What I want to see, is cheaper high transistor devices, if there’s an oversupply of transistors, then. A device, with 64 GB of GDDR6, 50 billion transistors in the combined GPU, CPU, at 4nm, 4TB in PCIe5, should be possible, at a reasonable price, with lots of machine learning. For artificial intelligence neural network processing algorithms, for example video HDR, dynamic processing algorithms, chip designs, that can change architecture, according to software demands.
In mini computers, large chips, running at lower power rates, more transistors, lower frequencies, more massively parallel, scalable, running on efficiency cores most of the time, high end, for very short periods, to keep down heating inefficiencies. The jobs get done faster, the chip, doesn’t heat up and become inefficient, give us some AR, linked with WiFi 6, so we can interact with our friends and environment, at a reasonable price, chips around the house, interacting, transceiver relaying amongst themselves, so we can see the avatar, we’re talking to, in our environment.