Liqid delivers software-defined composable infrastructure that enables enterprises to dynamically allocate GPUs and CXL-based memory, unlocking higher performance, agility, and efficiency.
Liqid delivers software-defined composable infrastructure that enables enterprises to dynamically allocate GPUs and CXL-based memory, unlocking higher performance, agility, and efficiency.
Using Liqid, organizations pool, share, and attach GPUs or DRAM on demand across PCIe and CXL fabrics, enabling up to 30 GPUs or 100TB of memory per pod with no performance loss.
Liqid architecture solves power, budget, and scalability constraints with truly dynamic AI infrastructure that supports 600W GPUs, integrates with Kubernetes, VMware, Nutanix, and Slurm, and maximizes utilization while reducing TCO.
Liqid’s composable GPU platform dynamically pools and assigns GPUs to any server, enabling higher utilization, 30-GPU scale-up, lower TCO, and maximum performance for AI workloads across PCIe fabrics.
Liqid’s composable CXL memory dynamically expands and shares DRAM across servers, delivering massive capacity, low latency, higher utilization, and dramatic performance and TCO gains for AI, analytics, and database workloads.
Liqid Matrix software orchestrates dynamic GPU and CXL memory pooling, enabling automated provisioning, multi-fabric management, and seamless integration with Kubernetes, VMware, Nutanix, and Slurm for fully agile, efficient AI infrastructure.
