Connecting GPUs. Delivering AI compute.

WebCal is a hybrid distributed GPU cloud and AI compute marketplace.

We bring together first-party data centers, independent GPU owners and professional compute providers. Through a shared platform for resource management, scheduling and service delivery, this capacity becomes compute that enterprises and developers can use for real AI workloads.

Explore how the platform works

From available hardware to usable compute.

Different sources bring different capabilities. WebCal connects supply with workload requirements, bringing resource assessment, execution and usage records into one service workflow.

Select a resource source to understand its role.

A directly operated infrastructure base for workloads that need a more controlled environment.

Managed through WebCal

Assess & onboard
Review hardware specifications, network conditions and operating capabilities before including a resource in the available supply.
Match the workload
Consider GPU memory, capacity, location, availability and isolation requirements when selecting a suitable environment.
Run & deliver
Provision the agreed resources or execution environment, then manage the workload within the service scope.
Measure & settle
Use service records and the agreed billing rules to account for usage, customer charges and provider settlement.

Delivered as

GPU rental
Dedicated compute
AI workloads

Three sources. Different strengths.

The platform combines a controllable infrastructure base with additional distributed capacity. Resources are assessed for their intended use; they are not treated as interchangeable machines.

First-party data centers

WebCal’s own data-center resources form the foundation of its compute supply. Direct management of hardware, networking and operations provides a basis for reserved capacity, dedicated servers and tailored deployments.

Typical fit: Long-running enterprise workloads, dedicated servers and projects with specific network or operating requirements.

Independent GPU owners

Individuals and teams with available GPU equipment can explore supplying capacity to the platform. Suitability depends on the hardware, connection quality, availability and service requirements. Onboarding and task allocation follow the assessment and cooperation terms.

Typical fit: Compatible workloads that can use distributed capacity within agreed availability and data-handling boundaries.

Professional compute providers

Professional providers contribute additional GPU configurations, deployment locations and operating capacity. Their resources are evaluated against the requirements of the service, including network connectivity and ongoing support.

Typical fit: Additional capacity, specific hardware configurations and deployments requiring a suitable provider location.

Connecting those who need compute with those who supply it.

Customers bring workloads. Providers bring equipment and operating capacity. WebCal coordinates the resources and service workflow between them.

For teams building with AI

Enterprises, model teams, researchers and individual developers can start with the workload they need to run, rather than assembling every part of the infrastructure themselves.

Define your model size, expected usage, data requirements and delivery timeline. These requirements guide the choice between GPU rental, dedicated infrastructure and a managed service.

Explore GPU rental

For compute providers

GPU owners, data centers and professional suppliers can discuss making suitable capacity available through WebCal.

Cooperation starts with resource specifications, availability and operating responsibilities. Allocation depends on suitability and customer demand; settlement follows actual service records and the agreed terms.

Discuss supplying compute

Choose the service around the workload.

Some projects need access to hardware; others need an environment for a model or a defined computing task. The delivery model follows what you need to run.

GPU servers & compute instances: Access GPU resources for development, experimentation and ongoing computation. Consider memory, configuration, network and availability requirements; the rental page lists the configurations and terms currently offered.
Dedicated infrastructure: Discuss reserved servers, bare-metal deployments or dedicated clusters for sustained workloads. Enterprise requirements such as networking, storage, isolation and support are scoped with the team before delivery.
Model training, fine-tuning & inference: Use compute for training or adapting models and for serving model requests. Deployment environments, API access and operational support depend on the selected service and its supported capabilities.
Generation, batch & scientific workloads: Image, video, speech and other compute-intensive tasks need an environment suited to their software and hardware requirements. Consider task size, processing time, storage and data transfer together.

How services are charged

Charges follow the selected product or agreement, including the applicable usage measure, duration and service scope. Review pricing, resource specifications and settlement terms before ordering. Provider settlement follows agreed service records and cooperation terms; connecting equipment alone does not imply a fixed return.

Clear delivery starts with clear responsibilities.

Hardware alone does not define a service. Workload fit, data handling and operational responsibilities must be considered together.

Resource suitability: Different GPUs and network environments suit different workloads. Memory, software compatibility, connection quality and availability inform resource selection. A node’s presence in the network does not make it suitable for every task.
Data & isolation requirements: Projects involving sensitive data need an appropriate deployment environment and agreed access boundaries. Specify isolation, storage location and data-handling requirements before choosing a service, especially when considering distributed resources.
Operations & support: Delivery scope, support arrangements and any service-level commitments should be set out in the selected product or enterprise agreement. For dedicated projects, establish operating responsibilities, the incident process and capacity needs with the team.

A few things you may want to know.

Do I need to own a GPU to use WebCal?

No. Customers can use the compute services offered by the platform. Supplying GPU equipment is a separate cooperation path for resource owners and professional providers.

Will every workload run on a distributed node?

No. Resource choice depends on the workload and the service agreement. Requirements around isolation, availability, networking or data handling may call for a dedicated or more controlled environment.

Where should I start with a custom requirement?

Prepare the model or application you plan to run, expected capacity, timeline and any data or network constraints. The enterprise team can then discuss a suitable delivery scope and the applicable commercial terms.

Tell us what you need to run.

From a first experiment to a dedicated deployment, start with your workload, capacity needs and operating requirements.

Explore enterprise solutions Talk to the team