Announced at CES 2026, full production as of Q1 2026, availability 2H 2026.
A suite of 6 platform products: Rubin GPU, Vera CPU, NVLink 6 Switch, ConnectX-9, BlueField-4, and Spectrum-6.
NVIDIA have been focusing more on system & rack level design. They hope to make racks a unit of compute, with the hardware and software in extreme co-design.
Chip
What it is
Why it exists
Rubin GPU
Flagship datacenter GPU (dual-die, HBM4)
The core compute engine; ~5x Blackwell inference
Groq 3 LPU
Language Processing Unit. SRAM-based inference chip (from NVIDIA’s Dec 2025 Groq acquisition)
500MB on-chip SRAM @ 150 TB/s per chip. Fights the HBM memory wall for decode. The 7th chip of the platform
Vera CPU
88-core custom Arm CPU
Power-efficient host CPU, tightly coupled to GPUs via NVLink-C2C (chip-to-chip interconnect)
NVLink 6 Switch
Scale-up interconnect
Makes 72 GPUs in a rack act as one memory/compute domain
ConnectX-9 SuperNIC
Scale-out Network Interface Card
Node-to-node networking across racks
BlueField-4 DPU
Data processing unit
Offloads networking/storage/security from CPUs; software-defined infra
Spectrum-6 Ethernet Switch
Scale-out Ethernet (co-packaged optics)
Connects racks into pods/clusters efficiently
Board
What it is
Why it exists
HGX Rubin NVL8
8-GPU baseboard with NVLink
For orgs that want Rubin GPUs but their own servers/CPUs — pairs with x86 (Intel/AMD) or Vera. The OEM building block
Vera Rubin NVL4
4 GPUs + 2 Vera CPUs module
Smaller HPC/scientific-computing form factor for MGX servers
Rack
What it is
Why it exists
Vera Rubin NVL72
Flagship rack: 72 GPUs + 36 Vera CPUs, one NVLink domain
Max scale-up — whole rack runs as a single coherent AI engine, no model partitioning. Fully liquid-cooled, cable-free modular trays
Groq 3 LPX
256 Groq 3 LPUs in a rack; deployed alongside NVL72
Decode is bandwidth-bound — LPX offloads token generation (FFN/MoE) to SRAM while Rubin GPUs handle prefill + attention. ~35x throughput/MW claim. Supersedes the earlier Rubin CPX concept
Vera CPU rack
256 Vera CPUs, no GPUs
CPU capacity for RL and agentic AI — sandboxes, tool calls, evals, orchestration (~22,500 concurrent sandboxes)
TurnKey
What it is
Why it exists
DGX Vera Rubin NVL72
The NVL72 rack as a full NVIDIA appliance
Hardware + Mission Control software + support, zero integration work