Catalytic AITroy, New York
REFERENCE DESIGNSPd · Pt · Rh

Machines we will build you.

Three starting points, from an entry fine-tuning node to a dual-GPU training rig. None of them is a fixed product — we adjust the spec to whatever you actually plan to run.

They are named for the three metals in a catalytic converter: palladium, platinum and rhodium, in ascending order of how rare and expensive they are. It seemed more useful than a number.

Why own one

Two reasons,
and only one is money.

The obvious one. If you would keep a node busy most of the month, owning beats renting inside a year, and after that the marginal hour costs you electricity. We will do that arithmetic with you honestly, including the case where it says rent.

The one that actually decides it. Some work cannot go anywhere else. Export-controlled drawings, patient records, fab process data under NDA, privileged documents. For that, a machine you own, sitting on a network you control, is not a cost optimisation — it is the only configuration that is allowed. We build those, document them for an audit, and can hand one over having never connected it to the internet at all.

What sovereign deployment involves →
What you get

Specced, sourced, built, burned in.

A documented machine

The full parts list with part numbers, thermal numbers under sustained load, and a build guide detailed enough that someone else could put it together.

A working software stack

OS, NVIDIA drivers, CUDA, Docker with the container toolkit, and whichever frameworks you actually use — installed, pinned and checked, not left for you.

A burn-in report

We hammer it before it ships and record thermals and clock behaviour, so a flaky part shows up on our bench instead of three weeks into your training run.

Ongoing support

Optional, and worth having. Hardware fails eventually; when it does you will be talking to whoever put the machine together.

Reference designs

Three starting points.

The three reference builds compared by VRAM: PALLADIUM-16 at 16 GB on one RTX 5080, PLATINUM-96 at 96 GB on one RTX PRO 6000, and RHODIUM-192 at 192 GB on two RTX PRO 6000 Max-Q cards. PALLADIUM-16 16 GB 1 × RTX 5080 PLATINUM-96 96 GB 1 × RTX PRO 6000 RHODIUM-192 192 GB 2 × RTX PRO 6000 Max-Q 96 192 GB
Fig. 01 — Reference designs by VRAM
PALLADIUM-16 Entry fine-tuning node
GPU
RTX 5080
VRAM
16 GB
System RAM
64 GB
Load power
~550 W

Best for QLoRA fine-tuning of 7B–14B models and bf16 LoRA up to about 8B, small-batch inference, RAG indexing, and Jupyter-based research. Fits under a desk in a mid-tower, idles around 85 W, and takes about three hours from boxes to first boot.

Bill of materials
GPU1× RTX 5080 — 16 GB GDDR7, Blackwell SM, PCIe 5.0 ×16, 256-bit, 960 GB/s, 360 W TDP
CPUAMD Ryzen 7 7800X3D — 8C/16T, 96 MB L3 V-Cache, 4.2 GHz base / 5.0 GHz boost, Zen 4
Memory64 GB DDR5-6000 CL30 (2×32 GB), EXPO, dual-channel
Storage1× 2 TB PCIe 4.0 NVMe (WD Black SN850X) + 1× 4 TB SATA SSD for datasets
PowerCorsair RM850x — 850 W 80+ Gold, fully modular, ATX 3.1
SoftwareUbuntu 24.04 LTS · CUDA 12.8 · PyTorch 2.7 · vLLM
PLATINUM-96 Professional workstation Most common
GPU
RTX PRO 6000
VRAM
96 GB ECC
System RAM
128 GB
Bandwidth
1,792 GB/s

Best for QLoRA fine-tuning past 100B at 4-bit, full-weight fine-tuning to roughly 8B on one GPU, production inference with vLLM continuous batching, molecular dynamics, diffusion training, and multi-user JupyterHub. 96 GB also holds a 70B model at 4-bit with a long context. ECC memory earns its keep once a run lasts days. This is TROY-1 a generation on — we run the earlier revision ourselves.

Bill of materials
GPU1× RTX PRO 6000 Blackwell Workstation Edition — 96 GB GDDR7 ECC, 1,792 GB/s, 512-bit, PCIe 5.0 ×16, 600 W maximum power
CPUAMD Ryzen 9 9950X — 16C/32T, Zen 5, TSMC 4 nm, 80 MB cache, 4.3 GHz base / 5.7 GHz boost, 170 W TDP
Memory128 GB DDR5-6000 CL36 (2×64 GB), EXPO — two DIMMs rather than four, because four dual-rank sticks will not hold 6000 on AM5
Storage2 TB PCIe 5.0 NVMe (Crucial T700) on the CPU-attached M.2 + 8 TB PCIe 4.0 NVMe for datasets. AM5 does not have the lanes for a second Gen5 drive plus U.2 alongside a ×16 GPU — that configuration needs TRX50.
PowerSeasonic Vertex PX-1200 — 1,200 W 80+ Platinum, ATX 3.1, native 12V-2×6. ~940 W under load, so about 78% of rating.
MotherboardASUS ProArt X870E-Creator WiFi, or an equivalent AM5 board with PCIe 5.0 ×16 plus a Gen5 M.2
Cooling & chassis360 mm AIO on the CPU; full-tower with clearance for a 304 mm card
SoftwareUbuntu 24.04 LTS · CUDA 12.8 · Docker 27.x · k3s · MLflow
RHODIUM-192 Dual-GPU training rig
GPU
2× RTX PRO 6000
VRAM
192 GB ECC
System RAM
256 GB ECC
Interconnect
PCIe 5.0

Best for tensor-parallel serving of 70B–123B models, full-parameter fine-tuning to roughly 10–13B with DeepSpeed ZeRO-3 across two GPUs, LoRA and QLoRA well past 70B, large-scale embedding and GNN training, and multi-tenant inference partitioned with MIG or MPS. The 192 GB is two 96 GB address spaces sharded in software, not a hardware-pooled pool — there is no NVLink on these cards.

Bill of materials
GPU2× RTX PRO 6000 Blackwell Max-Q — 192 GB GDDR7 ECC total, 300 W each, PCIe 5.0 ×16 each with ~64 GB/s peer-to-peer. RTX PRO Blackwell does not support NVLink; the blower-style Max-Q is also the SKU meant for stacking two cards.
CPUAMD Threadripper 7960X — 24C/48T, Zen 4, 152 MB total cache, 4.2 / 5.3 GHz, 350 W TDP, sTR5, 48 PCIe 5.0 lanes
Memory256 GB DDR5-5600 RDIMM ECC (4×64 GB), registered, quad-channel — 5600 is an EXPO profile; AMD's official rating for this CPU is 5200
Storage2× 2 TB PCIe 5.0 NVMe for scratch, plus 2× 7.68 TB U.2 NVMe (Kioxia CM7-V) in RAID-0 for datasets — ~26 GB/s aggregate read across the U.2 pair
PowerSeasonic Prime TX-1600 — 1,600 W 80+ Titanium, single +12V rail, 3× EPS12V and 2× native 12V-2×6. ~1,150 W under load with Max-Q cards.
NetworkingOnboard 10 GbE, and a spare PCIe 5.0 ×16 slot for a future 25/100 GbE NIC. IPMI is available on TRX50 only as an optional expansion card, which costs a slot.
MotherboardASUS Pro WS TRX50-SAGE WIFI — sTR5, quad-channel RDIMM, onboard 10 GbE
Cooling & chassissTR5 air or AIO cooler; full-tower or 4U rackmount with spacing for two dual-slot cards
ElectricalDraws ~10 A at 120 V under load, so it fits a standard 15 A circuit. Specced with 600 W Workstation Edition cards instead, it needs a dedicated 20 A circuit or 208/240 V.
SoftwareUbuntu 24.04 LTS · CUDA 12.8 · NCCL 2.25 · DeepSpeed 0.16 · Slurm 24.05 · Enroot + Pyxis
On price. Part costs move week to week and the right spec depends on what you run, so a published number would be out of date and probably wrong for you. Tell us the workload and what you have to spend and you get an itemised quote — parts at cost, build fee listed separately, nothing hidden in the bill of materials.

Get a build quoted.

Tell us what you plan to run and what you have to spend, and we will come back with a spec and an itemised quote.