I have a native Intel GPU programming environment:

✅ oneAPI 2026
✅ DPC++ (SYCL)
✅ Level Zero
✅ OpenCL
✅ Intel Compute Runtime
✅ MKL
✅ MPI
✅ VTune
✅ Intel Debugger

This is actually a better environment for Intel hardware than CUDA. Both c++ and FORTRAN compilers are fully operational.
When customers insist on using CUDA code I have a reach tru available.