AMD has orbited the Sun and is back to the pps I considered buying in at last July. Doesn't pay a dividend so no point in jumping in quick. This is the most unusual market I have seen in 50 years of playing this game. Insane, actually.
For this industry, people will have to accept the fact that Moore's law, in the classical sense, is finished. There are three directions to go in...the third dimension, topology, and qubits.
Value quantum computing, meaning having a market, is decades away. The only "problem" it solves is very large prime number factorization (this is a class of problems, like binary addition, which is the basic "problem" that a classical computer solves). Quantum computers will output a probability distribution, ie, Z=XY where X and Y are very large primes, but you only get an output range of many probable X's and Y's...you then check the results of candidate X times Y with a classical machine. Algorithmically
this can be exponentially faster than a classical sieve.
So, the industry will consist of finding apps for this problem paradigm.
The third dimension is a novel way of trying to stay on a Moore-like trajectory, and may even be faster. My guess is a chip can become a grain of sand, but thermodynamics becomes the limitation...theres a lot more entropy in D=3 than in D=2.
Topology is the really intriguing future way to go. Basic circuit design may change profoundly; Ohms law, normally framed with complex numbers, becomes a far more sophisticated thing when cast into quaternions , octonions, unitary groups, etc. electron spin will become part of the current; current carriers become topological collective modes..solitons, monopoles, etc. Edge states and winding numbers come into play. This generally involves the D=2/D=3 interface, so it is 3 dimensional in a sense, but the induced effects occur at the boundary. photons and electrons both participate in switch.
There's a future and it isn/t who has the smallest node.
