In effect, memory becomes a record of the agent's reasoning process, where any prior node may be recalled to inform future ...

But why, though?

Google researchers have revealed that memory and interconnect are the primary bottlenecks for LLM inference, not compute power, as memory bandwidth lags 4.7x behind.

On-package memory would've put Intel in a tough spot if not.

Intel said Thursday it expects its CPU shortage to peak in the first quarter as the company indicated that the ongoing AI ...

It won't come online until 2028, so its effect on the current memory shortage will be negligible.