CloudRift
CloudRift provides hardware-agnostic GPU orchestration and AI infrastructure for sovereign AI deployments, serving datacenter operators, telcos, and regulated enterprises across 5 countries with on-demand GPU rentals, enterprise orchestration software, and OpenAI-compatible inference APIs.
- Company typePrivate
- Founded2024
- HeadquartersSan Francisco, United States
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What CloudRift does
CloudRift is a GPU orchestration and AI infrastructure platform founded in 2024 and headquartered in Santa Clara, California, built on QEMU/KVM open-source virtualization. The platform provides a hardware-agnostic control plane supporting NVIDIA (RTX 4090/5090/PRO 6000, L40S, H100/H200, B200) and AMD (MI350X via ROCm) GPUs across VM, container, MIG, and bare-metal isolation modes. Its product portfolio centers on three offerings: on-demand GPU rentals (console.cloudrift.ai, hourly billing, no minimum commitment), an Enterprise GPU Orchestration Platform for datacenter operators and regulated enterprises, and an LLM-as-a-Service Inference API with OpenAI-compatible endpoints and pay-per-token pricing.
The company positions itself as infrastructure for sovereign AI deployments, targeting datacenter operators, telecom providers, and regulated industries (banks, governments) that require data sovereignty, air-gapped operation, and on-premise control. CloudRift serves 8+ sovereign AI operators across five countries (USA, Canada, UK, Italy, Taiwan) with Central Asia (Kazakhstan) also represented, aggregating 3,400+ GPUs across 12 global data center locations. Pricing is usage-based — per-hour for GPU rentals with 10–15% discounts for 1–3 month reservations, and per-token for inference — with no egress or API-call fees. Revenue also flows from platform licensing to datacenter operators who monetize their own GPU capacity and set their own pricing, with CloudRift taking a platform fee or revenue share.
The business operates a hybrid go-to-market combining product-led growth (self-serve console for individual developers and researchers) with enterprise field sales (demo-driven procurement for datacenter operators and regulated industries). CloudRift is led by three co-founders — Dmitry Trifonov (CEO, ex-Apple/Roblox/HP), Slawomir Strumecki (Systems Lead, ex-Roblox/Ubisoft), and Dimitrios Verraros (Cloud Lead, ex-Apple) — and is pursuing SOC 2 Type II certification via Vanta. The company raised a $2.75 million seed round from Surface Ventures in March 2025 and is currently building out engineering, marketing, and business development functions to scale from a founding team to a multi-functional organization.
CloudRift firmographics
Firmographics- Name
- CloudRift
- Legal name
- CloudRift, Inc.
- Website
- https://cloudrift.ai
- Company type
- Private
- Founded year
- 2024
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- CloudRift provides hardware-agnostic GPU orchestration and AI infrastructure for sovereign AI deployments, serving datacenter operators, telcos, and regulated enterprises across 5 countries with on-demand GPU rentals, enterprise orchestration software, and OpenAI-compatible inference APIs.
- Ownership category
- akta.pro rank
CloudRift industry classification
Industry- Product category
- GPU Cloud Infrastructure
- NAICS
- Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518), Computer Systems Design and Related Services (54151)
- SIC
- Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Integrated Systems Design (7373), Services-Computer Rental & Leasing (7377)
- akta.pro primary industry
- AI Compute Cloud & GPU-as-a-Service (HDAAAAAK)
- akta.pro secondary industries
- AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG), Private Cloud Platforms (On‑Prem Cloud Stacks) (HDABABAA), Hybrid Cloud Compute & Virtualization Stack (HCI/VMs/Kubernetes) (HDABAMAF), Cloud Storage Gateways & Hybrid Storage (HDABAEAI)
Keywords
Where CloudRift is headquartered
LocationHeadquarters
- HQ city
- San Francisco
- HQ country
- United States
- HQ region
- North America
Offices2 records
Markets served
CloudRift business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- GPU Instance Rentals (On-Demand): Hourly billing for GPU compute instances. Users rent NVIDIA RTX 4090/5090/PRO 6000, L40S, V100, H100/H200, B200, and AMD MI350X GPUs by the hour with no minimum commitment. Billing is per-second of active runtime. On-demand pricing with no discounts; reserved 1-month and 3-month commitments offer 10-15% discounts.
- LLM Inference (Pay-per-Token): Serverless LLM inference with OpenAI-compatible API. Users pay per million tokens for input and output. Models include Qwen3.6-35B-A3B-FP8 at $0.15/M input tokens and $1.00/M output tokens. No idle GPU costs — users only pay for actual token generation.
- Enterprise Platform / Datacenter Management: Platform licensing for datacenter operators and enterprises to manage GPU fleets. Operators monetize their GPU capacity through the platform, set their own pricing, and own customer relationships. CloudRift provides the infrastructure layer while operators retain control over branding and commercial terms.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Usage-based | Pay-as-you-go | NVIDIA V100 SXM3 32GB — entry-level datacenter GPU |
| Usage-based | Pay-as-you-go | NVIDIA RTX 4090 24GB — cost-efficient AI training and fine-tuning |
| Usage-based | Pay-as-you-go | NVIDIA L40S 48GB — datacenter GPU for inference and vGPU |
| Usage-based | Pay-as-you-go | NVIDIA RTX 5090 32GB — Blackwell consumer GPU for AI workloads |
| Usage-based | Pay-as-you-go | NVIDIA RTX PRO 6000 96GB — Blackwell enterprise workstation GPU |
| Usage-based | Pay-as-you-go | NVIDIA RTX PRO 6000 Max-Q 96GB — power-optimized variant |
| Usage-based | Multi-year contract | NVIDIA H100 SXM 80GB — datacenter Hopper GPU (reserved only) |
| Usage-based | Multi-year contract | NVIDIA H200 SXM 141GB — high-memory datacenter Hopper GPU (reserved only) |
| Usage-based | Pay-as-you-go | AMD Instinct MI350X 288GB — CDNA 4 datacenter AI accelerator |
| Usage-based | Multi-year contract | NVIDIA B200 192GB — Blackwell datacenter GPU (reserved only) |
| Usage-based | Pay-as-you-go | LLM Inference — Qwen3.6-35B-A3B-FP8 (pay-per-token) |
Go-to-market motion3 records
Distribution channels4 records
Marketing channels7 records
CloudRift product offering
Product offeringCore offering
CloudRift provides a GPU orchestration and AI operations platform that enables enterprises, datacenter operators, and telcos to manage large-scale GPU infrastructure within their own security perimeter. The company sells on-demand hourly GPU rentals across NVIDIA and AMD hardware, an OpenAI-compatible LLM inference API with pay-per-token billing, and an enterprise platform license that includes a unified control plane, white-label console, persistent multi-datacenter storage, and full virtualization (VMs, containers, bare metal).
Product overview
CloudRift is a sovereign AI platform consisting of a unified core product portfolio: the Enterprise GPU Orchestration Platform (the AI ops layer for managing GPU infrastructure), GPU Rentals (on-demand hourly access to NVIDIA and AMD GPUs), and LLM-as-a-Service Inference API (OpenAI-compatible inference with pay-per-token pricing). These core offerings are supported by dedicated workload products including Image & Video Generation Infrastructure, Multimodal Search Infrastructure, and Game Development GPU Infrastructure, plus operational add-ons including White-Label Console for datacenter branding, AI Grant Program for developer credits, and Bug Bounty Program for community security research. The platform powers 8+ sovereign AI operators across 5 countries, enabling datacenters, telcos, and enterprises to run AI on their own infrastructure with full control over data, pricing, and customer relationships.
Differentiator
Problem solved
Functional benefit
Products and services
- GPU Rentals On-demand GPU rental platform offering NVIDIA RTX 4090/5090/RTX PRO 6000, L40S, H100/H200, B200, V100, and AMD MI350X GPUs billed by the hour. Supports VM, container, and bare-metal deployment modes with no minimum commitment and per-second billing. Targeted at developers, researchers, and AI practitioners who need instant access to GPU compute.
- Enterprise GPU Orchestration Platform AI ops layer for enterprise datacenters providing a single control plane for managing GPU clusters, tenants, and workloads across multiple data centers. Supports VMs, containers, MIG, bare metal, and OpenAI-compatible inference with RBAC, quotas, and audit logging. Designed for datacenter operators, telcos, and regulated enterprises that need to run AI on their own infrastructure.
- LLM-as-a-Service Inference API Serverless LLM inference with OpenAI-compatible API supporting Llama, DeepSeek, Qwen, Mistral, GLM, and Kimi models. Pay-per-token pricing with autoscaling, no idle GPU costs, and drop-in replacement for existing OpenAI-based code. Targeted at AI developers and enterprises needing sovereign inference without managing GPU infrastructure.
- Image & Video Generation Infrastructure GPU infrastructure offering tailored for image and video generation workflows, supporting ComfyUI, Invoke, FLUX, SDXL, and video models (LTX-Video, Mochi, Hunyuan Video, Wan). Provides persistent storage for models and checkpoints. Targeted at creative teams and AI practitioners building image and video generation pipelines.
- Multimodal Search Infrastructure Embedding and reranking infrastructure for multimodal search, supporting open embedding models for images and text with vector indexing and low-latency search pipelines on GPU-accelerated infrastructure. Targeted at teams building semantic search, RAG, and visual retrieval applications.
- Game Development GPU Infrastructure GPU compute offering for game studios covering rendering, lightmap baking, automated QA, CI/CD GPU runners, NPC dialogue research, and AI-assisted asset creation. Includes persistent storage for assets and checkpoints. Targeted at game development studios needing burst GPU capacity for dev/test environments.
Quantifiable outcome
- Time to first internal workload: Days (vs 12-24 months for in-house build, 6-12 months for hyperscaler private cloud)
- +4 more outcomes
Companies that use CloudRift
Customer profileNamed customers11 records
Segments4 records
Ideal customer profiles3 records
CloudRift technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration9 records
AI capability10 records
Feature6 records
CloudRift partnerships and signals
Strategic signalPartnerships
Five partnerships are on record, tiered core, major and minor.
- NVIDIAcoreNVIDIA Inception program member with full support for NVIDIA AI Enterprise (MIG, vGPU), RTX consumer GPUs (4090/5090/PRO 6000), and datacenter GPUs (H100/H200/B200). Platform built to leverage CUDA ecosystem and NVIDIA virtualization stack alongside AMD alternatives.
- NeuralRackcoreProfessionally managed high-capacity GPU provider operating CloudRift-powered infrastructure. Tier 3 ISO 27001, SOC II type 2, and PCI-DSS compliant facilities with up to 100Gbps network. Customizable hardware and proactive on-site support. One of the primary GPU rental providers on the CloudRift platform.
- HyperCloudmajorSubsidiary of a major telecom provider across Central Asia, Europe, and the Middle East. Multi-region data centers with enterprise-grade GPU capacity. Serving regulated industries including banking and government. One of the flagship sovereign AI operators on the CloudRift platform.
- Kazteleportmajor25+ years operating telecom and data center infrastructure in Central Asia. 5 TIER III certified data centers with 500+ engineers on staff. Flagship production deployment with Halyk Bank (largest bank in Central Asia) as an enterprise customer.
- KonstminorAI datacenter operator with 8 data center locations across Taiwan, Japan, Singapore, USA, and Europe. ISO 27001 certified with H100, H200, and RTX 5090 GPU fleet. Delivers turnkey AI data centers in 4-6 months at a third of hyperscaler cost.
Scale indicators6 records
Recent moves6 records
Expansion highlights7 records
CloudRift competitors and assessment
Company assessmentEmerging players
- Modal Labs: Serverless GPU compute platform for AI/ML workloads with Python-native deployment and per-second billing. Overlaps CloudRift's serverless inference and developer-focused GPU experiences but without an on-prem/sovereign story.
Direct peers
- Fluidstack: GPU cloud provider offering large-scale H100/H200 rentals and reserved capacity to AI labs and enterprises, often via partnerships with hyperscalers. Competes for enterprise AI training and inference capacity buyers.
- Crusoe: Vertically integrated GPU cloud operator using stranded energy for large-scale NVIDIA clusters, selling H100/H200 capacity to AI startups and enterprises. Competes on raw GPU capacity at hyperscaler-adjacent scale.
- CoreWeave: Largest independent GPU cloud provider offering NVIDIA H100/H200/B200 on-demand rentals and a managed Kubernetes platform. Closest direct competitor to CloudRift's GPU rental business, though CoreWeave targets hyperscaler-grade enterprise workloads rather than sovereign on-prem deployments.
- RunPod: GPU cloud offering on-demand and serverless GPU rentals (including H100/H200) with container and VM deployment. Direct competitor on self-serve AI developer rentals with similar pay-per-hour and per-second billing.
- Vast.ai: Marketplace-style GPU rental platform aggregating consumer and datacenter GPUs at low hourly prices. Competes with CloudRift on cost-sensitive, self-serve GPU rentals for AI developers.
- Together AI: GPU cloud plus hosted open-model inference API with OpenAI-compatible endpoints and per-token pricing. Directly overlaps CloudRift's LLM-as-a-Service offering and its pay-per-token inference model.
- Lambda Labs: GPU cloud and deep-learning workstation provider with on-demand H100 rentals and cloud instances aimed at AI researchers and startups. Comparable to CloudRift's self-serve GPU rental motion and developer-led GTM.
Broad incumbents
- Paperspace (DigitalOcean): DigitalOcean's GPU cloud arm offering on-demand GPU VMs and notebooks. Competes in the broader self-serve AI developer rental market though without CloudRift's on-prem sovereignty positioning.
- Red Hat OpenShift AI: Enterprise hybrid-cloud AI platform from Red Hat/IBM for deploying and managing AI workloads on-prem and across clouds. Competes for the same enterprise GPU orchestration buyer that CloudRift targets with its sovereign-AI platform, though OpenShift is a broader Kubernetes stack rather than GPU-specialized.
Market position
Strengths4 records
Weaknesses4 records
Competitive moat5 records
Key risks5 records
Key highlights6 records
Customer concentration
CloudRift social profiles
Digital presenceCloudRift compliance and trust
Trust signalCompliance2 records
CloudRift financial estimates
Financial estimateRevenue estimate
Valuation estimate
CloudRift leadership team
Management profileNumber of profiles
Profiles3 records
CloudRift funding detail
Funding detailFunding overview
Funding rounds1 record
Investors1 record
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
CloudRift M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about CloudRift
What does CloudRift do?
CloudRift provides a GPU orchestration and AI operations platform that enables enterprises, datacenter operators, and telcos to manage large-scale GPU infrastructure within their own security perimeter. The company sells on-demand hourly GPU rentals across NVIDIA and AMD hardware, an OpenAI-compatible LLM inference API with pay-per-token billing, and an enterprise platform license that includes a unified control plane, white-label console, persistent multi-datacenter storage, and full virtualization (VMs, containers, bare metal).
Is CloudRift a public or private company?
CloudRift is a private company. It is classified as venture growth investor backed and is currently operating.
When was CloudRift founded?
CloudRift was founded in 2024. It employs 1 to 10 people.
Where is CloudRift based?
CloudRift is headquartered in San Francisco, United States, in the North America region.
How does CloudRift make money?
Three revenue lines are on record. GPU Instance Rentals (On-Demand) is the primary driver. The others are LLM Inference (Pay-per-Token) and enterprise Platform / Datacenter Management.
Who are CloudRift's main competitors?
Modal Labs is listed as an emerging player. Direct peers are Fluidstack, Crusoe, CoreWeave, RunPod, Vast.ai, Together AI and Lambda Labs. Broad incumbents are Paperspace (DigitalOcean) and Red Hat OpenShift AI.
Does CloudRift have an API?
Yes. CloudRift provides REST APIs and OpenAI-compatible inference endpoints. The REST API enables programmatic control over GPU infrastructure — automate deployments, manage instances, volumes, users, and billing. The OpenAI-compatible API serves LLM inference with drop-in replacement capability (switch base URL and keep existing code), supporting models like Llama, DeepSeek, Qwen, and Mistral. The inference API is accessible at https://inference.cloudrift.ai/v1 with per-token billing. Developer documentation is at docs.cloudrift.ai.
What industry is CloudRift in?
CloudRift's product category is GPU Cloud Infrastructure. Its primary akta.pro industry code is HDAAAAAK, AI Compute Cloud & GPU-as-a-Service, with a secondary code of HDAAAAAG, AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers). Its NAICS code is 5182 and its SIC code is 7370.