October 12 2025. I read some announcement by microsoft that they managed to use microfluidic channels to cool their chips. It was neat, and in a mems class I took with Lin liwei he brought up something similar one day, and I didn't think much of it since I figured (and was right) that it was possible with existing processes, and what was more important was developing new processes.
Somehow thinking about this microsoft announcement led to the rowhammer exploit coming to mind. When I first learned about rowhammer I was completely shocked that phenomena at the semiconductor physics level could be exploited at the application level. It takes quite a bit of ingenuity to strike through abstraction.