FriendliAI Showcases AI Inference Optimization and Cost Savings at GTC

According to a recent LinkedIn post from FriendliAI, the company is using its presence at NVIDIA’s GTC conference to showcase tools for optimizing AI inference on open-source and open‑weight models. The post highlights technical sessions focused on continuous batching, online quantization, high-speed benchmarking with the Friendli Suite, and containerized inference on AWS EKS.

The content suggests FriendliAI is positioning itself as an infrastructure partner for enterprises looking to reduce GPU costs while increasing throughput for AI workloads. For investors, the emphasis on 2x–3x throughput and 50%–90% GPU cost reductions, if realized at scale, could strengthen the company’s value proposition in the competitive AI infrastructure market and support customer acquisition in cost-sensitive enterprise deployments.