If CloudFlare is running IPFS naked, then IPFS seems to be scaling.
If CloudFlare is caching IPFS, then it is CloudFlare saving their butts.
Can we confirm which it is?
I'm trying to chat with CloudFlare about my own P2P protocol, which I've seen handle HackerNoon's 15M monthly users (I saw about 10K concurrent users per second at peak load), to run GUN ( https://github.com/amark/gun ).
Because I do think it is important to test all these protocols out at bigger/larger scales, that certainly makes it easier to then debug and then fix problems (assuming CF isn't caching).