Hexi Pod 1
Your own AI compute, on the power you already have.
504 GPUs earning on your site in months, not years. A factory-built, liquid-cooled facility designed for your site and delivered against a date backed by our manufacturing partner.
The free check takes about five minutes. You'll hear back within two business days.
The GPUs are ready. The power isn't.
How long until your GPUs run, drawn to scale.
Hexi Pod 1, from order, on power you already have
26 to 40 weeks
Estimate, where suitable electrical plant is already on site.
London colocation vacancy is forecast to fall to 5.5% by the end of 2026 (CBRE). But plenty of factories, warehouses, industrial estates and battery sites already have spare import. Pod 1 puts your compute there.
One pod. Seven racks. Built in a factory, tested before you pay the final 5%.
504
GPUs
Seven NVIDIA GB300 NVL72 racks, which you own, plus three support racks. We provide the space, power, cooling and monitoring.
1.05 MW
Total IT envelope, about 140 kW per GPU rack.
1.5 MVA
Of your existing import, preferably your existing 11 kV plant. No new grid connection to wait for.
Liquid-cooled
Direct-to-chip through isolated loops. N+1 coolant units. Dry coolers sized for UK heatwaves and checked at 40°C.
Factory-built
Power room, IT modules and cooling compound, built and tested in the factory, then craned onto a prepared base.
Acceptance tested
Full-load, leak, cooling-failure and UPS-shutdown tests before handover.
How your pod arrives.
Scroll to follow it from the factory floor to your site.
Built and tested in the factory
Our manufacturing partner builds the power room, IT modules and cooling compound, then runs factory acceptance tests before anything ships.
Delivered by road
The modules travel as transportable units. No years of site construction.
Craned onto your prepared base
A crane sets each module on a base prepared while the pod was being built.
Energised and acceptance tested
Connected to your existing supply, then load-tested on site before you pay the final 5%.
Every question a buyer asks, answered before you commit.
-
Will my site even work?
A free spare-capacity check. Send us the site and its supply; within two business days you'll know whether Pod 1 fits.
Free
-
What will it really cost?
A firm price, backed by our manufacturing partner's quotation.
In the reservation
-
When will it actually run?
A dated programme, backed by a named factory slot.
In the reservation
-
Will it work when it arrives?
Factory and site acceptance tests and a full-load test. You pay the last 5% after site acceptance.
In the order
-
Is it designed properly?
Designed for your site by the Hexi engine, and independently reviewed by a senior data-centre electrical engineer.
Included
-
Planning and permits?
The planning route and every interface are part of your design pack.
Included
-
What if my £50,000 is wasted?
Firm offer or your money back.
Guaranteed
Reserve, see the firm price and date, then decide.
Step 1
Reserve: £50,000
£20,000 at signing. Within 10 business days we check your import, plant, access and our partner's willingness to quote.
£30,000 once the check passes. Within 30 business days: your site design, a firm price and a dated programme backed by the factory slot.
Credited in full against your order.
Step 2
Order
Accept the firm offer and pay against milestones, never ahead of progress. The manufacturer's warranties pass through to you.
Step 3
Run
Handover after site acceptance. Optional maintenance and site response through a local partner. The GPUs stay yours.
Facility price £4M to £7M excluding VAT and GPUs, depending on the site (estimate). It's made firm during the reservation.
Firm offer or your money back.
If we can't give you a firm, partner-backed price and delivery date within 30 business days of your site check passing, you get the full £50,000 back.
Pay the last 5% after it works.
The final payment is due only after the pod passes its site acceptance tests, including a full-load test, on your site.
You bring the site and the GPUs. We do the rest.
You
- Share your site and supply details
- Bring your GPUs, or your financier
- Agree the firm offer
Hexi
- Site check, design and the independent engineering review
- Manufacturer, factory slot and the firm price
- Planning route, installers and the HV connection works
- Delivery, commissioning and acceptance tests
- One accountable contact from first call to handover
Designed for your site, not picked from a catalogue.
The Hexi engine simulates how a facility gets from an empty plot to working GPUs: grid and equipment lead times, cooling against real hourly weather, outages and reliability, cost and cash flow. It searches the design options for the one that gets your GPUs working soonest at the lowest risk-adjusted cost, and records every assumption.
We tested it on three real Texas sites.
24 vs 51
months to first GPUs: the engine's plan against waiting for the grid, for the same 100 MW campus
$4.20
per useful GPU-hour, against $4.62 for waiting for the grid
3 reviews
independent rounds of review of the model before these numbers were published
What was assumed
A screening exhibit, with nothing confirmed by the utility. Each plan is the median of 300 simulated futures, with 2024 hourly weather and costs from 475 sourced or labelled table rows. Title, gas supply and connection capacity were not verified. Read the method and caveats.
Pod 1 is right for you if...
- you've bought or financed about 500 GB300 GPUs, or are about to;
- you control a UK site with spare grid import, or can get one;
- you need dedicated capacity sooner than a new connection or a colocation slot allows.
...and probably not if:
- you need Tier III resilience through utility outages (we can design it, but it isn't Pod 1);
- your workload is short or uncertain. Cloud is often cheaper, and we'll tell you so.
We're building the delivery network for Pod 1.
Manufacturers
Factory-built, liquid-cooled modules for 140 kW racks. We bring site-designed orders with deposits behind them.
Installers and engineers
Civils, HV and electrical installation, commissioning, maintenance and site response across the UK.
Site owners
Spare grid import at an industrial, logistics or storage site? Host a pod and earn from capacity you aren't using.
Questions
Who owns the GPUs?
You, or your financier. We never take title to them.
Is Pod 1 Tier III?
No. It has one utility supply and a UPS for safe shutdown. If you need to run through utility outages, we design and price a resilience option.
What happens if my site fails the check?
You'll know within 10 business days of giving us the site information. We keep only our earned costs, up to £20,000, and refund the rest.
Where is it built?
In our manufacturing partner's factory. It's then delivered, installed on your site and tested before handover.
How long does it take?
26 to 40 weeks from order where suitable electrical plant is already on site, or 52 to 78 weeks if a new transformer and switchgear are needed. These are estimates; we confirm the programme with our partner before you order.
What about networking and storage?
Your cluster networking, storage and software are yours to choose. We provide the space, power, cooling and interfaces.
Have GPUs coming and a site with power?
Start with the free check. In two business days you'll know whether Pod 1 fits your site.