Through the C++ foundations, PyArrow already has support for representing non-CPU data (e.g. you can have a pyarrow.Table object backed by buffers in GPU memory), but this is not yet exposed very well in the Python bindings. Specifically, over the last years, several features were added on the C++ side (the Device and MemoryManager classes, more recently a set of Copy functions that take a target memory manager) that don't yet have equivalent bindings in PyArrow.

• Add bindings for DeviceAllocationType, Device and MemoryManager and expose as attributes on the Buffer class: GH-41126: [Python] Basic bindings for Device and MemoryManager classes #41685
• Test Device and MemoryManager with CUDA (and integrate with existing pyarrow.cuda APIs?): GH-41126: [Python] Test Buffer device/device_type access on CUDA #42221
• [Python] Add bindings for CopyTo on RecordBatch and Array classes #42222