Our software company is not even 10 people big and we are able to deliver extremely high quality software multiple times per day to many customers, while supporting a wide range of business contexts. The annual budget for our whole team is something like $750k. Our part of the AWS bill is less than $400/m, but we seem to make due and get shit done.
Blizzard in particular has something like 2 billion dollars in free cash flow to work with. Obviously not exactly the same problem domain but effectively the same sort of cost structure around core product development. Nothing ever works out linear at scales like this, but even assuming a 10x loss in efficiencies of scale (i.e. due to organizational overhead), you would still have resources to run ~200 copies of our software team in parallel on ~200 unique projects. How many teams would have to succeed before that model pays off? 1/200 figuring out how to build a proper WoW killer could get the job done...
To Blizzard I say: Take all that capital the gaming community has been injecting into your money bin for the last 2 decades, and invest in an internal startup culture. Spin out tens or hundreds of smaller teams with their own independent creative direction to try out new ideas. If you find some hits on one of these teams, maybe you take a less successful team and add their resources to the more successful one. In this way, you can dramatically expand the portfolio & diversity of the product base. This has caveats for any "common" creative vision, but judging by the comments in this thread, you may want to re-evaluate any convictions around that idea.