News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud
Thunder Compute

Thunder Compute

GPU sandboxes and instances for AI developers

Thunder Compute provides GPU instances and isolated Ubuntu sandboxes for machine learning experiments and reinforcement learning tasks. The company bills usage per second, aiming to reduce costs for developers running CUDA code and agent evaluations.

Share on WhatsApp

  • Raised $13 million in Series A funding
  • Offers Nvidia H100 GPUs for developers
  • Bills GPU usage per second for cost efficiency
  • Provides isolated Ubuntu sandboxes for RL tasks

What it is

Thunder Compute offers computing resources tailored for artificial intelligence developers. It provides long-lived GPU servers for hands-on coding and scalable micro virtual machines for isolated tasks. These environments run full Ubuntu operating systems, allowing users to execute CUDA code and manage machine learning experiments. The service targets reinforcement learning rollouts, model evaluations, and continuous integration jobs that require graphics processing units. By billing per second, the platform aims to provide cost efficiency for variable workloads.

How it came about

The company secured thirteen million dollars in a Series A funding round. This capital supports its mission to utilise idle graphics processing units. Thunder Compute positions itself as a solution for putting unused compute power back into productive use for developers and researchers. The firm focuses on enabling faster iteration for engineering teams through accessible GPU infrastructure.

The compute

Users can access Nvidia H100 graphics processing units through the platform. The service offers two primary forms: persistent instances for development and ephemeral sandboxes for automated tasks. Sandboxes are isolated Ubuntu virtual machines designed for reinforcement learning environments and evaluation pipelines. The system supports parallel execution of coding agent assessments. Developers can configure CPU, memory, and storage allocations alongside their GPU requirements.

Customers and deals

The platform serves machine learning engineers and reinforcement learning researchers. Customers include staff at series B startups, bootstrapped data science consultancies, and independent AI laboratories. Academic users, such as doctoral candidates, also utilise the service for research deadlines. Healthcare AI startups and robotics companies use the infrastructure for fine-tuning jobs and experimental sweeps without facing queue delays.

Money and ownership

Thunder Compute raised thirteen million dollars in Series A financing. The pricing model charges per second for exact usage duration. Rates for an H100 GPU instance are listed at approximately zero point zero zero zero eight one nine dollars per second. This structure allows bootstrapped teams and large organisations to manage budgets by paying only for active compute time rather than reserved blocks.

Its place in AI

The company focuses on the developer experience within the AI infrastructure market. It addresses the need for rapid testing and iteration in machine learning workflows. By providing on-demand access to high-performance chips, it enables teams to move away from job queuing systems. This approach supports faster shipping of software and more frequent experimental cycles for engineering groups.

Questions people ask about Thunder Compute

What type of GPU hardware does Thunder Compute provide?

The platform offers Nvidia H100 graphics processing units. Users can select these chips when creating instances or sandboxes for their machine learning and reinforcement learning workloads.

How does Thunder Compute charge for its services?

The company bills customers per second for the exact length of time their resources are active. This model applies to both long-lived instances and temporary sandboxes, helping users control costs.

What is the primary use case for Thunder Compute sandboxes?

Sandboxes are isolated Ubuntu virtual machines designed for reinforcement learning rollouts, model evaluations, and continuous integration jobs. They allow secure and scalable execution of GPU-dependent tasks.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Thunder Compute yet, so the facts on this page come from the sources listed above: thundercompute.com/about. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

    Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.