Here's a straightforward approach:
data want;
set have;
by subid;
length all_names $ 200;
if first.subid then all_names = names;
else all_names = catx(',' , all_names, names);
if last.subid;
retain all_names;
run;
The only thing I would warn you about is to consider whether this is really a good step to take. If you merely want to print the data in this new form then go right ahead. But if you are analyzing or summarizing the data for some other purpose, SAS has many tools that let you process multiple observations per SUBID. It may not be necessary to change the form of the data.