Where did you get that number?

Vitalik gave goals of 1.33 MB per second in blob space, and a compressed tx size of 25 bytes. This gives around 50,000 transactions per second, which seems like a worthy goal.

I would have sworn I read it in the linked article, but it's sure not there when I read it again. And it's not in the oldest archive.org copy either. I can only conclude I hallucinated it, which makes me uncomfortable.

Thanks for responding

I'm curious what you think the latencies of 50,000 distributed transactions becoming a consistent chain add up to?

The reason L2s can scale that high is every machine doesn't have to process every transaction.

Each L2 is its own chain that uses its own sequencer/s. The blocks from this chain are then compressed into a single blob of data with a ZK proof that the transactions are valid. The validators on L1 only need to verify that ZK proof matches the hash of the submitted data, which can be done in a few ms even if the L2 did 1M+ transactions.