I doubt their API would handle 100k requests per second. That math was roughly indictive of what the cost to send 100k requests per second would look like. He did mention that that was assuming the target didn't have rate limiting, either intentional or just pure limits of the hardware.