Specifications
Model
Gigabyte R1C7-K0A-AS1
Form Factor
1U rackmount
Node Count
40 nodes in one 1U chassis
CPU
40 x Intel Core Ultra 7 258V (Lunar Lake)
Total Cores
320 cores across the chassis
Memory per Node
32 GB LPDDR5X
Total Memory
1.28 TB across the chassis
Storage per Node
2 x M.2 slots, PCIe Gen5 x2
Total SSDs
80 x M.2 SSDs
Integrated GPU
40 integrated GPUs (Lunar Lake Xe-LPG)
Cartridge Design
GPU-like cartridges — 5 sleds of 8 nodes each, 3 front + 2 rear row
Rear Connectivity
2 x QSFP28 ports consolidated behind a large rear heatsink
Internal Fabric
Each cartridge carries two MCIO 8i connectors
Management
Chassis management controller with rear management port
Power
2 x 3.2 kW Titanium-rated power supplies
Shown At
Computex 2026
Density Use Case
Maximum node count per rack unit rather than per-node performance
Supply
B2B volume supply; quote per configuration
Overview
The Gigabyte R1C7-K0A-AS1 is a density play: 40 complete compute nodes, 320 cores, 1.28 TB of memory, 80 M.2 SSDs and 40 integrated GPUs inside a single 1U chassis. It is the opposite of a GPU server — the goal is node count and power efficiency per node, not per-node floating-point throughput.
Each node is an Intel Core Ultra 7 258V, a Lunar Lake part, with 32 GB of LPDDR5X and two PCIe Gen5 x2 M.2 slots. Gigabyte packages eight nodes onto one board and loads five of those cartridges into the chassis — three in the front row, two in the rear. The cartridges look and slide like GPU sleds.
Consolidation happens at the back: each cartridge appears to bring out two MCIO 8i connectors, and a large rear heatsink covers what is likely an internal switch or Ethernet fabric feeding the two QSFP28 rear ports. Gigabyte did not publish the internal topology, so whether those are Ethernet or PCIe is unconfirmed. Power comes from two 3.2 kW Titanium supplies, and a chassis management controller provides a dedicated management port.
Key Benefits
Forty nodes per 1U — the highest node density in this catalogue, at 320 cores and 1.28 TB per chassis. Lunar Lake efficiency with LPDDR5X on-package memory and integrated GPU per node. Two QSFP28 uplinks consolidating forty nodes onto the network. Titanium-rated power at 2 x 3.2 kW for a density-first deployment.
Applications
Scale-out micro-node clusters for web and edge serving; CI/CD and container farms needing many independent nodes; distributed inference where per-node efficiency beats per-node throughput; isolated multi-tenant lab environments; network simulation and test beds; density-constrained colocation where rack space, not compute, is the limit.
Request a Quote — GIGABYTE R1C7-K0A-AS1 40-NODE 1U CLUSTER — 320 CORES, 1.28TB MEMORY
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →