I tried to reproduce the same in my environment.
My sample data:

to convert the column with nested Json
in the form of table. First, I created variable with nvarchar(max)
. set the select Querys value to it.
DECLARE @json nvarchar(max)
SET @json = (SELECT
TOP 100 *
FROM
OPENROWSET(
BULK 'https://dlsg2p.dfs.core.windows.net/fsn2p/jsoncolumn.csv',
FORMAT = 'CSV',
PARSER_VERSION = '2.0',
firstrow=3
) AS [result])
with below statement checking the value is assigned properly to variable.
select @json as JSON
Using CROSS APPLY for converting Json to table format:
SELECT b.id as ID ,b.name as Name
FROM
OPENJSON(@json)
WITH
(
friends NVARCHAR(MAX) AS JSON
) AS a
CROSS APPLY
OPENJSON(a.friends)
WITH
(
id INT,
name VARCHAR(MAX)
) AS b;
Execution:

Output:
