SQL Server 2019

By using db<>fiddle, you agree to license everything you submit by Creative Commons CC0.

create table mytable (item varchar(1), itemb varchar(1), date date);
insert into mytable values
	('X', 'A', '2014-01-01'),
	('X', 'B', '2014-01-02'),
	('X', 'C', '2014-01-02'),
	('X', 'D', '2014-01-04'),
	('Y', 'A', '2014-01-01'),
	('Y', 'B', '2014-01-02'),
	('Y', 'C', '2014-01-02'),
	('Z', 'B', '2014-01-02'),
	('Z', 'C', '2014-01-02'),
	('Z', 'D', '2014-01-04');
10 rows affected

select * from mytable


item
itemb
date


X
A
2014-01-01

X
B
2014-01-02

X
C
2014-01-02

X
D
2014-01-04

Y
A
2014-01-01

Y
B
2014-01-02

Y
C
2014-01-02

Z
B
2014-01-02

Z
C
2014-01-02

Z
D
2014-01-04


with 
	tab as (
		select 
			t.*, 
			dense_rank() over(order by item) drn
		from mytable t
	),
	cte as (
		select t.* from tab t where drn = 1
		union all
		select t.*
		from tab t
		inner join cte c on t.drn = c.drn + 1 and t.itemb > c.itemb
	)
select item, itemb, date
from cte c
where itemb = (select min(itemb) from cte c1 where c1.item = c.item)


item
itemb
date


X
A
2014-01-01

Y
B
2014-01-02

Z
C
2014-01-02

item	itemb	date
X	A	2014-01-01
X	B	2014-01-02
X	C	2014-01-02
X	D	2014-01-04
Y	A	2014-01-01
Y	B	2014-01-02
Y	C	2014-01-02
Z	B	2014-01-02
Z	C	2014-01-02
Z	D	2014-01-04