Max number of L2s using ETH as an L1?

Hey, so Messari recently put out a report where they claim that an optimistic rollup can run at a max tps of 5.2k, and that's only if rollup batches occupy the entire ETH L1 block (here's the tweet about it) (see the second row "ETH + Optimistic Rollups"). So this got me thinking, does this mean that if activity of a single optimistic rollup takes off for like a web3 game or DEX and it does run at 5k tps, then no other rollups or L1 transactions would theoretically occur on ETH? Taken to its extreme, this would mean that only one high-throughput L2 could settle on ETH, or maybe only a handful of L2s running at high tps (5 rollups running at 1000 tps?), and this can't be correct, right? Otherwise, why are all these groups building optimistic rollups. I know I must be missing something, or I'm hoping that the Messari report is incorrect, but please help me out here. Thanks in advance!

Edit: Added words to increase clarity

Edit2: Okay, so I believe that I may be correct, and there is a limit to the tps, but with the addition of danksharding, maybe we need to multiply this max tps number by 64 shards (here's the Ethereum website explaining danksharding). So basically, the Messari report states that rollups vary from being able to run at 2k (zk rollup), 5k (optimistic rollup), or 20k (validium) tps. So maybe if there's only one validium rollup per shard, running at max 20k tps, it gets ETH overall to 1,280,000 tps. More realistically, there will be a mix of zk and optimistic and validium rollups distributed across the 64 shards, yielding a range of tps between 128,000 (only zks) to 1,280,000 (only validiums). If correct, this is much more understandable. Maybe we need to keep this in mind when people post that there will be thousands of L2s on ETH, they likely don't understand the network limitations as soon as tps greatly increases on a few rollups. For example, in the future after danksharding is implemented, just 64 zk rollups running at a max of 2k tps would completely fill the entire available blockspace on ETH.

Edit3: Just realized L3s will settle on L2s, further amplifying the scaling. Feeling much better about ETH's future now.