Vitalik just re-priced the whole L2 trade in public. The easy era of cheaper Ethereum is over.

Two numbers explain the shift:
→ L2 user activity is down about 50% from peak, while mainnet activity snapped back hard once fees got cheap again
→ blobs made rollup data so cheap it’s not the bottleneck anymore, around $0.04 per MB
So what happens when the only edge is low fees, and the base chain starts handing out low fees too?
A lot of quiet deaths. Major L2 token market cap is down to ~$6.8B, with $ARB, $OP and friends down ~90% from highs.
Look at who’s still alive on pure activity:
▫️ Base is printing pure retail scale
~11.0M tx/day, ~1.5M DAU, $9.9B TVS, ~$5M fees in 30d. Did this without a token. Coinbase funnel + apps that actually drag normies onchain.
▫️ Arbitrum is the DeFi bank shard
~2.9M tx/day, ~1.2M DAU, ~$15B TVS, ~$700K fees in 30d. Also doing the real economic activity. Stablecoin transfers are a massive chunk of non-L1 flow, and Arb is eating that lane.
▫️ Optimism grinding OP Stack empire
~2.4M tx/day, ~800k DAU, ~$1.66B TVS. Even if OP mainnet isn’t the hottest, the Superchain direction is the meta for how mediocre OP forks survive, by merging into something bigger instead of pretending they’re special.
▫️ Lighter is the new horse that has product-market fit
39.1% user retention, ~$1B TVL, $6.12M fees in 30d. Users actually using the chain to trade perps.
Now the ugly part, who’s getting killed:
– Metis: ~3k tx/day, ~500 DAU, $38M TVS, dead Vitalik mom narrative
– Boba: ~1k tx/day, ~300 DAU, $5.6M TVS, ghost town with RPC endpoints
– Polygon zkEVM: ~5k tx/day, $9.8M TVS, no ecosystem gravity
– Loopring: ~200 tx/day, $11M TVS, market moved on
– Blast, zkFair, Manta Pacific, Mode, plus a pile of tiny OP Stack and Polygon CDK clones with fading TVL and usage down ~60% from top
Even the more “legit” names show the same pattern if they don’t ship differentiation fast enough.
– Linea is ~70k txs/day, $406M TVS, but still Stage 0 with centralized keys, and never really formed a native culture or app magnet
– Scroll is ~200k txs/day, $99M TVS, Stage 1, but still feels like a zkEVM lab more than a place users live
– zkSync Era is ~50k txs/day, $399M TVS, Stage 0, and hasn’t found its niche yet
– Starknet has good tech, Stage 1, but hasn’t proven it in usage
Only a handful are Stage 1. Nobody meaningful is Stage 2 yet. Vitalik basically said some operators privately don’t even plan to remove training wheels, which means they’re a separate chain with a bridge.
That callout forces the L2 landscape into:
→ Generic EVM L2s with no moat either join a bigger stack or fade. OP Superchain absorbs, Orbit absorbs, everyone else becomes a footnote
→ L2s need a reason beyond scaling, like privacy, app-specific efficiency, extreme scale, or non-financial design
→ Stage 1 becomes table stakes for any chain holding real assets
→ Token models either evolve or get ignored. Governance-only doesn’t work. If a token exists, it probably needs to map to something concrete: shared sequencing, revenue routing, provers, interoperability economics
The copy-pasta of rollups is over. The market doesn’t need 50 Ethereums-but-cheaper anymore. It needs execution environments with a reason to exist.