Why Lower L2 Fees Don’t Necessarily Bring Profits
Lower L2 fees don't necessarily bring profits. The fundamental reason is that fee reductions cause a sharp decline in "per-transaction profit," while user numbers and total transaction volume haven't grown enough to make up the shortfall. Simply put, the pie hasn't gotten bigger, but each slice has become thinner.
How falling fees squeeze profits
Many people see "lower fees" as a good thing—users pay less, so more people should use the network. But the reality is that a key variable sits between L2 fees and L2 profits: transaction volume.
Profit = (revenue per transaction – cost per transaction) × transaction volume.
Revenue per transaction comes from the fees users pay. Cost per transaction comes from the expenses L2s incur when posting data to Ethereum mainnet (including blob fees and calldata fees).
After EIP-4844 (Proto-Danksharding) introduced blob space, the cost for L2s to post data to mainnet dropped by over 90%. The sharp decline in per-transaction costs gave L2s room to significantly cut user fees, delivering on the promise of a "cheap experience."
The problem is: costs fell more than revenues, but that doesn't mean total profits grew.
Data shows that before the Dencun upgrade, top L2s were making a profit of $0.06 to $0.60 per transaction. Six months after the upgrade, per-transaction profit had fallen to $0.01 to $0.04—a profit compression of over 90%.
The direct cause of this profit squeeze is competition among L2s. When everyone has access to cheap blob space, fees become the only weapon. L2s slashed user fees dramatically, passing virtually all of the cost savings on to users and keeping very little for themselves.
Why "lower margins, higher volume" didn't work here
The premise of "lower margins, higher volume" is that thinner margins can drive enough additional volume.
But in reality, user numbers didn't surge just because fees fell. Data shows that the Ethereum ecosystem's user count has been declining since 2021. Even with fees dropping by a factor of 100 or more, the expected user growth didn't materialize.
To understand why lower fees failed to attract users, you can look at it from two angles:
First, user growth requires use cases, not lower transaction fees.
Fees dropping from $10 to $0.50 is great for existing users, but for someone who "has no idea what Ethereum is," that information holds no appeal. Users don't show up because the product itself hasn't changed.
Second, L2s are fighting a price war, but they're competing for the same existing users.
The three major L2s—Base, Arbitrum, and Optimism—are all cutting fees, but their target audiences overlap heavily: they're all crypto natives, not newcomers. Cutting fees is just a battle for share within the existing pool; it can't bring real growth.
The result: per-transaction profit dropped 90%, but transaction volume didn't grow 10×. As a consequence, total profits fell sharply.
The data tells the story: a real ledger for 2025
Here's a complete snapshot:
In 2025, total Ethereum L2 revenue fell by 53% to $129 million
Fees paid by L2s to Ethereum mainnet dropped sharply to just $10 million
L2 operators retained roughly $119 million in profit
On the surface, L2s still show $119 million in profit, but it's important to note: this is "retained profit," meaning the difference between what L2s collect from users and what they pay to mainnet. This figure declined significantly year-over-year compared with 2024.
From another angle: in 2024, Base led by a wide margin with about $85 million in protocol fee revenue, exceeding the combined total of Arbitrum and Optimism. However, most of Base's fee revenue became profit for Coinbase; very little went to Ethereum mainnet—on an annualized basis, Coinbase's earnings from Base approached $100 million.
L2s make money, but that money is not flowing back to mainnet.
A more fundamental mechanism: L2s actively avoid high fees
If you think "lower fees are temporary—once the network gets congested, fees will rise again and L2 profits will recover naturally," you may be overlooking the behavioral logic of L2 operators.
An L2 is essentially a business. Businesses proactively avoid creating high-cost environments for themselves.
Blob space has an automatic fee adjustment mechanism: if a block contains more than 4 blobs (the target is 3), the base blob fee for the next block rises by about 12%. In other words, if L2s collectively post blobs at high frequency, they drive up their own costs.
As a result, L2 behavior becomes predictable: when blob fees start rising, L2s will choose to:
Reduce posting frequency
Stagger posting times to avoid peaks
Temporarily switch back to traditional calldata (if cheaper)
Stop posting altogether (for zk rollups, there is no mandatory posting cadence)
A real-world example: during the LayerZero airdrop in June 2024, Arbitrum transaction volume surged and blob fees spiked. Scroll stopped posting blobs entirely, and Taiko slowed its batch transaction speed by 30–50%.
L2s will constantly adjust their behavior to avoid high-fee environments. This means L2 fees can never really rise—at least not to levels that would make L2 operators comfortably profitable.
As one researcher wrote in an analysis: "L2s will continuously adjust their behavior to avoid high costs. They are businesses, of course they will … this goal (making ETH deflationary again) will never be reached."
Risk warning
If you hold L2-related assets (such as OP, ARB tokens, or ETH), one structural risk to be aware of is: lower L2 fees challenge ETH's deflationary narrative. The shift from ETH as "ultra sound money" to ETH as a "scalable computing network" means that the amount of ETH burned is unlikely to return to historical highs in the foreseeable future. Token price may decouple from network activity—2025 is an example: Ethereum mainnet TVL accounted for 64%, yet the ETH price fell 10% over the year.
How to verify this yourself
Open your on-chain data tool (such as Dune or Token Terminal) and check the following three metrics for the L2 you follow over the past 90 days:
Transaction Count — has it grown significantly?
Average Fee Per Transaction — has it continued to decline?
Protocol Revenue — has it increased?
If you find that the L2's transaction volume isn't growing fast, fees are steadily falling, and revenue is stagnant, you're looking at a real-life example of "lower fees without higher profits." If you use an L2 for everyday transactions, check whether the fees you pay have actually come down—if they have and the user experience hasn't changed, at least you're saving money, even if the L2 operator isn't making much.
Next steps: If you hold L2 native tokens, keep an eye on whether the project has a "revenue conversion mechanism"—for example, using sequencer revenue to buy back tokens or distribute fees. If not, the benefit of lower fees accrues only to users, not to token holders. Also watch the timeline for the Pectra upgrade—the next upgrade may expand blob capacity by 2–3×, driving fees even lower and further compressing profit margins.
