Skip to content
Arete Heat Cloud

Heat-recovery compute · London

The GPU works.The heat stays useful.

Arete Heat Cloud installs managed compute where a hotel already heats water and a pool — and inside one compartment of a parcel locker that is already standing there.

Hotels and poolsLocker nodes
Hotel plant room with hot-water cylinders and a compact forest-green compute cabinet
Plant room node01
Hotel pool at dusk with steam rising from the water
Pool loop
A row of street parcel lockers, one bay marked by a small amber lamp
Locker bay
Domestic hot water preheatPool circuit preheatPlant-room GPU nodeOne locker bayDistributed AI serverDouble-wall heat exchangerParcels stay thermally isolatedZero construction cost for the hostServer rental carries the nodeDomestic hot water preheatPool circuit preheatPlant-room GPU nodeOne locker bayDistributed AI serverDouble-wall heat exchangerParcels stay thermally isolatedZero construction cost for the hostServer rental carries the node

01 · Hotels

Hot water and the pool, from the same node.

A cabinet in the plant room. GPU heat moves through a heat exchanger into the preheat loop — guest cylinders, and the pool circuit when the survey says it can.

See the hotel pilot

02 · Lockers

The locker stays. One box becomes the server.

Parcel lockers are already on the pavement and in buildings. We take a single compartment and run it as a distributed AI server. The other doors keep taking parcels.

See the locker node

Principle · heat

Where the heat is made

A GPU is a heater that also computes. Electrical energy in, heat out, then a preheat loop. One kilowatt-hour at the chip is about one kilowatt-hour of heat. The meter, not the diagram, says what crossed the exchanger.

01

Power in

A metered circuit feeds the cabinet. What the meter records is what the GPU is allowed to draw.

02

GPU

The chip does the compute. Nearly all of that electricity leaves the silicon as heat. That is the source.

03

Cold plate

Liquid lifts the heat off the package. This loop stays on the IT side of the cabinet.

04

Exchanger

A double wall. Coolant and building water meet as temperatures, not as fluids.

05

Water and pool

Preheat for guest hot water, and the pool circuit where the plant can take it. Boilers still finish the setpoint.

Principle · compute

One training run, many rooms

A distributed AI server is not a new hall. A training job is split across nodes that already have an address — a hotel plant room, a locker bay — and the pieces come back as one run. Inference uses the same nodes between training jobs.

Training job

One run, sharded

Scheduler sends a shard out. Gradients and checkpoints return.

H1

Hotel plant

Shard A

L1

Street locker

Shard B

H2

Leisure hotel

Shard C

L1

Building locker

Shard D

Return path — the run stays one job. Parcel data, guest Wi-Fi, and payment networks are not on this path.

Revenue model

Zero cost to build. Server rental covers the expenditure.

The host does not fund construction. Arete builds the node, then rents the GPU. That rental is the income that carries the hardware, the power, and the service. The hotel receives heat and a report. It is not sold a share of the server.

  1. 01

    Host pays £0 to build

    No construction cost for the hotel or the locker operator. They provide the place: a plant room, or one bay.

  2. 02

    Arete funds the node

    Cabinet, GPU, exchanger or sealed bay, meter, and the connection. The hardware stays ours.

  3. 03

    The server is rented

    Training runs and inference are sold as managed GPU time. That rental is the income.

  4. 04

    Rent covers the cost

    Hardware, electricity, and service are carried by the server rental — not by a bill to the host for the build.

The site agreement sets the meter and the term. This is the structure — host build at no construction cost, server rental as the income — not a forecast for a building we have not surveyed.

Two businesses, one rule

Heat is why a hotel opens the door. Compute is what we sell.

Hotel plant room

H1-Lite

Smaller independents. First measured pilot where headroom is limited.

Draw
0.5–0.7 kW
Heat
12–17 kWh/day

Hotel plant room

H1

The main hotel node. Power-capped professional GPU. DHW, and a pool where the circuit allows.

Draw
0.8–1.0 kW
Heat
19–24 kWh/day

Hotel or leisure plant

H2

Two GPUs, liquid-cooled. Larger hotels, hostels, and sites with a pool loop.

Draw
1.4–1.8 kW
Heat
34–43 kWh/day

Parcel locker bay

L1

One compartment. Sealed compute module. Neighbouring doors stay parcel doors.

Draw
0.6–0.9 kW
Heat
Rejected, not into parcels

Hotel figures are typical ranges for a power-capped professional GPU, not a forecast for a named site.

What this is not

A measured install. A removal option. No theatre.

Arete sells compute. A hotel receives metered heat and a report. A locker operator keeps a locker, with one bay reserved under a site agreement.

  • Not the hotel’s only hot water

    Boilers or heat pumps still finish the temperature. Guest hot water does not depend on the GPU.

  • Not on guest, PMS, or payment networks

    The node uses a segregated connection. It does not sit on the hotel’s guest Wi-Fi.

  • Not a mining box

    Arete sells managed compute — inference, rendering, batch. The hotel is not an investor in the hardware.

  • Not a heat source for the parcels

    In a locker, the compute bay is thermally isolated. Waste heat is rejected away from the other doors.

Tell us which door you are.

Hotel or poolA plant-room survey for hot water, a pool, or both.Locker networkOne bay in cabinets that are already installed.Compute buyerUK-sited GPU capacity. Inference, render, batch.