The business case seems to be to finally kill Falcor [1] which had a lot of similarities to GraphQL but a much smaller maintenance and developer community than GraphQL and I would assume looked a lot like tech debt to Netflix at this point.

[1] https://github.com/Netflix/falcor