Federal and defense organizations need to deploy AI, HPC, secure VDI, simulation, and other data-intensive workloads inside constrained, secure, and power-limited environments. Liqid turns fixed infrastructure into a dynamic software-defined pool of memory and GPUs, helping agencies improve utilization, increase performance, and deploy mission workloads faster without overprovisioning scarce resources.
Request a Federal AI Infrastructure DemoTalk to a Federal Infrastructure Specialist
Federal and defense workloadsare becoming more GPU-intensive, memory-intensive, and latency-sensitive. AIinference, ISR analytics, simulation, secure engineering VDI, cyber analytics,and scientific workloads need high-performance infrastructure that can adapt tochanging mission demand. Traditional servers lock expensive GPUs and memoryinto fixed configurations, creating stranded capacity, overprovisioning, andslower deployment cycles.
Defense, lab, and edge environments often need more performance without more racks or more power.
Expensive accelerators sit locked to individual servers and workloads, leaving valuable capacity idle when mission demand shifts.
Long-context models, graph workloads and KV cache demands push beyond standard server memory limits.
Liqid disaggregates GPUs and memory from fixed servers, pools them at rack scale, and dynamically assigns them to workloads through Liqid Matrix software. Existing or new servers can draw from shared GPU and memory pools, allowing teams to configure capacity in seconds, improve utilization, and scale resources independently.
A confidential DoD industrialbase environment needed high-performance engineering VDI with strong GPUacceleration inside severe power and cooling constraints. Liqid SmartStackenabled maximum GPU density per server while optimizing the CPU-to-GPU ratiodynamically.
Confidential DOD customer was buying 8 Way servers but could only fit one per rack (8-way is 1.5 kVa, rack was limited to17.5 kVa). Liqid designed a system with the customer to take full advantage ofthe available power - 4 2U servers with 28 NVIDIA H200s under the same power ceiling. This is the equivalent of 3 8-way and 6 4-way servers.
Liqid works with selected enterprise and public-sector sales partners to help Federal and Defense customers evaluate, procure, and deploy GPU and memory pooling infrastructure. Talk to Liqid to identify the right partner route for your agency, program, or mission environment.
Our partner ecosystem includes Carahsoft, CTG Federal, ClearShark, StorageHawk, Dell Technologies
Cisco and Liqid collaborate to deliver unprecedented GPU density and agility within 600W devices tailored for AI, HPC and edge inference.
Download
Cisco and Liqid deliver software-defined composable infrastructure that unifies GPU pooling and sharing across Cisco UCS rack servers.
Download
LIQID enables enterprises to dynamically pool and scale memory across servers, eliminating bottlenecks while improving throughput, efficiency, and overall database performance.
Download
Hardware that has been tested and verified to work with Liqid Command Center™. Keep checking back for updates to this list.
Download
See how Liqid can help your teampool GPU and memory resources, increase utilization, and deploy secure on-premAI, HPC, VDI, and other data-intensive workloads with the infrastructure ecosystem you already trust.
Request a Demo
LIQID’s innovative software-defined composable infrastructure platform enables users to manage, scale out, and configure physical bare-metal server systems in seconds.
