You can create two columns one for the string data and one for the numeric data. Assuming you have SQL Server 2016 or higher you can use the TRY_CONVERT or TRY_PARSE function.

SELECT
    [INVMB.MB110_INT] = TRY_CONVERT(INT, [INVMB.MB110])
    ,[INVMB.MB110_STRING] = [INVMB.MB110]
    ,[IsNumberDataType] = CASE WHEN TRY_CONVERT(INT, [INVMB.MB110]) IS NULL THEN 0 ELSE 1 END

The TRY_CONVERT function will convert the string to a numeric if it is possible otherwise it will return NULL.

Note that the TRY_CONVERT function will return a numeric value of 0 if the source string is an empty string. If you don’t want 0 to be retuned for an empty string, the use the TRY_PARSE function.

If you absolutely require the values in a single column, you can convert the values to a SQL_VARIANT types (after you converted the string to the desired base type). I personally avoid SQL_VARIANT types like the plague, there is just too many performance and implicit conversion issues associated with them. I have included an example of a query that returns some sample data.

WITH VariantSampleData
AS
(
    SELECT 
        variant_value = TRY_CONVERT(SQL_VARIANT, TRY_CONVERT(INT, '123') )
    UNION ALL
    SELECT
        variant_value = TRY_CONVERT(SQL_VARIANT, 'abc' )
)
SELECT
    variant_value
    ,BaseType = SQL_VARIANT_PROPERTY(variant_value,'BaseType')
    ,[math] = TRY_CONVERT(INT, variant_value) + 1
FROM
    VariantSampleData