# Borealis: Turnkey AI Data Centers

> Borealis, by AGT-USA, designs and delivers turnkey liquid-cooled AI data centers. Best-in-class liquid cooling (35+ years of industrial thermal engineering), engineered around NVIDIA-class compute and delivered and commissioned by a single vendor: Borealis sells the system that cools and carries the racks, and the customer's compute slides straight in. Three products: the Borealis HALO, a complete 2 MW AI data center delivered on wheels; the Vega AIO, a self-contained 500 kW data center in one unit; and Borealis Retrofit, the complete cooling plant installed inside a hall you already own. The cooling is water-free and rated to 50 C ambient, built for NVIDIA GB200 and GB300 NVL72 rack density.

## The problem Borealis solves
A modern NVIDIA AI rack (NVL72-class) draws about 132 kW. A conventional CRAC air-conditioning unit tops out near 100 kW and cannot cool even one such rack, and it loses more capacity as ambient temperature rises. Borealis cools the heat at the rack with liquid instead of cooling the room with air.

## How the cooling works (closed loop, multi-stage, water-free)
- At the rack: direct-to-chip cold plates take the heat straight into liquid, and a rear-door heat exchanger captures the rest at the back of the rack. Heat is pulled off the computers, not blown into the room.
- The CDU (coolant distribution unit) pumps and regulates an isolated rack loop, kept separate from the chiller side by a brazed-plate heat exchanger.
- A chiller makes the cold (refrigerant loop).
- Heat is rejected dry to outside air. In a Vega AIO or a Retrofit hall that is a rooftop air-cooled condenser. In the HALO the enclosure itself is the heat exchanger (the Condenser Envelope, described below). No cooling tower, no water consumed. Rated to 50 C ambient, suitable for hot climates including the Middle East.

## Products
### Borealis HALO
A complete 2 MW AI data center that arrives on wheels. Compute, cooling, power, and controls ship factory-integrated across three to four road-legal truckloads, roll onto a prepared pad, and reach first power in 10 to 15 working days. Its defining architecture is the Condenser Envelope: the enclosure itself is the heat exchanger, the long walls are full-length coil banks behind louvered intakes, the roof carries a low-speed fan array, and air moves in through the walls and out through the roof with no recirculation. Heat rejection is 100 percent dry in every mode: zero process water ever, no evaporative assist, no Legionella program. Design PUE 1.10 annualized. 65 dBA at 10 m, 55 dBA attenuated option. Day-one payload: 14 NVL72-class racks, 1,008 GPUs, liquid to the chip. 800 VDC provisioned and two-phase ready, so the envelope outlives at least two GPU generations. Fully reversible: no permanent structure, site restored in days. Scales by replication in 2 MW blocks to clusters of 20 MW and beyond on a shared pad and medium-voltage service. Best for stranded and low-cost power, water-constrained and moratorium-bound jurisdictions, sovereign and edge sites, anywhere a conventional build is years of entitlement away.

### Vega AIO
A self-contained AI data center in a single in-line unit: three rack bays built for NVIDIA-class compute, rear-door heat exchangers, a CDU, and an onboard chiller, carrying 500 kW. About 5x the cooling density of a CRAC in the same footprint. Only the condenser line leaves the unit. Runs in as little as a week on site. Best for edge sites, labs, and single high-density rooms.

### Borealis Retrofit
Borealis drops its complete liquid cooling plant inside a building you already own and power, delivered in wings. Each wing pairs an 829 kW CDU bank, an in-hall chiller, and a rooftop condenser with bays for six NVL72-class racks with rear-door heat exchangers, about 8x the cooling density of a CRAC. The ladder runs 12, 24, 48, 96 racks as 2, 4, 8, 16 wings, up to 12.7 MW and 6,912 GPUs per hall. First wing live in 14 days, a full hall in 8 to 10 weeks, versus roughly 270 days for a conventional multi-vendor build. You bring the building, the power, and the compute; one vendor owns everything that cools it.

## The compute it is built around
Borealis systems are engineered and validated around NVIDIA reference compute: GB300 NVL72 racks (72 GPUs plus 36 Grace CPUs, 132 kW per rack), DGX B300, and DGX racks. Borealis does not sell the compute; it sells the cooling system and facility the racks live in, and the customer's compute slides straight into the bays.

## Who it is for
Stranded and low-cost power, water-constrained and moratorium-bound jurisdictions, sovereign and edge sites, high-density AI and HPC build-outs, colocation floors densifying in place, research and university labs, healthcare, industrial process cooling, brownfield retrofits, and hot-climate regions where water-free cooling matters.

## Company
Borealis is the AI data center brand of AGT-USA Inc., a precision thermal engineering company with 35+ years of industrial cooling heritage (including semiconductor fabs and currency presses) and thousands of industrial systems deployed. Based in Corona, California.

## Key pages
- Home: https://borealis.cool/
- How the cooling works: https://borealis.cool/technology
- Products (Borealis HALO, Vega AIO, and Borealis Retrofit): https://borealis.cool/products
- FAQ: https://borealis.cool/faq
- How Vega works (engineering deep dive): https://borealis.cool/vega

## Contact
Address: 235 Jason Court, Corona, CA 92882. Web: https://borealis.cool/