Heterogeneous compute solves where AI runs.
The next question is who authorizes what it is allowed to do.
As intelligence spreads across CPUs, GPUs, NPUs and edge devices, execution-time authority needs to be just as portable — and ideally hardware-rooted.
Caught up with @clattner_llvm right after ModCon. Expanded hardware support, Mojo now open source, Modular Cloud live, and a new industry alliance. As he put it: heterogeneous compute now has a software platform.- Heterogeneous compute solves where AI runs. The next question is who authorizes what it is allowed to do. As intelligence spreads across CPUs, GPUs, NPUs and edge devices, execution-time authority needs to be just as portable — and ideally hardware-rooted.
I keep coming back to this question. If AI workloads are going to move across CPUs, GPUs, NPUs and edge devices, can authorization really stay only in software? Or does it eventually have to be enforced in hardware, right where the action actually happens?
