GB300 DGX Station: What 748GB of Coherent Memory Buys

GB300 DGX Station: What 748GB of Coherent Memory Actually Buys | GPU Partner Skip to content Build a systemReach a human

Article· Aug 7, 2026 GB300 DGX Station: What 748GB of Coherent Memory Actually Buys The GB300 Grace Blackwell Ultra Desktop Superchip puts 748GB of coherent memory on a desk. Which workloads earn it back, and which ones don't.

The NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip puts 748GB of coherent memory under a desk. No rack, no cloud contract, one 1,600 W wall circuit. That memory number is the reason anyone buys one. We're quoting GB300 DGX Station configurations now. NVIDIA showed the Station at GTC 2025 and it took most of a year to reach desks. Dell shipped first in March 2026, ASUS followed in June, and the OEM list has been filling in since. The machine is expensive and it has a narrow job. Worth knowing which workloads pay it back before the PO goes out. What's actually in the GB300 Grace Blackwell Ultra Desktop Superchip One package, two dies. A 72-core Grace CPU on Arm Neoverse V2 cores, a Blackwell Ultra GPU, and NVLink-C2C running 900 GB/s between them. Because both sides sit in one coherent memory domain, either processor addresses the whole 748GB without you writing a copy. No staging buffer, no host-to-device hop in the hot path. The pool is split by type, not by owner. 252GB of HBM3e at 7.1 TB/s holds weights and activations. 496GB of LPDDR5X at 396 GB/s holds everything else. Spec GB300 DGX Station GPU 1x NVIDIA Blackwell Ultra CPU 1x Grace, 72-core Arm Neoverse V2 GPU memory 252GB HBM3e, 7.1 TB/s CPU memory 496GB LPDDR5X, 396 GB/s Total coherent memory 748GB NVLink-C2C 900 GB/s MIG instances Networking ConnectX-8 SuperNIC, up to 800 Gb/s Ethernet ports 2x QSFP112 400GbE, 1x RJ45 10GbE, 1x RJ45 1GbE for BMC PCIe slots 1x Gen 5 x16, 2x Gen 5 x16 at x8 electrical Storage 4x M.2 Gen 5 Video out Mini DisplayPort, system management only Total system power 1,600 W OS Ubuntu with NVIDIA AI Developer Tools

Two corrections worth making, because the spec gets repeated wrong. Those are not three full x16 slots. One is x16 electrical and two are x16-length wired at x8. And NVIDIA supports up to one additional RTX PRO Blackwell GPU alongside the Superchip, not a slot full of them. The precision ladder, and the one number nobody prints GB300 is a 4-bit machine that can do other things. NVFP4 is where the silicon lives, and the ladder falls off exactly as you'd expect until you get to the bottom, where it falls off a cliff. Precision Throughput FP4 Tensor Core 20 PFLOPS FP8 / FP6 Tensor Core 10 PFLOPS FP16 / BF16 Tensor Core 5 PFLOPS TF32 Tensor Core 2.5 PFLOPS INT8 Tensor Core 330 TOPS FP32 80 TFLOPS FP64 and FP64 Tensor Core 1.3 TFLOPS

FP64 at 1.3 TFLOPS is the line to read twice if you were thinking about classical HPC. A GB300 is not a double-precision machine and was never trying to be. Computational chemistry, CFD, FEA and anything else that lives in FP64 should not go anywhere near this box. NVIDIA publishes the figure plainly. Nobody quotes it. How much model fits in 748GB of coherent memory The ability to stop quantizing, mostly. An RTX 5090 gives you 32GB of VRAM. Four RTX PRO 6000 Blackwell cards get you 384GB. That second number is where the comparison usually goes wrong. 384GB is four separate 96GB pools, your framework has to shard a model across them, and every shard boundary costs you something. The GB300's 748GB is one address space. Different problem. Llama 3.1 405B in full BF16 wants roughly 810GB. With the coherent pool and some optimization, that class of model is on the table, and NVIDIA rates the platform to a trillion parameters at FP4. Serving is unremarkable in the good way: point vLLM at it, expose an OpenAI-compatible endpoint on your own network, and the tooling your team already wrote keeps working. Mixture-of-experts models suit this box especially well. MoE trades a large resident parameter count for a small active one, and resident parameter count is precisely what 748GB is for. GB300 vs GB200, DGX Spark, and a quad RTX PRO 6000 workstation There is no GB200 desktop part, so “GB300 vs GB200” in practice means desk versus rack. If you want GB200-class capacity you are buying rack infrastructure. Option Memory a model can see Honest fit RTX 5090 workstation 32GB VRAM Single-GPU dev, 7B to 13B DGX Spark, GB10 128GB unified One developer, prototyping, learning the stack Quad RTX PRO 6000 Blackwell 384GB across 4 separate pools Strong all-rounder, real display output, sharding required DGX Station, GB300 748GB coherent Frontier-scale models on one private node GB300 NVL72 Rack scale, 72 Blackwell Ultra GPUs Production serving at enterprise scale

GB300 NVL72 vs DGX Station is the easy call of the group. The NVL72 is a rack, with rack power, rack cooling, rack lead time and a services engagement attached. The Station plugs into a wall. It's what you deploy while the rack quote is still in legal, or instead of a rack that was always more than you...

GB300 DGX Station: What 748GB of Coherent Memory Buys

Related Articles

(no title)

Scientists reverse brain aging, with a nasal spray

NHS apologises and admits Palantir have access to identifiable patient data

AI has torched the market for junior programmers

Al Vigier: Canada's AI strategy shouldn't include secret Palantir bills