IBM Research has developed a chip using phase-change memory (PCM) encoding the weights of a neural net on the chip itself.

The in-memory computing hardware showcased here increases the speed and energy-efficiency needed for the next generation of AI.

https://aihw-composer.draco.res.ibm.com/

https://www.technologyreview.com/2019/06/06/239031/training-a-single-ai-model-can-emit-as-much-carbon-as-five-cars-in-their-lifetimes/

https://soccernurds.com/uncategorized/916650/memristor-market-outlook-2020-global-topmost-companies-size-trends-and-upcoming-forecasts-2027/

We’re changing the game with a unique combination of building blocks that work together to deliver unparalleled performance and new computing possibilities across a breadth of markets.

https://www.intel.com/content/www/us/en/architecture-and-technology/intel-optane-technology.html

https://www.youtube.com/watch?v=_2Wiql4QSLQ

https://www.sciencedaily.com/releases/2019/05/190504130308.htm

The first programmable memristor computer—not just a memristor array operated through an external computer—has been developed at the University of Michigan.

More information:

https://news.umich.edu/first-programmable-memristor-computer-aims-to-bring-ai-processing-down-from-the-cloud/