Own US-based AI Infrastructure

Own the hardware outright. We build it and host it in Oregon. Put a rack to work under our offtake contracts and keep 85% of what it earns, or keep any machine entirely to yourself for a flat monthly fee.

Own US-built AI hardware

NVIDIA H200 NVL

The frontier machine. DeepSeek-V3 and Llama 405B on a full 8-GPU node.

VRAM: 141 GB HBM3e · Bandwidth: 4.8 TB/s · NVLink · up to 8× · 4U rack

Own it today · Starting at $59,136 earns about $1,259/mo · Configure & buy →

NVIDIA RTX PRO 6000

The 70B workhorse: a whole Llama 70B on one 96 GB card

VRAM: 96 GB GDDR7 · Bandwidth: 1.8 TB/s · Blackwell · up to 4× · Workstation tower / 4U rack

Own it today · Starting at $34,766 earns about $574/mo · Configure & buy →

NVIDIA RTX 5090

The entry box: 8B–32B models, render + AI

VRAM: 32 GB GDDR7 · Bandwidth: 1.8 TB/s · Blackwell · up to 2× · Prosumer tower / Workstation tower

Own it today · Starting at $7,880 · Configure & buy →

How it works

You configure and buy it

You pick the GPU, the memory, the storage, and whether it racks with us or ships to you. One charge and the machine is yours, with nothing owed to us afterwards. On select machines you can start with 30% down instead, arranged on a call.

We build and QA it

Parts land in Eugene, Oregon and your machine is assembled by hand, stress-tested for 48 hours, and racked. Typical lead time is 6–8 weeks from order.

You choose how it runs

Earning puts it to work under our offtake contracts; private keeps it to your workloads alone. You choose when you buy, and can change it by telling us. For the hours you make it available, we bring the demand and handle routing, billing and the SLA.

You see what it earned

Your dashboard shows what the machine earned, our 15%, Managed Hosting and what is owed to you, paid monthly. Want the machine on your own floor instead? Tell us and we'll ship it.

How ownership works · Earning Calculator · Earnings