Thank you for the article! Nice work. Wouldn't this preferred approach increase latency by adding an extra hop and magnify the single point of failure (assuming it's not microsite but microsites)? If we take a performance first approach I don't understand why we need GraphQL beyond naive service layers that don't consolidate network requests. Also, SSR in NextJS and React Server Components are failing to provide a clear value proposition for GraphQL, as did you. Maybe it's time to let GraphQL go? Or can it provide a clear value proposition in the face of SSR and SSS?