For a more robust solution, don’t store VALID_UNTIL because VALID_FROM is enough. When you need to obtain the prices for a given date just use the following query:

select top 1 *
from MA_COSTS
where ITEM_ID = @ItemId
and VALID_FROM <= @DateOfInterest
order by VALID_FROM desc;

There is no advantage in storing VALID_UNTIL – it just makes things more complicated.

And ensure you change the datatype of your columns – DO NOT USE FLOAT to store money! Float is an inaccurate datatype intended for scientific data… you need a precise datatype, you don’t want to lose money due to the behaviour of float. Use something like decimal(18,2)

I like Dale K’s idea of eliminating the VALID_UNTIL column and using the CROSS APPLY (SELECT TOP 1 ...) pattern for looking up prices.

However, if that isn’t an option, the following should satisfy your needs.

CREATE TRIGGER trg_update_valid_until
ON MA_COSTS
AFTER INSERT
AS
BEGIN
    SET NOCOUNT ON;

    UPDATE tPrior
    SET VALID_UNTIL = i.VALID_FROM
    FROM inserted i
    CROSS APPLY (
        -- Locate immediately prior cost row for this item
        SELECT TOP 1 t.*
        FROM MA_COSTS t
        WHERE t.ITEM_ID = i.ITEM_ID
        AND t.VALID_FROM < i.VALID_FROM
        ORDER BY t.VALID_FROM DESC
    ) tPrior
    WHERE tPrior.VALID_UNTIL = '2099-12-31'
    --or perhaps: WHERE tPrior.VALID_UNTIL > i.VALID_FROM
END;

For each inserted row, the prior row is identified and updated as needed.

An index in MA_COSTS(ITEM_ID, VALID_FROM) is needed to support direct lookup of the related rows using an index seek. If (ITEM_ID, VALID_FROM) is the primary key, no separate index is needed.

If you might be inserting rows with VALID_FROM values in the middle of the existing range, a second update may be needed to set the VALID_UNTIL values for the newly added row(s).

...
    UPDATE tCurrent
    SET VALID_UNTIL = tNext.VALID_FROM
    FROM inserted i
    JOIN MA_COSTS tCurrent
        ON tCurrent.ITEM_ID = i.ITEM_ID
        AND tCurrent.VALID_FROM = i.VALID_FROM
    CROSS APPLY (
        -- Locate immediately following cost row for this item
        SELECT TOP 1 t.*
        FROM MA_COSTS t
        WHERE t.ITEM_ID = i.ITEM_ID
        AND t.VALID_FROM > i.VALID_FROM
        ORDER BY t.VALID_FROM
    ) tNext
    WHERE tCurrent.VALID_UNTIL > tNext.VALID_FROM

Note that the above does not handle updates or deletes that may require adjustments to other rows.

Sample results after multiple insert operations with rows presented in a mixed order:

See this db<>fiddle for a demo.

I have slightly corrected the description of the table. It’s more consistent with your text.
Mainly – set default value for column VALID_UNTIL.

create table ma_costs(
  ITEM_ID VARCHAR(255) NOT NULL ,
  COST1 FLOAT NOT NULL,
  COST2 FLOAT NOT NULL,
  PRICE1 FLOAT NOT NULL,
  PRICE2 FLOAT NOT NULL,
  VALID_FROM DATE NOT NULL , 
  VALID_UNTIL DATE default '2099-12-31'  NULL,
  PRIMARY KEY(item_id,valid_from)
);
create index ix_ma_costs_item_id_valid_from on ma_costs(item_id,valid_from);

Your trigger works, when you insert 1 row in query. Then there is 1 row with the value valid_until=’2099-12-31′. JOIN is correct

ON t.ITEM_ID= i.ITEM_ID
 AND t.VALID_UNTIL = '2099-12-31' AND t.VALID_FROM < i.VALID_FROM

When you insert batch, and default value for valid_until is ‘2099-12-31’ INNER JOIN in trigger can not work correctly. There only 1 row in t match to all rows in i.

See example:
For test. Insert 1 row without trigger. Default for valid_until is correct.

INSERT INTO MA_COSTS (ITEM_ID, COST1, COST2, PRICE1, PRICE2, VALID_FROM)
VALUES
(2008201, 4062.71, 4062.71, 5803.81, 6732.41, '2024-10-01');
select * from ma_costs;

Your trigger:

CREATE TRIGGER trg_update_valid_until
ON MA_COSTS
AFTER INSERT
AS
BEGIN
    SET NOCOUNT ON;

    UPDATE t
    SET t.VALID_UNTIL = i.VALID_FROM
    FROM MA_COSTS t
    INNER JOIN inserted i
        ON t.ITEM_ID= i.ITEM_ID
        AND t.VALID_UNTIL = '2099-12-31'
        AND t.VALID_FROM < i.VALID_FROM;
END;

Insert rows one by one (item_id=2008201).

INSERT INTO MA_COSTS (ITEM_ID, COST1, COST2, PRICE1, PRICE2, VALID_FROM)
VALUES (2008201, 4062.7, 4062.7, 5803.85, 6732.47, '2024-12-01');
INSERT INTO MA_COSTS (ITEM_ID, COST1, COST2, PRICE1, PRICE2, VALID_FROM)
VALUES (2008201, 4119.57, 4119.57, 5885.09, 6826.7, '2024-12-02');
INSERT INTO MA_COSTS (ITEM_ID, COST1, COST2, PRICE1, PRICE2, VALID_FROM)
VALUES (2008201, 4150.85, 4150.85, 5929.78, 6878.54, '2024-12-06');

Output is correct

Insert several rows simultaneously (item_id=2008209)

INSERT INTO MA_COSTS (ITEM_ID, COST1, COST2, PRICE1, PRICE2, VALID_FROM)
VALUES
(2008209, 4062.7, 4062.7, 5803.85, 6732.47, '2024-11-01'),
(2008209, 4119.57, 4119.57, 5885.09, 6826.7, '2024-11-02'),
(2008209, 4150.85, 4150.85, 5929.78, 6878.54, '2024-11-06');

fiddle

My suggestion is

CREATE TRIGGER trg_update_valid_until
ON MA_COSTS
AFTER INSERT
AS
BEGIN
    declare @nn bigint;
    set @nn=next value for sq_trigger_work_count; -- for test
    SET NOCOUNT ON;
    UPDATE t
    SET t.VALID_UNTIL = i.VALID_FROM
          -- log operation to column testValue
       ,t.testValue=concat('(',i.valid_from,'-',i.valid_until
                  ,':',t.valid_from,'-',t.valid_until
                  ,':',(select count(*) from ma_costs)
          ,')'   ) 
    FROM MA_COSTS t
    INNER JOIN inserted i
        ON t.ITEM_ID= i.ITEM_ID
      and t.valid_from=(select max(valid_from) from ma_costs t2 
                        where t2.item_id=i.item_id and  t2.valid_from<i.valid_from)

      ;
END;

Details see in example fiddle