Friends of semiwiki: I am contemplating a framework for modelling enterprise/data centre computational demand (CPU/GPU)... we know that this market is dominated today by Intel, with Nvidia rising quickly in AI hyperscale environments... but other than simply extrapolating t-5 year worth of unit shipments (which is one method) is there a more sophisticated approach to think about modelling future silicon demand (surface area, transistor count, sockets, core)... based on the growth in workloads etc. Any guidance would be be helpful. rgds.