It's hard to imagine CS inventing something more impressive than modern AI.
So maybe the future of CS is mostly applied and mostly in biology such as giving blind people the ability to see and paralyzed people the ability to walk via a Neuralink type brain implant.
Or lean into the "science" part of CS. There is a lot to examine about how complex systems actually operate. Like what does "typical" network traffic look like? Is there even such a thing that can be broken down into categories like "enterprise", "home", etc.? It's been hard to create a tool to mimic or create synthetic traffic loads for testing.