Indeed, I completely screwed my math up. Looks like 10am is too early in the morning for a Sunday.

33 tps max token generation speed would be for 10ms of network latency in the above example.