Well theres not much point in just parsing a large json file, there will always be more work which cant be done using SIMD and that work time will be significantly larger than the portion spent parsing.
That depends on the application. Maybe I have 1TB of json files in S3 and I want to compute an aggregate value from one of its fields.