Private AI proof platform

Core 48

A practical first private GPU server - once the warranty route is settled.

Two independent 24GB GPU workers for private document assistants, coding, rendering and measured proof-of-value projects.

Validation-stage standard configuration.

This public proposal remains indexable while clearly qualifying its evidence state. Price, exact parts, compatibility, warranty, delivery and workload results require a current quotation and written acceptance record.

Plan around Core 48 Compare all systems

Price status: Indicative, ex VAT
Commercial model: Reviewed 25 July 2026
Range position: Standard progression

Supplier platform / proposed build

Inspect the platform

48GB aggregate VRAM* OEM 4U single-socket GPU server platform proposed physical platform GPU route 2 × refurbished NVIDIA RTX A5000 24GB

Buyer fit

Start with the reason to own it.

A measured first rack platform for organisations that have outgrown a light hosted trial but do not yet need a large professional-GPU endpoint.

Indicative purchase price

£27,500

ex VAT · £33,000 inc VAT

Proposed indicative price, excluding VAT. Final price, specification, availability, delivery, warranty and model fit require a current supplier quote and written order confirmation.

A credible fit

Departmental private chat and RAG trials
Independent coding, image or transcription workers
A measured route into rack-based private AI

Choose another route when

A guaranteed 48GB single-memory workload
An office without rack, heat and noise planning
A buyer who requires a settled multi-year GPU warranty today

Interactive system inspection

Inspect the platform behind Core 48.

Move from the complete Core 48 platform family into its internal layout, cooling, accelerator plane and management boundary. These views explain the engineering questions; the final ordered build and acceptance record remain decisive.

Three-quarter view of a 4U multi-GPU rack server platform — 01 **4U / 19-inch rack**

01 Platform

Start with the complete machine.

A serviceable 4U rack platform provides the physical boundary. Exact dimensions, rails, weight, power supplies and ordered components still belong in the written quotation.

Inspection focus 4U / 19-inch rack

02 Internal layout

See where density becomes an engineering decision.

GPU positions, processor sockets, memory, storage and cable paths compete for space and airflow. The final layout must be compatible as one system, not merely as a component list.

Inspection focus GPU / CPU / memory

03 Cooling

Treat heat removal as part of the product.

High-airflow fan modules serve a tightly controlled front-to-back path. Loaded wall power, room conditions, noise and heat rejection require a measured site and reference-build record.

Inspection focus Airflow / service access

04 Accelerator plane

Count independent workers, not imaginary pooled memory.

A dense accelerator layout can run separate jobs or supported parallel workloads. Per-GPU memory, runtime behaviour, context and concurrency decide what the system can actually serve.

Inspection focus Per-GPU fit first

05 Operations

Make management visible before handover.

Remote health, access, sensors, logs, credentials, recovery and update ownership are part of the appliance. The final interface and permissions are recorded for the ordered platform.

Inspection focus Observe / recover / own

Workload route

Every use needs its own acceptance test.

Model name and aggregate VRAM do not prove a business outcome. Each proposed route below states how the capacity could be used and what must be tested before the order treats it as suitable.

Private document assistant

Proposed use: Run a bounded RAG pilot over an approved document set on one 24GB GPU, with ingestion or another service kept separate.
Acceptance evidence: Test retrieval, source support, refusal and response time against a written question set.

Coding or transcription worker

Proposed use: Use the second GPU for a separate code, speech, embedding or image queue where the selected model fits 24GB.
Acceptance evidence: Confirm model fit, queue behaviour and useful output on representative non-confidential work.

Batch and render proof

Proposed use: Schedule independent jobs across the two GPUs without claiming one shared 48GB memory space.
Acceptance evidence: Measure completed useful work, wall power, temperature and failure recovery.

Specification certainty ledger

Known, proposed and still to be confirmed.

A proposal should expose missing facts. The final order replaces every confirmation row with an exact part, measured result or named customer decision.

Package proposal Supplier confirmation Customer decision

Proposed Core 48 specification and evidence status
Item	Current value	Status	What the record must show
Form factor	Rack server	Package proposal	The site, delivery and support route is designed around this physical class.
Physical platform	OEM 4U single-socket GPU server platform	Package proposal	Platform family selected for the proposed configuration.
GPU route	2 × refurbished NVIDIA RTX A5000 24GB	Package proposal	Exact manufacturer, part number, condition and serials belong in the final bill of materials.
Per-GPU memory	24GB	Package proposal	The safer model-fit starting point before any supported multi-GPU test.
Aggregate GPU memory	48GB across 2 GPUs	Package proposal	Not presented as one universal memory pool.
System memory	128GB	Package proposal	Memory population, speed and expansion route require the final platform bill.
Primary storage	2TB NVMe	Package proposal	Drive model, endurance, layout and backup destination are confirmed in the order.
Network	10GbE	Package proposal	Customer switching, cabling, storage traffic and segmentation remain part of site design.
Power planning	Plan around 1.0kW under load; verify the exact build	Customer decision	A qualified site review and measured reference build must replace the planning figure.
CPU and motherboard	To be stated in the final bill of materials	Supplier confirmation	No unverified processor, lane or motherboard claim is made on this proposal page.
Power supplies and leads	To be confirmed for the ordered build and UK site	Supplier confirmation	Include PSU count, rating, redundancy position, input requirements and lead specification.
Remote management	Platform route to be confirmed and access policy agreed	Supplier confirmation	The supplier interface shown in the gallery is a reference, not a promise of the final feature set.
Dimensions and weight	Exact ordered-system values required	Supplier confirmation	Placement, handling and the delivery route depend on the exact ordered-system values.
Hardware warranty route	Exact supplier and component warranty route required	Supplier confirmation	The final order records the whole-system, component, onsite, return and dead-on-arrival boundaries.

24GB

Per GPU is the first sizing boundary.

Core 48 has 2 GPUs and 48GB aggregate VRAM. A runtime may split a supported model or distribute independent jobs, but the page does not claim one physical 48GB memory pool. Quantisation, context, KV cache, batching and concurrency still change fit.

Software and security

The usable product is more than the chassis.

The final stack stays deliberately small. Versions, licences, access and recurring ownership are recorded so the customer is not left with an opaque collection of containers.

Proposed software baseline

Ubuntu LTS on a recorded operating-system version
NVIDIA driver, CUDA components and container support validated for the ordered hardware
Docker Engine and NVIDIA Container Toolkit
One primary model server selected from Ollama, vLLM or llama.cpp for the accepted workload
Open WebUI or another reviewed browser interface
Named authentication, TLS and reverse-proxy approach
GPU, node and service monitoring with an agreed log-retention period
Pinned versions, a software bill of materials and a model source and licence record

Security ownership

GPU RIGS baseline: Initial operating-system state, named administrator handover, host firewall baseline, agreed access route, secrets transfer and documented update state.
Customer or contracted operator: User lifecycle, network and VPN policy, backups, monitoring review, patch approval, incident response, data governance and lawful use.
Shared before acceptance: Model and software licence checks, retention and logging choices, recovery test, acceptance criteria and a named owner for every recurring task.

Local infrastructure can reduce disclosure to external AI APIs. It does not automatically make the service secure, accurate, confidential or UK GDPR compliant.

Testing and acceptance

The evidence pack is part of the machine.

No public benchmark is invented for this page. The accepted workload, exact build and disclosed test conditions decide what can be claimed after the reference system exists.

01
Record the final bill of materials, serial numbers and firmware versions.
02
Run at least 24 hours of GPU, CPU, memory and storage stress testing.
03
Capture temperature, fan, error, health and wall-power evidence under the agreed load.
04
Test cold boot, restart and the available remote-management route.
05
Check drive health, network throughput and the container and GPU runtime.
06
Run model and workload smoke tests against the written acceptance set.
07
Record any measured speed only with the model, quantisation, context, concurrency and runtime disclosed.
08
Test an agreed fault or recovery route and provide the resulting handover record.

What is not claimed today

No fixed users, tokens per second, latency, model size, accuracy, availability, savings or marketplace contribution is stated without the missing configuration and test conditions.

Read the evidence method

Site, power and cooling

The room is part of the specification.

Planning power is not a measured promise. The final pre-flight records the circuit, voltage, loaded wall power, heat route, noise tolerance, rack, network, UPS decision and operating owner.

Read the site guide

1.0kW

Current loaded-system planning basis

A suitable 4U rack or colocation position
An electrical circuit confirmed for the final measured load, planned around 1.0kW
Ventilation and cooling sized from measured wall power rather than GPU nameplate power alone
A noise review before any office-adjacent placement
10GbE network planning, remote-management access and a named technical owner

Delivery, support and warranty

A written boundary before money moves.

The proposed route covers discovery, configuration, evidence and remote handover. It does not quietly absorb facilities work, migration or an always-on managed service.

Included in the proposed baseline

Documented workload and site-fit review
Confirmed bill of materials before procurement
Configuration, burn-in and agreed smoke-test evidence
Asset schedule, admin notes and user quick-start material
Collection or the quoted kerbside or pallet-delivery route
Remote onboarding and 30-day configuration-defect support

Separate scope or customer responsibility

Building electrical work, rack, UPS, cooling or structured cabling
Nationwide on-site installation unless separately quoted
Migration of customer data, every integration or every application
Continuous managed operations, security monitoring or a 24-hour support agreement
Third-party model, API, marketplace or software charges
A compliance certificate, performance guarantee or income guarantee

Supplier and warranty gates

Unresolved items stay visible.

Written confirmation of the exact refurbished GPU condition and serialised bill of materials
A settled route for the supplier-stated one-year GPU warranty and Opace replacement reserve
Whole-system, component, dead-on-arrival and return responsibilities recorded before order
Exact GPU, riser, firmware, power and thermal compatibility confirmed in writing
Current delivered price, lead time and UK delivery terms confirmed in the quotation

Evidence and change record

Proposal inputs remain dated.

Package model: UK proposal reviewed 25 July 2026
Price state: Calculated from dated public supplier inputs, not a live quotation
Performance state: No public benchmark until the exact reference build is tested
OEM platform record: Exact platform identity and bill of materials provided in the written quotation.

Commercial reality

Tax and spare capacity are supporting questions.

Neither belongs in a guaranteed saving or payback claim. The purchase must stand on the accepted workload, control case and operating plan.

Finance, VAT and capital allowances

The displayed price is an indicative proposal based on dated public supplier inputs. The final bill of materials and supplier quote control the order.
Prices exclude VAT. VAT recovery depends on the buyer, its taxable activities and the normal evidence rules.
Qualifying equipment may be plant and machinery for capital-allowance purposes. The buyer's accountant decides eligibility and timing.
A third-party lease or hire-purchase route may be explored after partner validation and credit approval. GPU RIGS is not presented as a lender.
There is no generic capital-gains advantage and no automatic research and development relief because the equipment supports AI.

Read the guarded UK buyer notes

Optional idle capacity

Marketplace mode is off by default and is excluded from the purchase case.
A separate environment, no customer data mounts, network controls and a local kill switch would be required.
The customer, insurer, supplier warranty and marketplace terms must permit the proposed use.
Vast.ai, Render, Golem and direct batch work do not guarantee acceptance, demand, rate or income.
Any pilot must report achieved utilisation, fees, electricity, cooling, faults and operator time before continuation.

Review the marketplace decision gate

Limits and alternatives

A good specification leaves room for “no”.

The fit check can recommend a smaller system, hosted service, hybrid route, custom build or no purchase. That is preferable to making this package carry a workload it has not proved.

Package boundaries

The two GPUs provide 48GB aggregate VRAM, not one universal 48GB pool.
The supplier currently describes these GPUs as refurbished with a one-year GPU warranty.
This package remains proposed until the warranty and exact configuration are confirmed.
Each GPU has 24GB of local VRAM. The 48GB figure is aggregate capacity.
A model that needs more than 24GB is not assumed to run well across both GPUs.
No user-count, token-rate, latency or accuracy claim is made before the workload test.
The package is not suitable where a new-GPU warranty is mandatory and no replacement route has been agreed.

Hosted AI or a workstation

Often the better choice for a light team, frontier-model requirement or office without rack and cooling capacity.

Consider this route Business 128

Consider when four independent workers, greater throughput or a wider mixed-workload plan is justified.

Consider this route Custom specification

Use when new professional GPUs, different storage or a quieter form factor is a firm requirement.

Consider this route

Questions answered

Core 48 questions that affect the order

These answers preserve the validation boundary. The written quotation and acceptance plan replace proposal assumptions.

Can Core 48 run one 48GB model?

Not by default. It has two GPUs with 24GB each. A selected runtime may split some models across them, but model, context, cache and performance must be tested before that becomes an accepted use.

Are the RTX A5000 GPUs new?

The public supplier configuration describes them as refurbished. The exact condition, serials, GPU warranty and replacement route must be written into the final quotation and order.

Can it sit under a desk?

It is a 4U rack platform. Power, heat, airflow and noise need a site review, so a server room, equipment space or colocation route is normally more suitable.

Is the £27,500 price final?

No. It is an indicative ex-VAT proposal based on dated public inputs. The exact configuration, supplier quote, warranty and delivery terms determine the final price.

Prepare the next decision

Specify the work before the parts.

Record the workload, data boundary, users, site and acceptance test. No confidential documents or credentials are needed for the first brief.

Proposed indicative price, excluding VAT. Final price, specification, availability, delivery, warranty and model fit require a current supplier quote and written order confirmation.

Build the Core 48 brief Read the delivery boundary