What Cudo Compute does

Cudo Compute (legal entity: Cudo Ventures Limited) is a UK-based, privately held AI infrastructure company that designs, commissions, and operates enterprise-grade NVIDIA GPU clusters for production AI training and inference workloads. The company addresses end-to-end deployment constraints — land, power, and compute — by integrating site selection, hardware procurement through NVIDIA Preferred Partner channels, NVIDIA reference-aligned cluster design covering InfiniBand fabrics, high-performance storage (VAST, WEKA, DDN, NetApp), and direct liquid cooling for GB200/GB300 NVL72 rack-scale systems, with 24/7 SRE-led operations under a single accountable model. The product portfolio spans Hopper (H100, H200), Blackwell (B100, B200, B300, GB200 NVL72, GB300 NVL72), and pre-launch Vera Rubin (VR200) hardware, delivered in ISO 27001-certified data centres across North America, Europe, the UK, and MENA.

The business generates revenue through three streams: managed GPU cluster services on multi-year contracts, subscription or usage-based cloud GPU compute with OPEX consumption models compatible with bare metal, enterprise, and GPU-as-a-service structures, and one-time professional services for deployment, commissioning, and benchmarking. Customer segments include large enterprises requiring dedicated clusters from 128 to 40,000+ GPUs, AI-first companies and frontier model developers prioritising access to latest-generation hardware, and government and sovereign AI initiatives requiring jurisdiction-specific infrastructure with data residency, GDPR, CCPA, and FedRAMP-aligned controls. Commercial go-to-market is exclusively enterprise field sales with quote-based pricing, no public list rates, and primary intake through website enquiry forms targeting multi-year contracts.

The company has secured 250MW+ of contracted power capacity by end of 2026 with a 750MW+ target by end of 2027 and a multi-GW pipeline including exclusive European sites with power commitments and grid underwriting. Founder and CEO Matt Hawkins, who previously founded C4L, leads a senior team of seven executives including CTO Tim Dyce (ex-CERN and European Bioinformatics Institute), CFO David Bell, CSO Mark Denney (former CIO of HMRC and Barclays), and COO Lee Woodham with an explicit Series B mandate. No institutional funding rounds, revenue figures, or valuation disclosures are available.

Name

Cudo Compute

Legal name

Cudo Ventures Limited

Website

https://cudocompute.com

Company type

Private

Founded year

2021

Operating status

Operating

Headcount range

11–50 employees

Short description

Cudo Compute is a UK-based AI infrastructure company that designs, commissions, and operates production-scale NVIDIA GPU clusters for enterprise, government, and AI-first customers across North America, Europe, UK, and MENA, leveraging NVIDIA Preferred Partner supply chain access and an integrated land-power-compute delivery model.

Ownership category

akta.pro rank

Product category

GPU Cloud Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (51821), Computer Systems Design and Related Services (5415)

SIC

Electronic Computers (3571), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

AI Compute Cloud & GPU-as-a-Service (HDAAAAAK)

akta.pro secondary industries

GPU/Accelerator Compute Servers (HDABADAC), GPU-Accelerated & AI Training/Inference Servers (HDACABAG), GPU-Accelerated HPC Systems (HDACAOAC)

Keywords

• Gpu cloud infrastructure
• Ai infrastructure services
• Bare metal gpu
• Enterprise ai compute
• Data center operations

Headquarters

HQ city

Bournemouth

HQ country

United Kingdom

HQ region

Europe

Offices3 records

Markets served

GTM type

B2B

Offering type

Services

Cost components

Infrastructure, Supply Chain, Technology or R&D, Personnel, Operations, Marketing or Sales

Revenue model

1. Managed GPU cluster services: Dedicated bare-metal GPU clusters managed by CUDO from deployment through 24/7 operations. Includes cluster commissioning, monitoring, firmware lifecycle management, and L3 support. Revenue recognised as a recurring managed services contract.
2. Cloud GPU compute (subscription / usage-based): GPU compute access via subscription or consumption-based models. OPEX consumption models for predictable spend across hardware generations. Compatible with bare metal, enterprise, and GPU-as-a-service commercial structures.
3. End-to-end deployment and commissioning services: Deployment services covering site readiness assessment, rack deployment, cooling infrastructure, networking, OS and driver provisioning, and cluster benchmarking before handoff. One-time professional services revenue at deployment.

Pricing tiers

Go-to-market motion2 records

Distribution channels1 record

Marketing channels6 records

Core offering

CUDO Compute designs, commissions, and operates dedicated NVIDIA GPU clusters for enterprise AI training and inference workloads. The full-lifecycle offering covers site selection, power provisioning, hardware procurement through OEM channels (Dell, Lenovo, Supermicro, HPE), cluster deployment with NVIDIA reference-aligned architectures, and 24/7 operations across ISO 27001-certified facilities in North America, Europe, UK, and MENA. Deployments begin at 128 GPUs and scale to environments of up to 40,000 GPUs.

Product overview

Cudo Compute is an AI infrastructure company that designs, commissions, and operates production GPU clusters for enterprise AI workloads. The portfolio spans multiple GPU generations: Hopper (H100, H200), Blackwell (B100, B200, B300, GB200 NVL72, GB300 NVL72), and upcoming Vera Rubin (VR200). Core offerings include GPU Cluster Services (bare metal dedicated clusters), AI Factories (industrial-scale turnkey infrastructure), AI Infrastructure Managed Services (full lifecycle operations), and Enterprise AI Solutions (sovereign-ready multi-region deployments). All products are built on NVIDIA architectures with ISO 27001 and SOC 2-aligned security, deployed across North America, Europe, UK, and MENA. Cudo also publishes the 'Land. Power, Compute.' industry report covering infrastructure constraints in AI deployment.

Differentiator

Problem solved

Functional benefit

Products and services

• GPU Cluster Services Dedicated NVIDIA GPU clusters (H100, H200, B200, B300, GB200 NVL72, GB300 NVL72) for AI training, inference, and HPC workloads, deployed and operated by CUDO with 24/7 monitoring, firmware management, and NVIDIA escalation paths. Targets enterprises and government entities requiring clusters from 128 GPUs to 40,000+ GPUs.
• AI Factories Industrial-scale AI infrastructure designed, built, and operated for production AI workloads. Combines power-first site selection, high-density GPU deployment, and end-to-end operational management under a single accountable model.
• AI Infrastructure Managed Services Full-lifecycle AI infrastructure management from site selection through commissioning and 24/7 operations. Covers power provisioning, cooling, networking, storage, and cluster lifecycle including patching and firmware management.
• Enterprise AI Solutions Dedicated, managed AI environments for organisations transitioning from pilot to production. Includes sovereign-ready infrastructure with data residency, compliance controls, and multi-region deployment across ISO 27001 and SOC 2-aligned facilities.
• NVIDIA HGX B200 8-GPU Blackwell node delivering 3X faster training and 15X faster inference than H100, with 192 GB HBM3e per GPU. Designed for frontier model training, production inference at scale, HPC, and sovereign AI deployments.
• NVIDIA HGX B300 8-GPU Blackwell Ultra node with over 100 PFLOPS dense FP4, up to 288 GB HBM3e per GPU. Built for AI reasoning and large model training at node scale.
• NVIDIA GB200 NVL72 Liquid-cooled rack-scale system with 72 Blackwell GPUs and 36 Grace CPUs in a single NVLink domain. Delivers 30X faster real-time LLM inference for trillion-parameter models with up to 13.5 TB HBM3e.
• NVIDIA GB300 NVL72 72-GPU Blackwell Ultra rack-scale system delivering over 1,000 PFLOPS dense FP4 and 20 TB HBM3e. Up to 10X lower inference latency and 5X higher throughput per megawatt versus Hopper for AI reasoning, large-scale training, and agentic AI workloads.
• NVIDIA H100 Hopper-generation GPU available in PCIe and SXM form factors. 640 GB total HBM2e across 8 GPUs. Supports AI inference, deep learning, and HPC workloads on mature, widely available hardware.
• NVIDIA H200 Hopper-generation GPU with 1.1 TB total HBM3e (141 GB per GPU). Accelerates AI training and inference on larger models with enhanced memory bandwidth versus H100.
• NVIDIA B100 Blackwell-architecture GPU delivering up to 1.5 TB total HBM3e and 112 PFLOPS FP8 tensor performance. Accelerates trillion-parameter LLM inference and supports deep learning and HPC workloads.
• NVIDIA Vera Rubin (VR200) Next-generation NVIDIA platform built from six co-designed chips (Rubin GPU, Vera CPU, NVLink 6, ConnectX-9, BlueField-4, Spectrum-6). Up to 3.5X more compute and 2.8X more memory bandwidth per GPU versus Blackwell B200, with CUDO preparing managed Vera Rubin clusters for H2 2026.
• Bare Metal GPU Service Fully accessible dedicated GPU clusters providing physical machine control with no shared instances or virtual partitions. Includes renewable energy power, no noisy neighbours, NVMe SSD storage, SpectrumX local networking, and 300 Gbps external connectivity.

Quantifiable outcome

• 4,000+ GPU cluster deployed across North America, Europe, and the Middle East in approximately two months from contract to production acceptance
• +3 more outcomes

Named customers2 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration10 records

AI capability6 records

Feature5 records

Partnerships

Four partnerships are on record, tiered flagship and core.

• SingularityNET / Singularity ComputeflagshipStrategic or Co-development Partner · 4 December 2025Singularity Compute (for-profit infrastructure arm of SingularityNET) launched its first enterprise-grade NVIDIA GPU cluster in Sweden in partnership with Conapto. The cluster is managed and operated by CUDO and underpins ASI:Cloud, providing scalable AI inference via OpenAI-compatible APIs for enterprises and ASI Alliance partners. Phase I launched with early customer onboarding, additional hardware and new locations planned.
• ConaptocoreStrategic or Co-development PartnerSwedish data centre operator partnering with Singularity Compute and CUDO to host GPU clusters at a sustainable facility in Sweden. The facility hosts SingularityNET's ASI:Cloud underpinned by CUDO-operated infrastructure. Stefan Nilsson, COO of Conapto, noted alignment on operational excellence, customer experience, and transparent partnerships built on trust.
• NVIDIAflagshipTechnology or IntegrationAs an NVIDIA Preferred Partner, CUDO deploys platforms aligned to NVIDIA B200, B300, and GB300 designs, ensuring architecture integrity, supply certainty, and production-grade stability. Direct hardware access through established OEM channels (Dell, Lenovo, Supermicro, HPE). NVIDIA reference-aligned cluster architectures covering InfiniBand fabrics, high-performance storage, power, cooling, and rack layout.
• TelegramcoreStrategic or Co-development PartnerTelegram's Cocoon AI decentralised privacy-preserving AI network is powered by GPU infrastructure deployed by AlphaTON Capital and operated through CUDO Compute's marketplace. AlphaTON Capital structured GPU cluster investment specifically to support Telegram's Cocoon AI and generate revenue through external GPU rentals via CUDO's platform.

Scale indicators10 records

Recent moves6 records

Expansion highlights6 records

Broad incumbents

• NVIDIA DGX Cloud: NVIDIA's own DGX Cloud delivers dedicated NVIDIA GPU clusters as a service via partners like CoreWeave and Equinix. It competes for the same enterprise AI workloads as CUDO but from the hardware vendor itself, with broader NVIDIA portfolio integration.
• Microsoft Azure (ND H100 v5): Azure provides ND H100 v5 and Blackwell-based GPU virtual machines for AI training and inference, integrated with Azure AI services. Hyperscaler breadth and OpenAI partnership make it a broad incumbent competing for the same enterprise cluster deals as CUDO.
• Amazon Web Services (EC2 P5/P6): AWS offers H100 and Blackwell GPU instances via EC2 P5 and emerging P6 families, alongside Bedrock, SageMaker, and global data center footprint. As a hyperscaler, AWS competes with CUDO for enterprise AI workloads but with broader services and scale.
• Google Cloud (A3/A4 instances): Google Cloud's A3 and A4 instances offer H100 and Blackwell GPUs tightly integrated with Vertex AI and Google's networking stack. As a hyperscaler, Google is a broad incumbent competitor for dedicated enterprise AI cluster deployments.

Regional players

• OVHcloud: OVHcloud is a European-headquartered cloud provider offering H100 and earlier-generation GPU instances with strong data residency positioning. Comparable to CUDO's European sovereign and regulated AI focus, though with broader cloud services and weaker NVIDIA Preferred Partner positioning.

Direct peers

• CoreWeave: CoreWeave is a specialist GPU cloud provider built on NVIDIA hardware, offering bare-metal and virtualized GPU compute to AI labs and enterprises. It is the most direct competitor to CUDO Compute in dedicated NVIDIA GPU cluster services for training and inference.
• Lambda Labs: Lambda provides GPU cloud instances and on-premise GPU clusters (HGX H100, B200) for AI training and inference, with similar enterprise customer targeting and NVIDIA-centric stack. Comparable to CUDO's cluster and managed services offerings.
• Crusoe Energy: Crusoe operates large-scale, power-first AI cloud infrastructure sited near stranded or renewable energy. Its strategy of securing cheap, abundant power for GPU clusters mirrors CUDO's land-power-compute thesis and addresses similar enterprise AI workloads.
• Nebius: Nebius is an AI infrastructure provider operating full-stack GPU clusters across multiple geographies, with emphasis on NVIDIA hardware and enterprise AI customers. Comparable in vertical integration from data center to GPU cloud services.
• Together AI: Together AI runs an AI cloud platform on top of large-scale NVIDIA GPU clusters, serving frontier model developers and enterprises with inference and training. Overlapping customer base and GPU-as-a-service delivery model make it a close comparable to CUDO.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/CudoCompute

Compliance9 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles7 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Cudo Compute