Cube.AI support only Cortex-M4 and Cortex-M7 based STM32 taking advantage of the FPU and the DSP instruction set.
Running a neural network on a Cortex-M4 or Coretx-M7 will be much more efficient than on a Cortex-M0 and there are STM32 Cortex M4 at very cost effective price. Typically the F4 serie.