Flatkey launches dedicated NVIDIA B300 cloud infrastructure
Flatkey Inc. has introduced Flatkey Compute, a platform providing isolated NVIDIA B300 hardware clusters for artificial intelligence inference workloads.
San Jose-based Flatkey Inc. announced the launch of Flatkey Compute, a service that provides enterprise clients with dedicated hardware clusters utilizing NVIDIA B300 GPUs. Unlike traditional multi-tenant cloud environments, this service reserves entire physical clusters for individual customers.
The service is designed specifically for AI inference, allowing companies to run large-scale models in an isolated environment. Flatkey handles the infrastructure management and operation, positioning the product as a compliant solution for industries that require strict data separation and hardware consistency.
The deployment of dedicated GPU clusters responds to increasing demand for predictable compute performance in AI deployment. By moving away from shared resource models, Flatkey aims to reduce latency and infrastructure variance for clients focused on high-demand machine learning tasks.