Alexa Compute UK-sovereign, green-first compute.
Sovereign, sustainable, high-density compute — built in South Yorkshire.
Serious compute is scarce, expensive, and not on British soil.
UK organisations that need GPU/AI compute today face costly, queued US hyperscalers where data leaves UK control — or a handful of mostly-US specialist GPU clouds. Building it yourself is impractical. Demand has never been higher; supply has never been tighter.
Expensive
On-demand GPU hours and reserved blocks at hyperscalers are priced for scale buyers, not UK startups or mid-market.
Queued
Lead times for flagship accelerators run into months. Capacity is rationed.
Foreign & locked-in
Data exits the UK; contracts are under foreign law; vendor lock-in is by design.
Unsustainable
Carbon intensity and water usage of legacy compute are increasingly under scrutiny.
Rentable, sovereign, no-lock-in compute — built in the UK.
Bare-metal and GPU rental, reserved private clusters, colocation, and managed services. Cheaper than hyperscalers on targeted workloads. Data on British soil. Green-first. A real human answers the phone.
| What you care about | Alexa Compute | US hyperscaler | US GPU cloud |
|---|---|---|---|
| UK data residency | ✓ | Mostly US | US-centric |
| Green-first power | ✓ | Varies | Limited |
| No vendor lock-in | ✓ | ✗ | Partial |
| AMD + Arm options | ✓ | Niche | Rare |
| Real-human service | ✓ | Tier 1 only | Sometimes |
Best-in-class, vendor-neutral hardware.
AMD EPYC as the value default. AmpereOne Arm nodes as a perf/watt + British-tech differentiator. Both NVIDIA H200 and AMD Instinct MI300X GPUs — offer both, customers choose. 800G networking and direct-to-chip liquid cooling on the GPU rows.
AMD EPYC "Turin"
12-channel DDR5, 96–192 cores, PCIe Gen5. The default for general compute.
AmpereOne (Arm)
192 cores, 12-ch DDR5. Best perf/W and the "British tech" angle.
NVIDIA H200 / AMD MI300X
Offer both. Customers pick CUDA-locked or open stack.
Direct-to-chip liquid cooling
1.05–1.15 PUE on AI rows; quiet, low-water.
Hellaby, Rotherham — and the honest answer on "solar 24/7".
Long-term home: a ~25–40k sq ft warehouse in Hellaby, South Yorkshire — land and power at a fraction of London/Slough. Solar alone can't power a 24/7 compute facility in the UK (capacity factor ~10–11%); the honest answer is a renewable-first stack: grid PPA, on-site solar, battery storage, and HVO-fueled backup. That's both true and more fundable than the greenwashed version.
Grid PPA
Contractually renewable primary supply via fixed tariff or PPA.
600–900 kWp solar
Rooftop PV — offsets and peaks; not the whole story.
~1 MWh battery
Peak-shaving + ride-through; shaving grid draw and cost.
HVO backup
~90% lower net CO₂ than diesel, drop-in; used by UK operators.
Contract-led, capital-disciplined. Build it only once it's paid for.
We don't spend the next tranche until the previous milestone (usually signed revenue) is hit.
Foundation
Incorporate Alexa Compute Ltd, get SEIS/EIS advance assurance, raise pre-seed.
Launch via colocation
GPU + CPU fleet in an existing UK data centre. First paying customers in months, not years.
Hellaby fit-out
Acquire the warehouse, apply for grid, install solar/BESS/HVO, commission D2C cooling.
Scale
Expand fleet against demand, not vanity. ISO 27001, SOC 2, public-sector sales.
Two complementary founders. Lean by design.
A clear, non-overlapping split: one owns the outward — money, customers, story; one owns the inward — build it, run it, keep it up.
Alexa — CEO
Vision, fundraising, customers, partnerships. Face of the company to investors and government.
A. Fahmy — CTO
Hands-on infrastructure: hardware, network, automation, security, SRE. Builds and operates it.
Let's talk.
Investors, anchor customers, advisors, partners — all welcome. We respond quickly.