Concept of a dark 20-foot container on a concrete pad beside an industrial building at dusk. Not a photo of a built site.

Building a distributed datacenter for the future of AI.

Big datacenters are waiting years for permits and power. We containerize GPU capacity to run on excess power at thousands of small industrial sites around the world.

Join the waitlist Why a small site

What it is

Your property. Your power. Our box. Our GPU.

TWR sets a self-contained 20 ft container on a parking pad. Inside are one or two NVIDIA GB300 NVL72 racks, 72 GPUs each. Power and cooling stay in the box. No water needed. The building stays a building. You lease the pad and the power. We operate the racks.

What the property needs is spare 480 volt three-phase power, 250 amps for one rack or 500 amps for two, a concrete pad, and a business fiber line of about 10 gigabits per second. A licensed electrician makes the connection. You do not have to run the computers, and the site does not supply water.

20 ftSelf-contained container on a concrete pad beside the building.
72 GPUsEach NVIDIA GB300 NVL72 rack. One or two racks in the box.
480 V, 250 AThree-phase spare power. 250 amps for one rack, 500 for two.
10 GbpsBusiness fiber to the pad.

Why

GPU capacity is waiting on sites that take years.

New hyperscale power, substations, and generation take years, so GPU capacity is stuck waiting on big sites.

A 20 ft self-contained container on an industrial parking pad is the way through. Each box holds one or two NVIDIA GB300 NVL72 racks. The property brings the pad, 480 V, and fiber. There is no site water. Many of those small sites run as one distributed compute cloud.

Time to get it

  • New transmission 5–15 yr
  • Grid queue to operation 5+ yr
  • Substation transformer 3+ yr

Demand vs the grid, 2025

  • Data-center electricity +17%
  • All electricity +3%

Published ranges, not a TWR forecast. New transmission is 5–15 years (IEA). US projects that reached operation in 2025 spent a median of over 5 years in the interconnection queue (Berkeley Lab). Substation transformers are quoted past 160 weeks (Wood Mackenzie, 2026). Data-center electricity use rose 17% in 2025, against 3% for global electricity (IEA). About 20% of planned data-center projects are at risk of delay if grid bottlenecks stay (IEA).

Hardware

NVIDIA GB300 NVL72

  • 72 Blackwell Ultra GPUs
  • 20 TB GPU memory
  • 1,440 petaflops FP4
  • 130 TB/s NVLink

Per rack, from NVIDIA. The FP4 figure includes sparsity. Dense FP4 is 1,080 petaflops.

NVIDIA GB300 NVL72 rack, front view beside a view of the cable side.

Who it’s for

Join the waitlist.

  • Property owner

    You have industrial property with spare power and room for a pad. If the power is 480 volt three-phase with at least 250 amps to spare and fiber can reach the site, you lease the pad. TWR brings the box and runs it.

  • Investor

    Be the first to know when a site is ready to talk about. Joining the list is interest only, not an offer.

  • GPU reservation

    Capacity can be reserved for 1, 2, 3, or 4 years. A longer term has a lower hourly rate and a different share paid up front, and exact terms are shared with people on the waitlist.

Concept of the same container on a pad beside an older brick industrial building. Not a photo of a built site.

Waitlist

Property, investment, or a GPU reservation. We’ll follow up. Nothing here books a pad, takes capital, or reserves a rack.

Not a lease, an investment, or a signed reservation.