In the US at least, "sweat of the brow" does not by itself allow copyright protection; it is irrelevant how much work is done.
According to wikipedia, sweat of the brow implies:
Substantial creativity or "originality" is not required.
Designing APIs or instruction sets requires a ton of creativity and originality, since billions of lines of code are going to be using it and since changing them in the future is costly and difficult. The originality comes from improved API design that reduces coding time/debug time as is the case in Java API vs C API. Sweat of the brow type of work means something trivial like creating a database of people's names and phone numbers or printing phone books.
That being said, there is a lot of tedious, sweat of the brow concepts involved in designing APIs. For example, reading a text file still needs something like fileopen(), fileclose(), fileread() that is present in all programming languages in some form. But this API should still be copyrightable since copyright protects creative expression of an idea, but not the idea itself.
The TL;DR version: the API ideas/concepts are similar across many programming languages but the way they are expressed differently means they deserve copyright protection.
IP protection is poor and archaic because it was created in the 1700s where you only had to deal with protecting books and mechanical inventions. API/instruction set are a type of IP that seem like a blend of design patents and copyright. IP law should be extended to protect all types of intellectual property using different tools, not just patents and copyright.
I didn't read the NEC vs Intel case too much but it seems like copyright infringement because copying the body/implementation of a function API is similar to copying the microcode (or subassemly language) of a CPU.