GPU Accessibility Gap and Mojo

Every Engineer Is a Manager Now ๐Ÿค– โ€” with Chris Lattner

Topic 2 of 10

Published Apr 3, 2026 ยท Clip 7:10โ€“15:36 (8:26)

Chris describes the accessibility gap in accelerator programming: 99% of engineers program CPUs while GPUs remain the domain of few wizards. He draws an analogy to Swift replacing Objective-C, which multiplied the number of iPhone app developers. He explains Mojo's approach โ€” Python syntax with a new compiler stack built on MLIR, delivering performance parity with CUDA while remaining portable across chips, and enabling AI-driven Python-to-Mojo migration with huge speedups.