Devin McGloin

In general I think there will be a trend towards raw materials becoming smarter and more efficient as well as computing becoming more contextual.

Computation

With Moore’s law coming to a close, I expect chip designers to work towards embedding chips in physical materials. Instead of focusing on a single computational source, computation would be distributed across thousands of individual CPUs.
Heterogeneous computing EG FGPA, Heterogeneous cores, quantum etc. Key to be able to target multiple platforms with a single program.
Everything that is currently a "dumb" device will be aided by networked computers. The security model for such devices however leaves a lot to be desired. This is mostly due to the economics for how such goods are produced, and generally such devices should not be considered secure.
I expect cloud computing to become more fundamental to how computers operate. Devices that users interact with shift even more so towards displaying / networking instead of carrying out computations locally. Compute and storage resources will be fluid across multiple devices. This requires a definitive identity for each user, which along with the networking costs are the primary hurdles.
Standards were created very early on, and therefore don't realize the full potential of the machine. Therefore with better / more accurate standards, which are now possible we can have more accurate / useful data formats.

Solar power will continue to drop in price and will become the dominant roofing material in the developed world.
Embedded devices will continue to optimize energy consumption in the home, mostly powered by AI which will get its start in enterprise. There will however most likely be a few major security scares before we get to this point.
Increasing prevalence of localized power sources

Programs that allow for introspection, and explain themselves.
Computational Building Blocks, simple com posable operations that can be nested without considering the internal implementation.
Increasingly distributed applications that force developers to come up with better models for parallel computation. Distributed both on the same physical device (multiple cores) and also between networked computers.

There is a huge missed opportunity for fast transport into rural environments. Faster trains would work, opening up countless communities for commuters to live in outside of large cities. The faster the train the larger the number of sub communities are possible. Self Driving cars will be transformative, but I think there are larger gains to be made with faster, cleaner rail transport.