> IMO, Migrating to small-scale local LLMs would be a significant improvement over using data centers.
LLM serving is most efficient when you batch a lot of parallel requests together. Data center solutions also have the advantage of collecting queries from around the globe, so the hardware can be utilized around the clock.
Having everyone serve their own local LLMs would produce a lot more memory demand. Not less. The same memory would be idle most of the time, and when it was used it would be used for 1 person instead of a batch of requests.
There are other reasons to run local LLMs, but solving hardware demand problems is not one of them.
Inefficient per watt, yes, but local inference capacity is greatly underutilized in aggregate. If a model can run on a machine that already exists, that's a bunch of additional chips that don't need to be built.
You picked up on what I didn't write existing machines don't have to be built, you just use them.
It occurred to me much later that local machines could become a shared resource of a small co-op. Or it only runs off solar or energy harvested at times of excess generation (too much daytime solar, lots of wind, or overnight when demand naturally drops).
Exploring how to partition inference across many machines and shed load for idle sections of hardware would be a problem that could be solved by a community of smart people like us here
LLM serving is most efficient when you batch a lot of parallel requests together. Data center solutions also have the advantage of collecting queries from around the globe, so the hardware can be utilized around the clock.
Having everyone serve their own local LLMs would produce a lot more memory demand. Not less. The same memory would be idle most of the time, and when it was used it would be used for 1 person instead of a batch of requests.
There are other reasons to run local LLMs, but solving hardware demand problems is not one of them.