[waffle_ss]

I'm wondering if Google will use this as a way to expand their compute footprint into more POPs / edge network locations.

Everyone is hung up on latency, which is true. So if Google deploys their game rendering boxes at edge networking locations, like internet exchanges, that will result in lower latency to end users. Once Google has their rendering boxes there, they might as well sell other computing services as well. Why not sell to the game makers themselves, too - if they put their matchmaking/host servers in the same place, then latency is lowered even more.

IIRC Google already has the largest private network (most fiber-miles) with the most POPs compared to other cloud providers. The more POPs you have, the better you can do CDN and other geographic-sensitive computing since you are physically closer to your end users.

Having mini-datacenters at edge network locations could also be an interesting bet on future/emerging technologies that are also compute-heavy and latency sensitive, say VR or AR. Imagine being able to deploy code to thousands of locations (for a price) compared to the traditional couple dozen regions * a handful of availability zones where the huge datacenters are.

[wppick]

These are great points. My two cents: I think there's potentially a lot more behind this product than just the face value of game streaming. Amazon is growing their compute power with aws, and other people are paying for this. Google needs something similar to grow their compute resources, and to lower risk by also having others pay for it. It's an arms race for compute power. With stadia Google will have an advantage over Amazon in gpu resources. Amazon could grow gpu resources by getting into the video rendering business with massive shows like Lord of The Rings to finance it. The other big win for Google imo is having this be a passion project for good engineers to pour their energy into