Because the directory you specified does not exist!
Since you have a sufficiently recent version of CMake, you can simplify the build procedure: instead of creating a build folder and changing to the build folder, you can just stay in the main LAMMPS source folder and use: cmake -S cmake -B build ... and then cmake --build build to compile.
BTW: it is not advisable to use folder/path names with whitespace in them.
If you want to add a package to an executable/library in an existing LAMMPS build folder, you can just do: cmake -DPKG_NEWPKG=on -B build. It will first read and apply the cache from the previous configuration and then change the desired setting. Of course, after configuring, you need to compile again with cmake --build build to update library and executable.
P.S.: when building a shared library (i assume for use with python) it is strongly advised to also use the -DLAMMPS_EXCEPTIONS=on setting.
There are detailed installation instructions for compiling LAMMPS from source in the LAMMPS manual. Also there is a short CMake tutorial and CMake has a great documentation on their home page. There is no point in continuing this discussion until after you have studied those well enough so you can understand my explanations (most likely you won’t need them at that point).
Why do you then try to compile from source and do not use a precompiled installer package for Windows?
Then you don’t need the PYTHON package.
The https://rpm.lammps.org/windows/LAMMPS-64bit-stable.exe package for example should “just work™”. It has been tested with multiple Python distributions including those from python.org and the one included in Visual Studio or the Intel OneAPI compilers.