What Cirrascale Cloud Services LLC does

Cirrascale Cloud Services LLC is a private neocloud provider headquartered in San Diego, California, operating since its 2017 spin-off from Cirrascale Corporation. The company delivers bare-metal cloud infrastructure purpose-built for AI training and inference workloads, supporting NVIDIA (HGX H100, H200, B200, B300, RTX PRO 6000), AMD Instinct (MI250, MI300X, MI325X, MI350), Qualcomm Cloud AI 100, Cerebras Wafer-Scale Clusters, and Tenstorrent Galaxy Blackhole accelerators within a single platform. Its product portfolio centers on two offerings: the AI Innovation Cloud for training and fine-tuning, and the Cirrascale Inference Platform, an enterprise inference-as-a-service product with dynamic accelerator selection, cross-region workload balancing, and token-based pricing. The company serves AI/ML developers, regulated industries (healthcare, defense, financial services) requiring data sovereignty, research institutions, and enterprises seeking hyperscaler alternatives, with named engagements including the Allen Institute for AI's $152M computing cluster, an early OpenAI relationship, Google Public Sector research services, and Private Google Gemini deployments for regulated buyers.

Name

Cirrascale Cloud Services LLC

Legal name

Cirrascale Cloud Services

Website

https://cirrascale.com

Company type

Private

Founded year

2015

Operating status

Operating

Headcount range

11–50 employees

Short description

Cirrascale Cloud Services is a private neocloud that delivers bare-metal AI training and inference infrastructure across NVIDIA, AMD, Qualcomm, Cerebras, and Tenstorrent accelerators, serving enterprises, regulated industries, and research institutions that need multi-accelerator flexibility and data sovereignty.

Ownership category

akta.pro rank

Product category

AI Cloud Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182), Computer Facilities Management Services (541513)

SIC

Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Processing & Data Preparation (7374)

akta.pro primary industry

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

akta.pro secondary industries

AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB), AI Inference Hardware & Edge Accelerators (HDAAAAAC)

Keywords

• AI cloud infrastructure
• Bare-metal servers
• GPU cloud services
• Private AI cloud
• AI inference platform

Headquarters

HQ city

San Diego

HQ country

United States

HQ region

North America

Offices5 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

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

Revenue model

1. Cloud Infrastructure Services: Dedicated bare-metal cloud servers and clusters for AI training and inference workloads. Revenue generated through recurring server rental charges based on hardware configuration, GPU count, and contract term. Pricing follows flat-rate monthly billing with no surprise charges.
2. Inference Platform: Token-based pricing model for the Cirrascale Inference Platform, enabling customers to pay based on AI model usage with predictable costs for budget management.
3. Professional Services: Custom support services, consulting, and managed services for customers requiring additional expertise beyond standard cloud offerings. Billed on 30-minute incremental basis or per-service quotes.
4. Storage Services: Multi-tiered storage solutions including NVMe hot-tier, local bulk storage, and S3-compatible object storage offered as add-on services with associated recurring fees.
5. Multi-GPU Server Rental: RTX-powered virtual workstations and multi-GPU servers available starting at $1,999 per 8-GPU server per month with aggressive flat-rate billing.

Pricing tiers

Go-to-market motion2 records

Distribution channels4 records

Marketing channels6 records

Core offering

Cirrascale Cloud Services operates a private AI cloud platform providing bare-metal, dedicated server infrastructure purpose-built for AI/ML training and inference workloads. The platform supports multiple AI accelerators—NVIDIA (HGX H100/H200/B200, RTX PRO 6000), AMD Instinct (MI250/MI300X/MI325X/MI350), Qualcomm Cloud AI 100, Cerebras Wafer-Scale Clusters, and Tenstorrent Galaxy—in a single environment, and offers an enterprise inference platform with dynamic workload balancing and token-based pricing.

Product overview

Cirrascale Cloud Services offers a comprehensive private AI cloud platform with two primary product lines: the AI Innovation Cloud (for training and fine-tuning) and the Cirrascale Inference Platform (for model deployment). The AI Innovation Cloud provides access to multiple AI accelerators including NVIDIA GPUs (Blackwell B200/B300, Hopper H200/H100, RTX PRO 6000), AMD Instinct (MI325X, MI300X, MI250), Qualcomm Cloud AI 100, Tenstorrent Galaxy Blackhole, and Cerebras Wafer-Scale Clusters - the only cloud where customers can test and deploy on every leading AI accelerator. The Inference Platform enables enterprise deployment with intelligent accelerator selection and cross-region workload balancing. Additional offerings include Private Google Gemini for on-premises frontier AI, Ai2 Model Endpoints for open models, Google GPAR Services for research/government, Multi-GPU Compute, Cloud Storage Solutions (WEKA-based NVMe hot-tier, object storage), and Networking (InfiniBand). All services feature bare-metal delivery with no egress fees.

Differentiator

Problem solved

Functional benefit

Brands

• AI Innovation Cloud: Cloud service enabling testing and deployment on leading AI accelerators including NVIDIA, AMD, Qualcomm, and Tenstorrent GPUs.
• Cirrascale Inference Platform
• Private AI
• Cerebras AI Model Studio

Products and services

• AI Innovation Cloud Flagship bare-metal cloud platform providing dedicated AI training, fine-tuning, and inference infrastructure. The only cloud where customers can test and deploy across every leading AI accelerator (NVIDIA, AMD, Qualcomm, Tenstorrent, Cerebras) in one unified environment.
• Cirrascale Inference Platform Enterprise inference-as-a-service platform that automatically selects optimal AI accelerators for customer models, dynamically balances workloads across regions, and offers serverless deployments with RAG integration and token-based pricing.
• NVIDIA GPU Cloud AI Innovation Cloud offering featuring NVIDIA GPUs including Blackwell B200/B300, Hopper H200/H100, and RTX PRO 6000 for generative AI, LLM, and HPC workloads.
• AMD Instinct Series Cloud AI Innovation Cloud offering featuring AMD Instinct GPUs including MI325X, MI300X, and MI250 for high-performance AI training and inference.
• Qualcomm Cloud AI AI Innovation Cloud offering featuring Qualcomm Cloud AI 100 Pro and Ultra accelerators for cost-effective AI inference workloads with one-click deployment.
• Tenstorrent Galaxy Cloud AI Innovation Cloud offering featuring Tenstorrent Galaxy Blackhole servers engineered for real-world AI workloads with industry-leading performance at approximately half the cost of leading GPU alternatives.
• Private Google Gemini On-premises Gemini AI model deployment via Google Distributed Cloud through the Cirrascale Inference Platform. Enables public sector and enterprises to run advanced multimodal AI within their own infrastructure while maintaining data sovereignty, supporting connected and air-gapped environments.
• Qualcomm Inference Cloud Inference cloud powered by Qualcomm AI Inference Suite enabling one-click deployment of AI models with seamless model swapping and custom application building.
• Ai2 Model Endpoints Access to open AI models from the Allen Institute for AI including OLMo, Molmo, and Tulu models deployed on the Cirrascale Inference Platform.
• Cerebras AI Model Studio Cloud service for training GPT-class AI models (1.3B to 175B parameters) on Cerebras Wafer-Scale Clusters. Delivers 8x faster time to accuracy at half the cost of traditional cloud providers, supporting sequence lengths up to 50,000 tokens.
• Multi-GPU Compute Multi-accelerator cloud servers ranging from single-GPU development configurations to production/training applications in a cloud-hosted environment. Multi-GPU servers supporting NVIDIA RTX A4000, A5000, and A6000 professional GPUs available starting at $1,999 per 8-GPU server per month.

Quantifiable outcome

• Up to 15X faster real-time inference performance for multi-trillion-parameter models (GPT-MoE-1.8T) with NVIDIA HGX B200 compared to previous Hopper generation
• +7 more outcomes

Named customers4 records

Segments5 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration5 records

AI capability10 records

Feature8 records

Partnerships

15 partnerships are on record, tiered secondary and core.

• TenstorrentsecondaryTechnology or Integration · 4 May 2026Cirrascale added Tenstorrent Galaxy Blackhole servers to its AI Innovation Cloud, marking Tenstorrent's entry into broad commercial deployment. Servers engineered for real-world AI workloads with industry-leading performance at approximately half the cost of leading GPU alternatives.
• Google Distributed CloudcoreTechnology or Integration · 23 April 2026Cirrascale Inference Platform integrated with Google Distributed Cloud enabling public sector organizations and enterprises to run advanced multimodal, multilingual AI within their own infrastructure while maintaining data sovereignty.
• Google CloudcoreTechnology or Integration · 22 April 2026Cirrascale expanded partnership with Google Cloud to deliver Gemini AI models as fully private, disconnected on-premises appliance—the first neocloud provider to offer Google's most advanced AI model outside its own infrastructure. Built on Dell-manufactured hardware with eight Nvidia GPUs and confidential computing protections. Supports both connected and air-gapped environments via Google Distributed Cloud.
• Google Public SectorcoreGTM or Marketing Partner · 10 March 2026Partnership with Google Public Sector to deliver specialized research offerings and launch new Government Services Division. Google GPAR (Government and Public Sector Research) services partnership for secure AI infrastructure for research institutions and universities.
• Qualcomm TechnologiescoreTechnology or Integration · 6 March 2025Inference Cloud powered by Qualcomm AI Inference Suite and Qualcomm Cloud AI 100 Ultra accelerator. Previously launched Cloud AI Developer Playground. Offers one-click deployment of AI models and applications with efficient, scalable solutions.
• Cerebras SystemscoreStrategic or Co-development Partner · 29 November 2022Joint Cerebras AI Model Studio offering hosted on Cerebras Cloud @ Cirrascale. Enables customers to train GPT-class models (1.3B to 175B parameters) on Cerebras Wafer-Scale Clusters with 8x faster time to accuracy at half the cost of traditional cloud providers. First announced November 2022.
• NVIDIAcoreTechnology or IntegrationNVIDIA Partner Network Cloud Provider offering NVIDIA GPU Cloud services including HGX H100, H200, and B200 instances. Also NVIDIA Elite Cloud Services Provider for Visualization. Cirrascale deploys NVIDIA Quantum InfiniBand networking and offers NVIDIA RTX professional GPUs for visualization workflows.
• Dell TechnologiescoreTechnology or IntegrationDell Technologies Platinum Partner. Cirrascale deploys Dell's industry-leading storage and hardware technologies across its AI Innovation Cloud, providing customers with a reliable, high-performance foundation for demanding AI and deep learning workloads.
• WEKA (WekaIO)coreTechnology or IntegrationWEKA Data Platform certified as high-performance data store solution for NVIDIA Cloud Partners. WEKA's Matrix software provides fully parallel and distributed file system leveraging Flash technology for massively parallel NVMe access. Removes storage bottlenecks for inference and training datasets.
• Allen Institute for AI (Ai2)coreStrategic or Co-development PartnerAi2 Model Endpoints offering on Cirrascale platform. $152M project including $152M computing cluster managed by Cirrascale for Ai2, running Nvidia Blackwell Ultra chips. Supports open AI model development for scientific research in materials science, biology, and energy.
• IBM Power AIsecondaryTechnology or IntegrationIBM PowerAI Vision ecosystem partner. PowerAI Vision makes computer vision with deep learning more accessible with intuitive toolset for labeling, training, and deploying deep learning vision models without coding expertise.
• KubeflowsecondaryTechnology or IntegrationKubeflow ecosystem partner. Kubeflow project makes deployments of ML workflows on Kubernetes simple, portable, and scalable, enabling easy, repeatable deployments on diverse infrastructure.
• One Convergence (DKube)secondaryTechnology or IntegrationDKube from One Convergence ecosystem partner. Enables data scientists to find, download, and maintain ML frameworks, CUDA libraries, RDMA tools without extensive IT expertise. Installs on Cirrascale platform in under an hour.
• H2O AIsecondaryTechnology or IntegrationH2O AI ecosystem partner. H2O is open source, distributed in-memory ML platform with linear scalability supporting gradient boosted machines, generalized linear models, deep learning and AutoML functionality.
• CiscosecondaryTechnology or IntegrationMentioned in context of Secure AI Factory with NVIDIA. Cirrascale infrastructure potentially compatible with Cisco AI infrastructure solutions for edge deployments and enterprise AI production.

Scale indicators7 records

Recent moves7 records

Expansion highlights6 records

Direct peers

• Lambda Labs: GPU cloud provider specializing in NVIDIA-based AI training and inference infrastructure, with on-demand and reserved clusters comparable to Cirrascale's bare-metal AI Innovation Cloud offerings.
• Fluidstack: GPU cloud provider offering large-scale AI compute clusters and bare-metal GPU rental directly comparable to Cirrascale's AI Innovation Cloud multi-GPU server rentals.
• CoreWeave: GPU-focused neocloud built on bare-metal NVIDIA infrastructure serving AI training and inference customers. Closest direct competitor to Cirrascale in dedicated AI cloud services for enterprise and AI labs.
• Crusoe: AI-focused cloud provider running large-scale GPU clusters, often powered by stranded/clean energy. Competes with Cirrascale for large AI training workloads and GPU capacity deals.

Emerging players

• Together AI: AI cloud platform focused on open-model inference and training with multi-GPU clusters. Competes with Cirrascale Inference Platform for enterprise inference workloads and open-model deployments.
• RunPod: GPU cloud platform offering on-demand and serverless AI compute to developers and startups. Overlaps with Cirrascale's Inference Platform in serving flexible AI workloads with token/usage-based pricing.
• Vast.ai: Decentralized GPU marketplace aggregating third-party GPU capacity for AI training and inference. Targets similar AI developers as Cirrascale but via a marketplace rather than owned infrastructure.

Broad incumbents

• Paperspace (DigitalOcean): GPU-accelerated cloud now under DigitalOcean, targeting developers and AI/ML workloads with bare-metal and virtualized GPU instances comparable to Cirrascale's multi-GPU compute offerings.
• OVHcloud: European cloud provider with bare-metal and dedicated GPU servers targeting enterprise workloads. Provides broader geographic (European) coverage that Cirrascale lacks.
• IBM Cloud: Broader enterprise cloud and on-premises infrastructure provider with GPU offerings for AI/HPC. Competes with Cirrascale for regulated-industry and government AI deployments via hybrid deployment models.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/cirrascale
• Facebookfacebook.com/cirrascalecloudservices
• LinkedInlinkedin.com/company/cirrascale

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 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 Cirrascale Cloud Services LLC