LAS VEGAS — Startup ODINN has unpacked OMNIA, an artificial intelligence (AI) supercomputer the size of carry-on luggage that delivers data center-level performance without requiring months of construction or cloud-based infrastructure.
The new system, announced at the massive CES show here, targets institutions with strict data sovereignty requirements — defense agencies, government bodies, healthcare facilities, and financial institutions — that can’t risk sending sensitive information to remote cloud data centers. Until now, these organizations had only one option: Building costly, localized data centers from scratch.
OMNIA packs full data center specifications such as CPUs, GPUs, multi-terabyte memory, and petabyte storage into a compact unit deployable in minutes. Its proprietary closed-loop cooling system runs quietly while handling advanced AI workloads.
The system connects to standard power and networking infrastructure, requiring no specialized server rooms.
For larger processing demands, ODINN offers the Infinity Cube, which combines multiple OMNIA units in a modular glass enclosure. Each unit maintains independent cooling and computing capabilities without external infrastructure like cooling plants or raised floors.
The NeuroEdge software layer integrates with NVIDIA Corp.’s AI ecosystem and other frameworks, handling job scheduling and deployment to optimize cluster performance. This allows institutions to focus on building AI models rather than managing infrastructure.
Founded in 2023, ODINN set out to eliminate dependency on traditional data centers by rethinking how power, cooling, and reliability could be integrated into a single deployable system. The company positions itself as an AI infrastructure provider delivering low-latency, high-performance capabilities for security-conscious organizations.
“ODINN’s OMNIA reframes where AI infrastructure can live. OMNIA compresses data center class capability into a deployable, on-premises system that can be stood up immediately,” said Mitch Ashley, vice president and practice lead, DevOps and AppDev, at The Futurum Group.
“That matters most where deployment speed, flexibility, and sovereignty are non-negotiable optimizations. For defense, government, healthcare, and financial institutions, this points to a broader shift in AI deployment models,” Ashley added. “As AI workloads move closer to sensitive data and operational edges, infrastructure must adapt to policy, risk, and mission constraints rather than the other way around. Systems like OMNIA show how AI compute is becoming situational, portable, and purpose-built, enabling advanced AI use without surrendering control or waiting for traditional infrastructure to catch up.”
