As far as I can see, I can make an UPDATE in PL/SQL without using a CURSOR. For example:
CREATE OR REPLACE PROCEDURE update_my_table (id NUMBER)
AS
...
BEGIN
...
UPDATE my_table SET column=NULL WHERE my_id = id;
...
END
I guess the cursor is only needed then if you want to perform a fetch of the updated row, is that right?
Oracle uses cursors under the hood for the update and for selects, which is going a bit beyond what you're asking. But no, you don't have to select the rows to be updated in a visible explicit or implicit cursor.
You don't even have to use a cursor to select the row though. you can do a select ... into:
CREATE ... AS
my_row my_table%rowtype;
BEGIN
select * into my_row from my_table where my_id = id;
-- do something with the row data
dbms_output.put_line(my_row.my_col);
END;
Or you can select individual columns into separate variables instead of into a rowtype variable. But you have to get exactly one row (or one row's worth of columns) from the query; if there are no matches you'll get a no data found error, if more than one then too many rows. You can also select multiple rows into a PL/SQL table and manipulate the data there, still without a visible cursor.
However, if you want to select and update the row, you probably do want a cursor with a select ... for update clause, and then update ... where current of ... to prevent someone else modifying the data between the select and the update.
A simple update as you've got it now will be the most efficient way, whether you're updating one row or millions.
select for update the rows are locked until you either commit or rollback the transaction; whether you actually update them or not.