Are L2s Dying?

It’s all 3 at once: growth at the system level, saturation at the general-purpose layer, and a reset in security + economics.
First, growth is undeniable.
On an activity frame, rollups are doing ~2.13K user ops/sec vs Ethereum L1 ~33.13 UOPS. That’s ~100x scaling in raw execution demand.
Converted, that’s ~184M user ops/day on rollups vs ~2.86M on L1.
-----
But the structure just changed.
- #Base alone is sitting >60% of all L2 txn share
- Base roughly 47% of rollup DeFi TVL, Arbitrum ~31%, OP mainnet ~6%, everyone else fighting for scraps.
- Base is migrating off the Optimism OP Stack to operate its own chain.
- In January 2026, total gas fees across OP Stack were ~68.2 ETH (~$199.7K). Base alone accounted for ~$192.9K - ~96.5%.
- In 2025, Base generated an estimated 71–90%+ of Superchain sequencer revenue.
Base was already the only L2 with a clearly positive 2025 bottom line (~$55M after blob costs). No token required.
If a chain is just a generic EVM-but-cheaper rollup, idk who exactly they’re stealing users from.
Because they’re not outcompeting Base’s distribution funnel, not out-liquiding #Arbitrum’s DeFi bank vibe, and not out-network-effecting #OP Stack as infra.
-----
That reminds me of Vitalik’s callout. If it’s just another EVM chain with a bridge, it’s even worse. Nobody needs more copy-paste EVMs just to scale.
He reset everything.
→ Security gets repriced socially.
L2s are still heavy Stage 0: 14 rollups at Stage 0, 7 at Stage 1, only 4 at Stage 2 in that rollup set.
Stage 1 or bust is turning into a coordination point.
→ The economics got nuked by success.
-----
Post-Dencun blobs made DA cheap enough that fee wars turned into a race to the floor.
In 2024, L2s did ~277M in revenue and ~113M of that (41%) went back to Ethereum for DA/security.
In 2025, L2 revenue fell ~53% to ~129M even with higher usage, and payments back to Ethereum collapsed to ~10M. The rest stayed with L2 operators.
Sequencers got paid, tokens mostly didn’t.
For example, Base dominated usage and revenue without even needing a token, and was the only L2 with a clearly positive 2025 bottom line (~$55M profit) after blob costs.
Meanwhile, most L2 tokens are still priced like they’re equity, when they’re mostly governance with dilution.
And smaller rollups can’t efficiently fill blobs. They either waste capacity or post less frequently, which worsens UX.
Activity on smaller rollups also fell ~61% since June 2025 in some datasets. A lot of them are just zombie chains now.
The long tail is structurally disadvantaged.
-----
If L2s want to survive, they need to bring something new to the table. There are niches they can own like privacy, app-specific efficiency, ultra-low latency, non-EVM specialized VMs…
But first, any chain that can’t move to Stage 1 is probably going to die this year.