SME workstation · two 32GB GPUs
Company 64
Two independent RTX 5090 workers in a supplier-validated AI tower.
A supplier-validated dual-GPU workstation for teams needing two local workers or qualified PCIe sharding, with no NVLink or pooled-memory claim.
Current quotation and acceptance plan required.
Price, exact parts, compatibility, warranty, delivery and workload results are indicative until confirmed in a written quotation and acceptance record.
Key buying facts
Company 64 at a glance
- Complete private AI deployment
- £18,500 ex VAT · final quotation required
- GPU configuration
- 2 × NVIDIA GeForce RTX 5090 32GB 2 GPUs
- Per-GPU memory
- 32GB physical VRAM per GPU
- Physical GPU total
- 64GB across 2 GPUs · not automatically pooled
- Physical class
- Workstation Dual-GPU full-tower AI workstation
How the memory works: 64GB is the physical total across 2 separate 32GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.
Power planning · Workstation
2050W PSU; measured wall power and site circuit requirement pending
Representative intended fit
Two independent private AI or creative queues
Buyer fit
Start with the reason to own it.
A practical two-worker tower for an SME that needs more throughput but cannot install a rack.
Complete private AI deployment
£18,500
ex VAT · £22,200 inc VAT
Includes workload sizing, the configured system, AI software stack, security baseline, burn-in, agreed workload testing, documentation, remote onboarding and 30-day configuration-defect support.
Package figures exclude VAT and cover the stated GPU RIGS deployment scope. Every final order requires a written quotation confirming the specification, availability, delivery, warranty and workload acceptance plan.
A credible fit
- Two independent private AI or creative queues
- A measured runtime that supports the exact dual-GPU topology
- A floor-standing deployment with suitable power, heat and noise tolerance
Choose another route when
- One model needs a guaranteed contiguous 64GB space
- The site cannot accept the measured dual-GPU load
- Rack serviceability or redundancy is required
System evidence
What is confirmed, calculated and still unknown.
Vendor-documented values, calculations, assumptions and unknowns remain separate. A capability is marked as measured only when a reproducible benchmark exists.
- Commercial status
- indicative
- Reviewed
- Review again
- Open specification items
- 4
01 / model evidence
Exact checkpoint profiles
0 measured-supported profiles. A named model appears as supported only after the exact checkpoint, runtime, deployment and limits have a reproducible benchmark record.
Current boundary
No exact tested model-fit profile is listed for this system. GPU memory, a supplier capability statement or a model family name must not be read as a compatibility guarantee.
This does not mean that every model is unsupported. It means untested memory arithmetic or a generic GPU claim is not treated as a model match.
02 / workload acceptance
Beyond language models
Each workload needs its own quality, latency, throughput, stability and resource test. The entries below define what would be measured; they are not performance claims.
01
Retrieval-augmented generation
Retrieval quality, citation support, permissions, refusal behaviour and latency against a versioned corpus and question set; document count alone is not a hardware metric.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
02
Software-engineering assistance
Task correctness, test pass rate, unsafe-change rate, reviewer effort and useful response time on a versioned repository evaluation set.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
03
Speech to text
Word error rate, real-time factor and failure rate on a versioned, representative audio set.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
04
Embedding
Vectors per second, query latency and retrieval-quality metric on a versioned corpus with the exact embedding checkpoint and dimensions.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
05
Image generation
Latency percentiles, images per second, peak memory, stability and accepted-output rate at exact checkpoint, resolution, steps, sampler, CFG and batch.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
06
GPU rendering
Frame latency, throughput, errors, wall power and temperature for an immutable scene using the exact renderer, version, resolution and sample count.
Acceptance definition only. No performance, quality, capacity or suitability result has been measured for this system.
03 / configuration
Specification with its certainty attached
The current sources show the basis for each value. The final quotation names the exact supplier, parts, warranty and bill of materials for the ordered system.
| Field | Public value | Evidence state | Qualification |
|---|---|---|---|
| Platform class | Dual-GPU full-tower AI workstation | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. This is a current public reference configuration, not a GPU RIGS trade quote or proof of every sustained customer workload. |
| Chassis class | Full-tower workstation | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. |
| GPU route | 2 × NVIDIA GeForce RTX 5090 32GB | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. Exact board part numbers and serials belong in the final bill of materials. |
| Physical GPU memory total | 64GB across 2 GPUs | Calculated | Calculated from disclosed inputs; not a measured system result. 64GB is the physical total across 2 separate 32GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing. |
| Per-GPU memory | 32GB | Vendor documented | Documented component capacity and the safer first sizing boundary before any supported multi-GPU test. |
| GPU interconnect | PCIe route; the selected supplier lists NVLink as absent, while exact lane allocation and peer-to-peer behaviour remain pending | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. No universally pooled-memory claim is made. Multi-GPU execution depends on the named model, runtime, topology and validation result. |
| System topology | 2 × 32GB over PCIe; no pooled-memory or NVLink claim | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. the selected supplier explicitly lists NVLink as absent. Supported sharding and peer-to-peer behaviour depend on the named model, runtime and measured topology. |
| CPU | AMD Ryzen Threadripper 9960X, 24 cores / 48 threads | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. |
| System memory | 128GB ECC DDR5 | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. |
| Primary storage | 2TB PCIe 5.0 NVMe SSD | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. |
| Network | 10GbE and 2.5GbE Ethernet | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. |
| Cooling | Active-air dual-GPU tower with 360mm CPU liquid cooling | Vendor documented | Documented by the component or platform vendor; not independently measured by GPU RIGS. Sustained dual-GPU AI thermals and acoustics remain a GPU RIGS acceptance test. |
| GPU RIGS package figure | £18,500 ex VAT | Assumption | Complete GPU RIGS private AI deployment for the stated reference configuration. Final written quotation confirms exact parts, availability, delivery, warranty and accepted workload scope. |
04 / site and facility
The room stays inside the system boundary
Nameplate power, supplier cooling descriptions and aggregate GPU TDP are not a site approval. The exact build and customer facility need written acceptance.
- Input power
- 2050W PSU; measured wall power and site circuit requirement pending Vendor documented Documented by the component or platform vendor; not independently measured by GPU RIGS. PSU capacity is not measured system draw or proof that an ordinary office circuit is suitable.
- Measured heat output
- Measured wall power and facility heat-rejection requirement pending reference build Unknown Not yet evidenced for the ordered system and must be resolved before acceptance.
- Cooling route
- Active-air dual-GPU tower with 360mm CPU liquid cooling Vendor documented Documented by the component or platform vendor; not independently measured by GPU RIGS. Sustained dual-GPU AI thermals and acoustics remain a GPU RIGS acceptance test.
- Dimensions and handling
- 245 × 515 × 575mm full tower Vendor documented Documented by the component or platform vendor; not independently measured by GPU RIGS.
- Facility acceptance
- Office placement still requires circuit, heat and noise acceptance Unknown The customer site has not been surveyed. The final ordered configuration and qualified facility design control this decision.
Pre-order site checks
- A ventilated floor or desk-side position with clear intake and exhaust paths
- A suitable UK circuit and an agreed UPS decision after measured-load review
- An acoustic and heat check under the accepted workload
- A named owner for accounts, updates, backups and incident response
05 / delivery and support
A handover boundary, not an implied managed service
The final quotation must name who owns procurement, facility work, acceptance, warranty, remote support, recurring operations and any on-site service.
- Hardware warranty
- 36-month listed the selected workstation supplier warranty; exact onsite versus return-to-base scope requires written quote Vendor documented Documented by the component or platform vendor; not independently measured by GPU RIGS. The supplier page contains both onsite-first-year marketing copy and return-to-base specification text, so GPU RIGS must reconcile the exact ordered warranty in writing.
- Availability and lead time
- Pre-order; 3–5-day ETA observed 28 July 2026 Vendor documented Documented by the component or platform vendor; not independently measured by GPU RIGS. Dated public availability only; it is not reserved stock or a customer delivery promise.
- GPU RIGS support baseline
- Remote onboarding and 30-day configuration-defect support Assumption The written quotation and contract confirm this service boundary. It is not a 24-hour managed service or an implied on-site warranty.
Standard service boundary
- 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
06 / sources and review status
Sources and unresolved items
Review dates show evidence freshness, not future availability. A current quote, serialised bill of materials and accepted test record control the order.
- Commercial state
- indicative
- Reviewed
- 2026-07-28
- Review again
- 2026-08-28
Unresolved before acceptance
- GPU RIGS package figure: £18,500 ex VAT
- Measured heat output: Measured wall power and facility heat-rejection requirement pending reference build
- Facility acceptance: Office placement still requires circuit, heat and noise acceptance
- GPU RIGS support baseline: Remote onboarding and 30-day configuration-defect support
-
Product record
Current product specification and commercial status
GPU RIGS
Retrieved 2026-07-28 · reviewed 2026-07-28
The current product record defines the specification and commercial status shown on this page.
-
Product record
Current product specification and commercial status
GPU RIGS
Retrieved 2026-07-28 · reviewed 2026-07-28
The current product record defines the specification and commercial status shown on this page.
-
Supplier reference
Dated supplier configuration evidence
System supplier
Retrieved 2026-07-28 · reviewed 2026-07-28
Dated supplier evidence. Configuration, stock, price and warranty require confirmation in the final quotation.
-
Supplier reference
Dated supplier configuration evidence
System supplier
Retrieved 2026-07-28 · reviewed 2026-07-28
Dated supplier evidence. Configuration, stock, price and warranty require confirmation in the final quotation.
Known limitations
- 64GB is aggregate capacity across two 32GB GPUs, not one universal memory pool.
- No multi-GPU model fit, thermals, noise or wall power has been measured.
- 64GB is aggregate capacity across two 32GB GPUs, not one universal memory pool.
- No multi-GPU model fit, thermals, noise or wall power has been measured.
- 64GB is aggregate capacity across two independent 32GB GPUs; the selected supplier lists no NVLink.
- A workstation is not presented as a redundant datacentre appliance.
- No model, speed, context, concurrency or output-quality promise exists without a controlled fit profile.
- Public supplier price and availability observations are dated inputs, not reserved stock.
Do not buy this route when
- One model needs a guaranteed contiguous 64GB memory space.
- A quiet desk-side role or ordinary electrical circuit cannot accept the final measured load.
- A rack appliance offers a better warranty, serviceability or operating model.
Inspect the evidence behind Company 64.
The system view shows the selected hardware reference where available. Technical diagrams explain memory, data boundaries, queues, power and handover. The final bill of materials and workload test confirm the ordered system.
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.
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, SGLang, TensorRT-LLM 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.
Performance is confirmed against the accepted workload, exact build and disclosed test conditions. No throughput, latency or quality figure is claimed before that test.
- 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.
See how systems are testedCommercial 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 figure is for a complete GPU RIGS private AI deployment, not unconfigured hardware. The final written quotation confirms the exact specification, delivery, warranty and accepted workload scope.
- 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.
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 intended use.
- Vast.ai, Render, Golem and direct batch work do not guarantee acceptance, demand, rate or income.
- Any dated estimate must identify its source date; a pilot must report achieved utilisation, fees, electricity, cooling, faults and operator time.
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
- A workstation is not presented as a redundant datacentre appliance.
- No model, speed, context, concurrency or output-quality promise exists without a controlled fit profile.
- Public supplier price and availability observations are dated inputs, not reserved stock.
- 64GB is aggregate capacity across two 32GB GPUs, not one universal memory pool.
- No multi-GPU model fit, thermals, noise or wall power has been measured.
- 64GB is aggregate capacity across two independent 32GB GPUs; the selected supplier lists no NVLink.
- A workstation is not presented as a redundant datacentre appliance.
- No model, speed, context, concurrency or output-quality promise exists without a controlled fit profile.
- Public supplier price and availability observations are dated inputs, not reserved stock.
Consider the next workstation only when the accepted memory or worker requirement justifies it.
Consider this route Value Rack 128Use the first rack route when shared operation, serviceability or several workers matter more than desk-side placement.
Consider this route Hosted or hybrid AIOften better for irregular demand, elastic frontier quality or teams without an internal operator.
Consider this routeQuestions answered
Company 64 questions that affect the order
The written quotation and acceptance plan confirm any price, specification, warranty or workload assumption discussed below.
Does Company 64 provide 64GB as one memory pool?
Only a single-GPU system provides its stated GPU memory on one card. Multi-GPU totals are aggregate physical capacity; usable sharding depends on the exact model, runtime and topology.
Which models are supported?
Only models with a reproducible fit profile are listed as supported. Capacity candidates are labelled separately; a memory calculation alone is not a support claim.
Is the displayed price final?
No. An indicative price is a transparent planning value based on dated inputs. Quote-only platforms and the final order require a current supplier quote, accepted bill of materials and delivery scope.
Can spare capacity earn income?
It may be evaluated as an optional isolated secondary use. Marketplace acceptance, utilisation, rates, fees, energy and income can change and are never guaranteed or included in the purchase case.
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.
Package figures exclude VAT and cover the stated GPU RIGS deployment scope. Every final order requires a written quotation confirming the specification, availability, delivery, warranty and workload acceptance plan.