When it comes to chip manufacturing, no single country has it all in-house. Neither US, nor Taiwan, nor China can make state of the art chips without machines, materials, chemicals and support from a bunch of other countries.
For personal assistant AI it seems to me that improvements in efficiency will make this a non-issue. We’re able to squeeze more and more out of smaller language models. Eventually the models that require 100s of GBs in GPUs and giant datacenters will not be able to provide enough additional capabilities to justify their cost. Most tasks will run on-device with 10s of GBs.
Don’t know if there’s hope of state of the art CPUs, GPUs and/or NPUs being designed and manufactured in Europe. It has a lot of the expertise. Imagination Technologies and ARM designs GPUs in Europe. But the scale is lacking
For personal assistant AI it seems to me that improvements in efficiency will make this a non-issue. We’re able to squeeze more and more out of smaller language models. Eventually the models that require 100s of GBs in GPUs and giant datacenters will not be able to provide enough additional capabilities to justify their cost. Most tasks will run on-device with 10s of GBs.
Don’t know if there’s hope of state of the art CPUs, GPUs and/or NPUs being designed and manufactured in Europe. It has a lot of the expertise. Imagination Technologies and ARM designs GPUs in Europe. But the scale is lacking