By default, tablespaces are now locally managed, which means we use a bit map system to map out the file. Which means extents are defined in preset equally sized chunks (64k, 1m, 8m, ... etc)

For example

SQL> create table T ( x char(1000) )
  2  tablespace demo;

Table created.

SQL>
SQL> insert into t values (1);

1 row created.

SQL>
SQL> select extent_id, bytes
  2  from   dba_extents
  3  where  owner = user
  4  and    segment_name = 'T'
  5  order by 1
  6  /

 EXTENT_ID      BYTES
---------- ----------
         0      65536

We gave you a 64k chunk. What if you asked for larger initial ? Now I drop the table and repeat

SQL> create table T ( x char(1000) )
  2  tablespace demo
  3  storage ( initial 20m next 7m);

Table created.

SQL>
SQL> insert into t values (1);

1 row created.

SQL>
SQL> select extent_id, bytes
  2  from   dba_extents
  3  where  owner = user
  4  and    segment_name = 'T'
  5  order by 1
  6  /

 EXTENT_ID      BYTES
---------- ----------
         0    8388608
         1    8388608
         2    1048576
         3    1048576
         4    1048576
         5    1048576

6 rows selected.

You asked for 20m, so we gave you 2 x 8M, plus 4 x 1M

This is a good thing, because as tables comes and go, grow and shrink, then they'll all be able to grab free chunks from each other, because all of the size allocations are consistent.