reth 2.0 hitting 1 gigagas/s execution is a real milestone. but on its own it doesn't move the needle much. execution has never been the binding constraint on L1 throughput. consensus and data propagation are, and have been for a while. a faster execution client running on top of the same consensus layer still inherits that ceiling. most L1s still order things sequentially: sequencer produces a block, consensus commits, and execution runs. when execution is fast, the limitation is how quickly you can agree on what to execute next. if consensus takes 500ms and execution takes 5ms, you've got more performance than you can actually use. the interesting architectural work is improving consensus, and then decoupling consensus from execution. async execution, where consensus reaches agreement on ordering without waiting for state, and execution runs in parallel. multi-proposer chains like Giga are designed to make these performance gains usable