i think, what we build now, was already envisioned way before we built it. up to 50-60 years. if you look at papers from early system designers they were definitely looking to.build neural networks etc. already. if you look at current new research concepts you might learn more about the future 50 years on. i think ibms light/photon based chips, and intels microfibers to connect chips etc. is super interesting. maybe well all have small devices with these photonic chips connected by microfibers. everything working at lightspeed seems a fair direction. these devices can even be made on current production lines. i also think, but am not sure about this,.that they might help with issues around power consumption which is a direction also being steered in. to try and get around moores law perhaps, or optimize the power for performance ratio