In particular simple case, when: 1) there not more than 4 columns, 2) column data fits nvarchar(128) (sysname), 3) no dots in data - you may use parsename function like this:

declare @t table(c varchar(100))

insert @t values ('AA,BB,CC'),('DD,EE,FF'),('GG,HH,II')

select

col1 = parsename(replace(c,',','.'),3),

col2 = parsename(replace(c,',','.'),2),

col3 = parsename(replace(c,',','.'),1)

from

@t

But for all cases, for universal stuation, you sould better use csv splitter function mentioned above.