An EVM chain without an optimistic bridge (an alt L1) is worse; the space doesn't need more copypasta EVM chains or more L1s. L1 scaling will bring significant EVM blockspace, though some use cases (e.g., AI) will still need more blockspace and lower latency.