mainframe -> pc -> laptop -> phone -> cloud/gpu/ai -> blockchain/vr/... -> quantum/self-driving-car/... .
Each of those bring massive high-margin software tides to ride.
In practice, life's still pretty sweet for cloud/gpu/ai startups, so we still seem to be roughly at that point. (We do a lot of gpu cloud accelerated security/fraud visual graph analytics: most enterprises still do it on-prem and basically just hope splunk can run regex without dying.) In GPU land, it's pretty clear they're being way underutilized, esp. due to slow SW rollout by various incumbents like cloud vendors.
We're still early on the transition to the next phase. The blockchain/vr/quantum startups raised a ton of funding, but have largely fallen flat. Some of the fundamentals in them seem sound, so they seem more of a matter of being early, not necessarily wrong. Current teams just couldn't get the tech together, or a true killer app: those are a question of time. Likewise, we know quantum etc. are coming, just need ~5 more years for what most general sw people start to get into.
IMO bigger risk is monopolies. Maybe the market reasoning is more from VC perspective, where everything feels 'expensive', and SaaS metrics era made investors forget how to take tech risks while the coming gen of platforms emerge.
EDIT: Some other potential near-term up-and-comers - synthetic bio (crispr,..), NN moving into program synthesis (so past perception and into automation), serverless. I don't follow agtech, but drones/robots/etc. are increasingly here.