en.Wedoany.com Reported - US-based distributed AI infrastructure company Zero Latency has adopted the AI Factory platform, co-designed by Red Hat and NVIDIA, as the enterprise-grade Kubernetes foundation for its nationwide network. The company, formerly known as Hyphastructure, recently launched its Zerogrid platform in closed beta. This platform acts as an AI inference orchestration layer, routing workloads to edge infrastructure based on latency, location, and capacity constraints.

Adopting the Red Hat AI Factory platform provides Zero Latency with a containerized foundation layer, enabling it to manage Graphics Processing Unit (GPU) resources distributed across different locations within a unified workflow. Joe Fernandes, Vice President and General Manager of Red Hat's AI Business Unit, noted: "By using Red Hat AI Enterprise Edition to manage distributed infrastructure, Zero Latency demonstrates how hybrid cloud technology can scale innovation without investing massive resources. We are collaborating with Zero Latency to jointly define the architecture for future low-latency distributed AI inference applications."

While distributed computing is not a new concept, it is gaining increasing attention as the demand for computation closer to the actual application deployment locations grows. Compared to the centralized monolithic infrastructure used by hyperscale clouds and new cloud providers, the Zero Latency team drew inspiration from virtual power plants to build its distributed platform, aggregating resources into what it calls a shared inference capacity pool. The company claims this enables democratized access to NVIDIA-grade GPUs, allowing users to drive long-context or autonomous AI applications while meeting latency or sovereignty-related requirements.

This distributed AI inference platform employs "prefix-resident indexing," enabling inference cache data to be distributed across GPU memory, system memory, and storage tiers within the distributed cluster. The company believes this approach addresses bottlenecks caused by large KV cache memory states spanning multiple systems and storage tiers. Michael Huerta, CEO of Zero Latency, stated: "For years we have believed that decentralized infrastructure is superior to centralized infrastructure for the workloads that need it most. AI inference is its next application domain: machine-driven, constraint-bound, and poorly served by centralized cloud services. Red Hat AI Enterprise Edition provides us with the containerized foundation to bring this architecture to enterprise customers, from factory floors to city streets."

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com