I use DXGI_FORMAT_B8G8R8X8_UNORM as the format of a D3D11_TEXTURE2D_DESC.

How does it come that a uint32_t of 0x000000FF corresponds to blue instead of black (0xFF000000)?

Is this due to endianness and if so why such an ambiguous name (since the majority of computer architectures uses little endian)? 

The DXGI names are little endian. B8 is listed first, meaning it's byte #0 - the byte at the little end.

D3D9 used a big endian naming convention, where that format would be written as "ARGB".

Thanks for the clarification.