What TensorDock does

TensorDock is a GPU cloud marketplace founded in 2019 and incorporated in Delaware as TensorDock.com, Inc. The company operates a horizontal marketplace that aggregates supply from independent hosts who install TensorDock's Libvirt-based hypervisor and set their own per-resource pricing; TensorDock takes a 25% revenue-share fee on host earnings and connects end customers to this supply via a self-serve web dashboard and a REST API v2. The platform spans 100+ locations across 20+ countries with up to 30,000 GPUs reservable through close partners and roughly 1,000+ GPUs available for immediate deployment, offering 45 GPU models from consumer cards (RTX 4090 from $0.35/hr) through to enterprise H100 SXM5 (from $2.25/hr) and H100 hostnode bare metal.

The technical stack is built on KVM virtualization with GPU passthrough, providing dedicated GPU access with root control, custom driver management, Windows 10 support, Docker pre-installed on all VM templates, and sub-30-second deployment via a multithreaded provisioning stack. Core products include GPU Cloud (flagship), CPU Cloud (AMD EPYC / Intel Xeon for HPC), Instant VMs (pre-configured fast-launch beta), a Whitelabel GPU Storefront licensing program, and a GPU Hosting Program for supply-side participants. Pricing is per-second usage-based with no ingress/egress fees, no quotas, and a $1 minimum balance; long-term reserved contracts (monthly, annual, 3-year) are sold through an enterprise field-sales motion for compliance-bound workloads.

Customer base is heavily weighted toward AI/ML developers and researchers (LLM training/inference, diffusion models, GANs), creative and rendering professionals, HPC and academic users, and cloud gaming platforms; selected logos include airgpu, ELBO, dstack, Runware, Livepeer, Creavite, and Florida State University. Strategic supply partnerships with Voltage Park (H100 SXM5 clusters at Evoque Dallas) and payment infrastructure built on Stripe Connect underpin scale. Certifications at partner data centers include SOC 2 Type II, PCI-DSS, HIPAA, and ISO 27001 options.

Name

TensorDock

Legal name

TensorDock.com, Inc.

Website

https://tensordock.com

Company type

Private

Founded year

2019

Operating status

Operating

Headcount range

1–10 employees

Short description

TensorDock is a GPU cloud marketplace connecting customers to independent hosts across 100+ locations, offering 45 GPU models from consumer cards to H100 SXM5 at up to 80% below hyperscaler pricing for AI, rendering, HPC, and cloud gaming workloads.

Ownership category

akta.pro rank

Product category

Cloud GPU Infrastructure

NAICS

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

SIC

Services-Miscellaneous Equipment Rental & Leasing (7350)

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), GPU-Accelerated & AI Training/Inference Servers (HDACABAG)

Keywords

• GPU cloud marketplace
• Cloud GPU rental
• AI infrastructure
• On-demand GPUs
• Virtual machine hosting

Headquarters

HQ city

Boston

HQ country

United States

HQ region

North America

Offices6 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

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

Revenue model

1. GPU Cloud Services (On-Demand): Usage-based revenue from renting GPU virtual machines to customers. Customers deposit funds and are billed per second. Hosts set their own competitive prices in the marketplace; TensorDock earns through its margin on the difference.
2. Host Revenue Share Fee: TensorDock takes a 25% revenue-share fee from host earnings. Hosts set their own a la carte per-resource pricing; TensorDock monitors usage, credits host accounts, and collects its platform fee.
3. Reserved/Long-Term GPU Contracts: For customers needing long-term GPU capacity, TensorDock offers reserved pricing at discounted rates (monthly, annual, 3-year terms). Revenue recognized as services are delivered over contracted periods.
4. Whitelabel GPU Storefront Licensing: Whitelabel service offered to select hosts and organizations. Organizations set up branded GPU storefronts powered by TensorDock infrastructure and take a commission on sales.
5. CPU Cloud Services: CPU-only virtual machines for scientific computing, HPC, transcoding, and batch processing workloads. Revenue generated through on-demand usage of Intel Xeon and AMD EPYC processor instances.

Pricing tiers

Go-to-market motion3 records

Distribution channels5 records

Marketing channels6 records

Core offering

TensorDock operates a global GPU cloud marketplace that aggregates supply from independent hosts to rent virtual machines equipped with 45+ GPU models (including NVIDIA H100 SXM5, A100, RTX 4090) and CPU instances to customers worldwide. The platform enables on-demand deployment of bare-metal KVM virtual machines with dedicated GPU passthrough, billed per second with no quotas or commitments at prices up to 80% lower than hyperscalers. Revenue is generated through per-second compute usage, marketplace commissions on host earnings, reserved long-term contracts, and a whitelabel storefront licensing program.

Product overview

TensorDock is a GPU cloud marketplace offering three core product lines: GPU Cloud (45+ GPU models including H100, A100, RTX 4090 for AI/ML training and inference), CPU Cloud (AMD EPYC and Intel Xeon processors for HPC workloads), and Instant VMs (pre-configured fast-deploy virtual machines). Additional offerings include a RESTful API for programmatic infrastructure management, a Whitelabel program for partners to create branded storefronts, and a Hosting Program enabling GPU owners to monetize hardware. The platform operates a marketplace model where hosts compete on pricing, with TensorDock taking a 25% revenue-share fee.

Differentiator

Problem solved

Functional benefit

Products and services

• GPU Cloud GPU cloud infrastructure service offering 45 different GPU models (H100 SXM5, A100 SXM4/PCIe, RTX 4090, RTX 3090, L40, V100, RTX 6000 Ada, RTX A6000, RTX A4000) for deep learning, AI inference, rendering, and gaming workloads at up to 80% lower cost than hyperscalers.
• CPU Cloud CPU cloud service providing Intel Xeon Scalable and AMD EPYC Milan processors for scientific computing, HPC, transcoding, and batch processing workloads with a la carte per-resource billing.
• TensorDock API RESTful API (v2) enabling programmatic management of virtual machines, resource monitoring, secrets management, billing, hostnode discovery, and integration with existing infrastructure. Bearer-token authentication, rate limited at 100 requests/min per organization.
• Whitelabel GPU Storefront White-label service enabling select partners and organizations to create branded GPU storefronts powered by TensorDock infrastructure, taking a commission on sales made through their branded portal.
• GPU Hosting Program Program that allows GPU owners to monetize idle hardware by listing servers on TensorDock's marketplace. Hosts install TensorDock's Libvirt-based hypervisor, set their own competitive prices, and earn revenue from customers deploying VMs on their hardware. TensorDock takes a 25% revenue-share fee.

Quantifiable outcome

• H100 SXM5 at $2.25/hr delivers 80% cost savings vs. AWS ($98.32/hr) and Google Cloud ($88.49/hr) for equivalent GPU configurations
• +4 more outcomes

Named customers7 records

Segments5 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability6 records

Feature7 records

Partnerships

Three partnerships are on record, tiered core and flagship.

• NVIDIAcoreTechnology or IntegrationTensorDock's entire cloud platform runs on NVIDIA GPUs including H100 SXM5, A100 SXM4/PCIe, RTX 4090, RTX 3090, RTX 6000 Ada, RTX A6000, L40, and V100. The partnership enables TensorDock to offer the industry's most cost-effective GPU cloud for deep learning, AI training, and inference workloads.
• StripecoreTechnology or IntegrationStripe Connect is TensorDock's payment infrastructure for host payouts. Hosts link their bank accounts via Stripe Connect to receive automatic withdrawals of their earnings. TensorDock covers withdrawal fees for all withdrawals above $10,000.
• Voltage ParkflagshipStrategic or Co-development PartnerVoltage Park supplies thousands of NVIDIA H100 SXM5 GPUs to TensorDock on-demand, enabling TensorDock to offer the industry's lowest H100 on-demand pricing at $2.25/hr. Voltage Park's Dallas data center hosts these clusters with enterprise-grade SLAs.

Scale indicators9 records

Recent moves6 records

Expansion highlights5 records

Direct peers

• RunPod: RunPod is a GPU cloud platform offering on-demand and serverless GPU compute for AI training, inference, and inference endpoints. Competes directly with TensorDock on developer-led AI workloads, pricing, and product-led growth distribution.
• Voltage Park: Voltage Park is a GPU cloud provider with thousands of NVIDIA H100 GPUs and a flagship strategic partnership supplying TensorDock. While a partner, Voltage Park also sells H100 capacity directly to enterprise customers, making it a partial direct competitor.
• Vast.ai: Vast.ai operates a similar GPU marketplace model aggregating independent hosts to offer on-demand GPU rentals at prices below hyperscalers. Directly comparable to TensorDock on marketplace structure, target customer (AI developers), and pricing positioning.

Broad incumbents

• Google Cloud (A2 instances): Google Cloud offers A2 instances with A100 GPUs for AI/ML workloads. Functions as a high-priced incumbent against which TensorDock's value proposition is explicitly defined (e.g., $88.49/hr H100 comparison).
• AWS EC2 (P5/P4 instances): AWS provides GPU instance types (P5 with H100, P4 with A100) at premium pricing against which TensorDock positions its 80% cost advantage. The dominant incumbent in the segment TensorDock targets for cost displacement.
• Azure (ND H100 v5): Azure's ND H100 v5 series provides enterprise-grade H100 GPU instances for AI training and inference. One of the three hyperscalers against which TensorDock competes on price, while offering narrower enterprise compliance depth.
• Paperspace (DigitalOcean): Paperspace (acquired by DigitalOcean) provides GPU-accelerated cloud computing for AI/ML and rendering. Offers a similar developer-focused experience but with a vertically integrated infrastructure model rather than a marketplace of independent hosts.
• CoreWeave: CoreWeave is a large-scale, well-capitalized GPU cloud provider specializing in AI workloads with thousands of NVIDIA H100/A100 GPUs. While serving the same AI/ML customer base, CoreWeave owns its infrastructure at scale and focuses on enterprise and hyperscale customers rather than a marketplace model.
• Lambda Labs: Lambda is a GPU cloud and deep learning hardware company offering reserved and on-demand GPU instances for AI training and inference. Targets similar customers but operates owned infrastructure (Lambda Hyperplane) rather than a host-aggregated marketplace.

Emerging players

• Together AI: Together AI is an AI cloud platform providing GPU-backed compute for training and inference, with a focus on open-source models. While it operates its own infrastructure, it targets a similar buyer (AI developers seeking lower-cost alternatives to hyperscalers) with higher-level abstractions.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/tensordock

Compliance4 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds1 record

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 TensorDock