The correct answers are C and D. Option C is correct because Snowpipe uses serverless compute, meaning Snowflake automatically manages the compute resources required to execute the COPY INTO commands, eliminating the need for manual virtual warehouse management. Option D is correct as Snowpipe maintains metadata to track which files have been successfully loaded, preventing duplicate loads. Option A is incorrect because Snowpipe via REST API supports both internal and external stages (e.g., S3, Azure Blob Storage). Option B is incorrect because not all COPY INTO options are available during pipe creation; specific options like FILES, ON_ERROR=ABORT_STATEMENT, SIZE_LIMIT, PURGE, and FORCE are excluded, as noted in Snowflake documentation and the community discussion.