Given that even fast connections between SF and Tokyo average around 110ms (1), is < 100ms an arbitrary goal that has been pulled out of thin air, or are you deploying new technology that's faster than a typical fiber connection?
1) https://hello.agora.io/rs/096-LBH-766/images/Agora_WP_SD-RTN...
In fiber, with light traveling at about (2/3) * c, the minimum latency from SF to Tokyo would be about 42 ms.
The physical fiber optic floor from SF to Tokyo is about 50ms, point-to-point, with no sub-fiber last miles, wifi access points, or delay from routers and switches.
Actual user-to-user latency is much higher than that. Agora, for example, cites:
> Low latency, 400ms average globally