as cheap blobs that don't compete with regular transactions for blockspace. The L1 transaction market got materially less crowded.
Retail and speculative activity rotation to other chains is the third factor. Memecoin trading went to Solana. Frame-driven onchain activity went to Base. Many casual users left Ethereum L1 entirely. The "long tail" demand that used to fill L1 blocks during sentiment peaks doesn't exist on L1 anymore.
These factors compound. Each one reduces L1 demand. Combined, they shifted L1 from a near-perpetually-congested state to a state where blocks rarely fill completely.
When Gas Spikes Actually Happen
Despite the lower baseline, gas does still spike. Tracking when:
Liquidation cascades drive gas spikes. When ETH moves sharply, leveraged positions on Aave V3, MakerDAO/Sky, and other lending protocols get liquidated. Liquidator bots compete for inclusion, which bids up gas. These spikes typically last hours rather than days.
Major DEX arbitrage events drive gas spikes. Sharp moves create cross-DEX price differentials. Arbitrageurs race to execute, bidding up gas in the process. Same event-driven character, similar duration.
Specific NFT drops or token launches that capture genuine attention drive gas spikes. Most NFT activity has died but occasional drops still draw real demand. The drop window typically lasts 2-6 hours.
Major CEX outflows or institutional movements drive gas spikes. When a major holder moves $100M+ in one direction, the related DeFi position adjustments cascade through L1.
Ethereum upgrade-related activity around hard forks drives gas spikes. Pectra rollout in 2025 caused a few high-gas days as everyone tested new functionality.
What doesn't drive sustained gas spikes anymore: ordinary retail volume. Even on days when crypto is broadly trending up and there's enthusiasm, gas typically stays under 20 gwei because retail flow is on L2s now. The structural change is that "bull market = expensive gas" relationship that defined 2017-2022 has weakened substantially.
The Cost Curve for Different Operations
In Q1 2026, here's roughly what common operations cost in USD-equivalent at typical 6-8 gwei base fee with normal ETH prices:
Plain ETH transfer: under $1. Standard Uniswap swap: $3-6. Wrapped ETH (WETH) operations: $2-4. Simple ERC-20 transfer: $1-2. wstETH supply on Aave V3: $4-8. Borrow against collateral on Aave V3: $5-10. Pendle PT mint or redeem: $4-8. NFT transfer (ERC-721): $2-4. Complex DeFi interaction (multi-step): $10-25. Worst-case complex contract interaction: $30-50.
Compare to 2021 peak when a Uniswap swap could cost $80-150 routinely. The cost-of-doing-DeFi has compressed dramatically.
For active DeFi users, this fundamentally changes the economics. Strategies that weren't viable at $80/swap (active rebalancing, frequent position adjustment, smaller-size DeFi operations) are now viable. The "DeFi for $5K positions" that was uneconomical in 2021 works fine in 2026.
My Own Operational Adjustments
Given the new gas environment, I run several practices:
Most of my L1 operations are now non-time-sensitive. I'll batch operations for late-night UTC or weekend windows when gas is typically lowest (often 3-6 gwei). The savings vs peak hours are meaningful at scale even though absolute amounts are small.
I use L2s by default for any active DeFi positioning. Aerodrome on Base for stable LP. Aave V3 on Arbitrum/Base for moderate leverage. Pendle on Arbitrum for fixed-yield positioning. L1 only for positions that genuinely need L1 (large size, specific protocol requirements, long-duration positioning).
I size DeFi positions assuming ~20bps in gas costs over a year of active management. Pre-2024 I used to assume 100-200bps. The cost compression means smaller positions are economical.
I don't bother optimizing single transactions for gas anymore. The amounts are small enough that 5 minutes saved isn't worth optimizing for $0.50.
The Forward Question
The interesting forward question is whether these new structural conditions persist or whether crypto returns to "expensive gas" mode in the next mania cycle.
The bear case for gas (i.e., gas stays low): L2 ecosystem keeps growing, EIP-4844 blob storage keeps absorbing L2 demand, retail keeps preferring Solana for speculative activity, and Ethereum becomes a settlement layer with consistently low fees. Even sentiment peaks don't drive sustained L1 spikes because there's nowhere for marginal demand to go that L2s can't handle.
The bull case for gas (gas comes back): some new use case emerges that genuinely requires L1 (institutional ZK rollup settlements at scale, novel financial primitives that don't suit L2, regulatory requirements that push activity to L1) and creates sustained demand. Or: L2 capacity hits limits that push activity back to L1.
I'm in the bear-case camp. The structural shifts of 2024-2025 look durable to me. The L2 ecosystem keeps maturing. EIP-4844 keeps being effective. Retail keeps preferring chains other than Ethereum L1 for most activity. I don't see the catalyst that brings sustained 100+ gwei back.
But I've been wrong about gas before. In 2020 I thought gas would stay manageable through the bull market. It didn't. So I hold this view loosely.
The practical implication for users: assume gas stays low for the foreseeable future, plan operations accordingly, but maintain optionality to adjust if structural conditions shift. Build DeFi positioning that benefits from low gas (more frequent rebalancing, smaller positions, more diverse strategies). Don't position around expensive gas as the assumption.
Footnotes
The gas figures and patterns are from etherscan, ultrasound.money, dune analytics, and my own observation through April 2026. The "$X cost" figures depend on ETH price (variable) and exact gas amount per operation (varies by contract). The historical comparison with 2021 peak gas reflects publicly observed data; specific peak gas levels varied across days. The forward speculation about gas trajectory is opinion rather than data — alternative trajectories are plausible. Personal operational observations reflect my own usage and aren't recommendations. Smart contract interaction gas costs can vary substantially by specific contract; the cited cost ranges are typical not guaranteed. None of this is financial or operational advice.