Run real-time and async inference on the same infrastructure with GKE Inference Gateway

As AI workloads transition from experimental prototypes to production-grade services, the infrastructure supporting them faces a growing utilization gap. Enterprises today typically face a binary choice: build for high-concurrency, low-latency real-time requests, or optimize for high-throughput, "async" processing. In Kubernetes environments, these requirements are traditionally handled by separat