What GPU.ai does

GPU.ai is a GPU cloud aggregation platform that sources NVIDIA hardware globally and routes customer workloads to the cheapest available provider with stock. It continuously polls inventory and pricing across providers in North America, Europe, and India on a 30-second cycle, normalizing rates to per-GPU-hour and automatically launching each job on the lowest-cost box matching the customer's specification. The platform is operated by BitCore AI Ltd, incorporated in the Dubai International Financial Centre, and serves AI/ML teams, research institutions, and enterprise AI operators with on-demand GPU compute, reserved capacity contracts, and custom supercluster buildouts.

The product portfolio comprises On-Demand GPUs billed per second, Reserved GPU Clusters with up to 60% discounts for multi-week to multi-year commitments, Custom GPU Buildouts for enterprise supercluster procurement, a Serverless Inference API (OpenAI-compatible chat, embeddings, image, and video generation), a GPU CLI, and a Supplier API that lets GPU providers join the marketplace by implementing three REST endpoints. Pricing is published transparently — for example H200 SXM at $3.81/GPU-hour, H100 SXM at $2.64, B200 at $5.29, and A100 80GB at $2.00, positioned at 48-65% below hyperscaler list rates. Architecture is a Go services layer with a Next.js frontend, FRP-based SSH tunneling for instance connectivity, and SOC 2 Type II compliant infrastructure.

Go-to-market combines product-led growth (self-serve CLI/API, $100 free trial without credit card) with consultative enterprise sales for reserved and custom-buildout engagements. Distribution is primarily developer-centric — engineering blog, API documentation, organic/community channels (Hacker News, Reddit, X), and event sponsorships such as the AGI Summit SF 2026 title sponsorship. Revenue streams include usage-based on-demand margin, reserved-instance subscription contracts, and project-based custom buildout services. Named customers span research/academia (Cornell, NYU, MIT, US Department of Energy) and AI-adjacent enterprises (Telegram, DeepMotion, DeepInfra, Exabits, Cocoon).

Name

GPU.ai

Legal name

BitCore AI Ltd

Website

https://gpu.ai

Company type

Private

Founded year

2026

Operating status

Operating

Headcount range

1–10 employees

Short description

GPU.ai is a GPU cloud aggregation platform that routes on-demand, reserved, and custom-built AI compute workloads to the cheapest available NVIDIA provider across North America, Europe, and India, serving AI/ML teams, research institutions, and enterprise AI operators with per-second billing and a self-serve CLI/API.

Ownership category

akta.pro rank

Product category

GPU Cloud Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (51821), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182)

SIC

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

akta.pro primary industry

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

akta.pro secondary industry

AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG)

Keywords

• GPU cloud services
• Cloud GPU aggregation
• AI compute infrastructure
• Serverless inference API
• GPU marketplace

Headquarters

HQ city

Dubai

HQ country

United Arab Emirates

HQ region

Middle East

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

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

Revenue model

1. GPU compute on-demand billing: Customers are billed per second for GPU compute usage. GPU.ai aggregates inventory from multiple providers and routes each launch to the cheapest available provider with stock. Revenue is generated as a margin on the per-GPU-hour rate charged to customers versus the cost paid to suppliers. No minimum commitments or hourly rounding.
2. Reserved GPU capacity contracts: Customers can reserve dedicated GPU capacity with guaranteed availability for weekly to multi-year commitment terms. Reserved instances offer discounted rates compared to on-demand pricing, with discounts of up to 60%. These contracts provide predictable recurring revenue and deeper customer lock-in.
3. Custom GPU buildouts and infrastructure services: GPU.ai offers end-to-end GPU infrastructure services including hardware procurement (NVIDIA B300, B200, H200, H100), dedicated bare-metal servers, deployment support, and AI infrastructure consulting. Revenue from buildouts is project-based and typically higher ACV enterprise engagements.

Pricing tiers

Go-to-market motion4 records

Distribution channels4 records

Marketing channels6 records

Core offering

GPU.ai operates a cloud GPU aggregation platform that continuously polls inventory and pricing across multiple GPU providers worldwide, then automatically routes each customer workload to the cheapest available provider with stock. The platform offers on-demand GPU instances billed per second, reserved GPU clusters with discounted contract pricing, custom enterprise GPU buildouts (bare-metal superclusters), and an OpenAI-compatible serverless inference API for chat, embeddings, image, and video generation.

Product overview

GPU.ai is a GPU cloud aggregation platform that aggregates GPU inventory from multiple cloud providers into a single unified interface. The core product portfolio includes: On-Demand GPUs for pay-per-second compute access with automatic best-price routing, Reserved GPU Clusters for guaranteed capacity at discounted rates, and Custom GPU Buildouts for enterprise hardware procurement and deployment. Supporting infrastructure includes the GPU CLI for command-line instance management, the Serverless Inference API for OpenAI-compatible LLM and media generation without instance management, the Supplier API enabling GPU providers to join the marketplace, and a Pricing Engine providing real-time competitive price comparison. The platform aggregates providers across North America, Europe, and India, routing workloads to the cheapest available box matching requirements. Per-second billing, no minimums, no lock-in.

Differentiator

Problem solved

Functional benefit

Products and services

• On-Demand GPUs Pay-per-second GPU cloud compute service aggregating inventory from providers worldwide. Deploys instances in under 60 seconds with automatic routing to the lowest-cost available provider. Supports H200, H100, B200, A100, L40S, RTX 4090, and other NVIDIA GPUs. Designed for AI/ML training, inference, fine-tuning, HPC, and rendering workloads.
• Reserved GPU Clusters Reserved GPU capacity service offering guaranteed availability and discounted rates (up to 60% off on-demand) compared to on-demand pricing. Commitment terms range from 1 week to 2+ years. Available for B200, B300, H200, H100, A100, and L40S GPUs with dedicated support and a named account engineer.
• Custom GPU Buildouts Enterprise GPU infrastructure procurement, deployment, and ongoing support service. Includes pre-designed supercluster configurations (GB300 NVL72, HGX H200, mixed multi-generation) and custom topology design for large-scale training workloads. Partners with NVIDIA, Supermicro, and NovaCore for hardware sourcing and installation.
• Serverless Inference API OpenAI-compatible serverless inference service for chat completions, embeddings, image generation, and video generation. No instance management required; billed per token or per output unit. Compatible with standard OpenAI SDKs and supports streaming responses.
• Supplier API REST API for GPU providers (small GPU farms, co-locations, regional operators) to list inventory, provision instances, and integrate with GPU.ai's aggregation platform. Providers implement three endpoints for inventory polling (every 30 seconds), provisioning, and lifecycle tracking, authenticated via OAuth2 client credentials flow.
• GPU CLI Command-line interface binary for GPU cloud management on macOS and Linux (amd64 and arm64). Enables launching instances, SSH connectivity via FRP tunneling, real-time pricing queries, serverless chat inference, and SSH key management from the terminal.

Quantifiable outcome

• 48–65% savings versus hyperscaler on-demand pricing (e.g., H200 SXM at $3.81/hr vs AWS at $4.97/hr and Azure at $12.99/hr)
• +3 more outcomes

Named customers9 records

Segments4 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability8 records

Feature7 records

Partnerships

Four partnerships are on record, tiered core and minor.

• NovaCore InnovationscoreStrategic or Co-development Partner · 3 September 2025Mumbai-based NovaCore Innovations and New York-based GPU.ai announced a strategic partnership to provide AI teams and enterprises seamless access to advanced GPU resources across the United States and India. The collaboration combines GPU.ai's U.S. east-coast cloud infrastructure offering A100, H100, and H200 instances with NovaCore's Hyderabad Blackwell cluster for worldwide remote deployment. A 'Builder's Express' program offers complimentary compute credits to select early-stage startups and research teams. NovaCore's Blackwell GB200 NVL72 racks appear in GPU.ai's availability feed alongside existing supplier inventory, and NovaCore tenants gain one-click access to GPU.ai's aggregated U.S. inventory.
• NVIDIAcoreTechnology or IntegrationGPU.ai partners with NVIDIA for GPU procurement. The Custom Buildouts page lists NVIDIA as a partner for accessing B300, B200, H200, and H100 GPUs with volume pricing and vendor-negotiated discounts. NovaCore's Hyderabad cluster, integrated into GPU.ai, is explicitly powered by NVIDIA Blackwell hardware.
• NovaCorecoreTechnology or IntegrationNovaCore is listed as a buildout partner on GPU.ai's Custom Buildouts page. NovaCore is India's first GPU cloud with NVIDIA Blackwell, co-founded by the same founders as GPU.ai. Their Hyderabad Blackwell cluster is integrated into GPU.ai's availability feed for worldwide deployment.
• AGI Summit SF 2026minorGTM or Marketing PartnerGPU.ai is the Official Title Sponsor of AGI Summit SF 2026 (July 18–19, 2026, Palace of Fine Arts, San Francisco). The two-day summit is expected to draw more than 15,000 attendees. GPU.ai will offer free GPU credits to attendees as part of its sponsorship initiative. Speakers and representatives from OpenAI, Anthropic, Microsoft, BlackRock, Tesla, and AWS are featured at the summit.

Scale indicators5 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Vast.ai: GPU marketplace that aggregates consumer and prosumer GPUs from hosts worldwide and rents them to AI/ML customers. Closest comparable to GPU.ai's aggregation model, Supplier API, and per-unit pricing.
• CoreWeave: Specialized GPU cloud provider built on NVIDIA hardware for AI training and inference. Direct competitor on H100/H200/B200 SKUs and on the price comparison page GPU.ai publishes against.
• Lambda: GPU cloud and on-prem AI infrastructure provider serving AI/ML teams with H100/A100 clusters. Comparable customer base (AI labs, research, enterprises) and overlapping SKU set.
• Crusoe: GPU cloud provider using stranded/renewable energy, focused on large-scale AI training clusters. Comparable on NVIDIA H100/B200 offerings and enterprise training workload positioning.
• RunPod: GPU cloud offering on-demand and reserved instances plus serverless endpoints for AI workloads. Comparable on per-second billing, multi-region inventory, and AI developer target market.
• Together AI: GPU cloud plus serverless inference platform for open and frontier models. Comparable because it bundles raw GPU access with OpenAI-compatible inference APIs, the same combination GPU.ai ships.
• Paperspace (DigitalOcean): GPU cloud for AI/ML training and inference, now part of DigitalOcean. Comparable target customer (developers and small AI teams) and overlapping GPU SKUs.

Broad incumbents

• Microsoft Azure (ND/NDv-series VMs): Hyperscaler GPU VM offering listed on GPU.ai's competitor pricing comparison. Overlapping H100/B200 SKUs but full enterprise cloud stack rather than aggregation focus.
• AWS EC2 (P-family GPU instances): Hyperscaler GPU compute offering used as the public benchmark on GPU.ai's pricing page. Comparable SKUs (H100, H200) but with much broader portfolio and no aggregation model.
• Google Cloud (A3 instances / TPU): Hyperscaler GPU/TPU offering for AI training and inference. Comparable on large-model workload targets and on enterprise procurement, but not specialized as a low-cost aggregator.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks7 records

Key highlights7 records

Customer concentration

Compliance1 record

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 GPU.ai