Weebit - 2023 and beyond, page-1604

yeah - so the rudimentary AI wafer scale engine laid out the possible ....

Nvidia and AMD ( i suspect tenstorrent as well) already have high performance AI compute memory processors with advanced interconnect and packaging without attached GPUs' ....this is essentially their own Neural Processor Unit (compute in memory if you will) and these are being implemented within the respective companies to produce and refine bench SoC solutions that achieve much improved system performance and cost flexibility.....they are end use /application specific and are on par now with high end smart phone apps.
Supposedly Oppo produced and manufactured an NPU in partnership with TSMC (6nm process node)...tsmc won't confirm as they don't discuss their customer business...Oppo have temorarily shut down wafer chip operations .....
STMicro have an MCU /NPU ....looking at market capture and IOT / edge applications into an exploding market segment
https://www.eenewseurope.com/en/st-to-launch-its-first-neural-microcontroller-with-npu/




id like to know more about the LG /Tenstorrent transaction ....seems to be a strategic partnership into exactly this space but getting details is difficult.

https://www.reuters.com/technology/canadian-ai-computing-startup-tenstorrent-lg-partner-build-chips-2023-05-30/
they'd have to be AI compute in memory chips....i suspect
these guys are working AI cim back into production values on exisiting lines to achieve productivity and cost gains at scale....its the first real AI market application and its exploding.

.....and weebit and partners are lurking there in rram AI cim.....work in parallel is a strategic masterstroke of this company....as we know