| Anz. | Price |
|---|---|
| 3-4 | $37,830 |
| 5-9 | $37,050 |
| 10-29 | $36,660 |
| 30+ | $36,270 |
Your models don't get deprecated, rate-limited, or pulled. The weights sit on your disk.
Prompts never leave the building. Nobody trains on your work. Not by policy, by physics.
Own the machine and agents can run around the clock. You pay for electricity, not tokens.
Fine-tune on what only you know. The model becomes an asset you own, and it sells with the company.
Four cards, tuned and burn-tested. We install RTX 5090s (32GB each) or RTX PRO 6000 Blackwell (96GB each) — installed, driver-loaded, and benchmarked as one system before it ships. Plug it in and run.
Milled from solid aluminum. A solid billet, milled inside and out until only the machine is left. It moves heat like a heatsink and holds true under a hundred pounds of hardware. 15.5 × 15.5 × 16 inch, anodized silver.
A full ×16 to every card. Each GPU reaches the board over its own server-grade PCIe 5.0 riser — shielded, custom-cut, no dropped lanes. Every power run is cut to length and sleeved by hand.
ASUS Pro WS W790E-SAGE SE. The hardest choice in a multi-GPU build, solved with a workstation board: every card gets a full ×16, and the BIOS ships pre-tuned so every card links at full width.
Two supplies, one clean bus. Four cards at full draw pull close to 3000W — cheap power is where multi-GPU builds die. We spec two 2000W supplies with native 12V-2x6 feeding one distribution board. Plan a 240V circuit before it arrives — normal at this scale, better sorted early.
Air moved on purpose. Eighteen fans on a single controller, mapped to the airflow path before a panel was cut — not bolted on after.
ECC memory, fast NVMe. Six 32GB DDR5-4800 ECC RDIMMs catch the silent bit-flip before it corrupts a run a whole team depends on. The 2TB PCIe 4.0 NVMe loads a 70B checkpoint in seconds, not minutes.
.webp)
Open-source hardware
Every CAD file, bill of materials, and BIOS setting — free on GitHub. Fork it, change it, build your own, even sell it. This is how the personal computer began: in the open. We’re doing it again, for AI hardware.
GitHub 1.1k2x-5090/ 4x-5090/ 8x-5090/ 4x-6000/
bom/ step_models/ stl-models/ photos/
README.md setup.md
MIT · 143 forksClone it. Build it. Sell it. We just want it built.
The lineup
