Skip to content
C-Gen.AI

For AI startups with cloud GPU credits

GPU credits for your startup? Claim $10,000* more in Nucleaton™ credits.

Whichever cloud gave you GPU credits, Nucleaton™ gives those GPUs a complete platform: training and inference on one pool, real-time observability, usage reports and project management. Your GPU cluster typically deploys in about 25 minutes. It runs on Nebius, Google Cloud, Azure, DigitalOcean, OCI and AWS.

$10,000*in Nucleaton™ credits

For startups and teams with active GPU credits from a cloud provider.

  • Works on popular GPU clouds
  • Official Oracle partner
  • From the inventor of AWS SageMaker HyperPod

Claim your Nucleaton™ credits

Claiming for a team with GPU cloud credits.

We only use your details to contact you. See our Privacy Policy.

Make GPU credits go further

Credits run out. Idle GPUs make it happen sooner.

Cloud providers bill GPUs by the hour, busy or not. Give training and inference their own GPUs and each set waits for work part of the day.

Nucleaton™ runs both together on one shared GPU pool, so your credits buy more training and more serving.

A week of cloud GPU hoursIllustrative
Separate GPUs for training and for inference each sit idle part of the time. With Nucleaton™, training and inference run at the same time on one GPU pool.
Without Nucleaton™: separate GPUs
Training
Inference
With Nucleaton™: both at once on one GPU pool
Both

Popular clouds

One complete GPU platform, whichever cloud your credits are on.

  • Hands-off cluster setup

    Nucleaton™ deploys and configures a Slurm-based GPU cluster on the cloud account you already use.

  • Training across GPU nodes

    Multi-node GPU jobs with queues, priorities and fair scheduling between your researchers.

  • Endpoints for your models

    Inference endpoints with request routing and demand-driven scaling, on the same GPUs as training.

  • Full GPU visibility

    Real-time telemetry on utilization, GPU health and job performance, across every node in the cluster.

  • Reports on every GPU hour

    Usage is attributed to projects, teams and users, so you can see what your credits paid for.

  • Access for the whole team

    Projects, users and storage managed in one place, with GPU quotas set per team.

  • GPU failures handled for you

    Real-time telemetry spots failing GPUs. The affected GPU node is isolated and replaced automatically.

  • Training and inference, one pool

    Both run at the same time on the same GPUs, by the priorities you set. Guaranteed GPUs are never touched.

  • Slurm and optional Kubernetes

    The cluster runs on Slurm, with a Kubernetes overlay available for container-based work.

Step by step

Three steps on popular clouds.

  1. Tell us where your GPUs are

    Share your cloud and setup in the form, and an engineer will usually reply within 4 business hours.

  2. Deploy in your own account

    Nucleaton™ installs on the GPUs in your existing cloud account. You keep full control of your data and models.

  3. Go live

    The cluster typically deploys in about 25 minutes, ready to take training runs and inference alike.

Your GPU clusterIllustrative
  • GPU poolProvisioned
  • Training runRunning
  • Model endpointServing
  • Live telemetryOn
  • Usage reportsCurrent
  • GPU node faultDetected
  • Replacement GPU nodeOnline

Led by a world-class team

We’ve designed, built and run some of the largest AI training and inference systems in production.

Dr. Sami Kama

Dr. Sami Kama

Founder & CEO
  • 17 years at CERN
  • Inventor, TensorFlow-TensorRT Integration (NVIDIA)
  • Inventor and Architect, AWS SageMaker
  • Inventor and Architect, AWS SageMaker HyperPod
  • Creator, SageMaker Profiler (world’s first distributed AI profiler)
Richard Vencu

Richard Vencu

Head of Engineering
  • Former Head of HPC & MLOps, Stability AI
  • Managed one of the world’s largest AI training clusters
  • Co-Founder & Engineering Lead, LAION

Offer FAQ

Quick answers

Which clouds are supported?
Nebius, Google Cloud, Microsoft Azure, DigitalOcean, Oracle Cloud Infrastructure (OCI) and AWS.
What’s part of the platform?
Cluster setup, training and inference on one GPU pool, live telemetry, usage reports, projects and access, and automatic GPU fault handling.
Is this the same as GPU credits?
No. Your cloud credits pay for the GPUs. You get up to $10,000 in Nucleaton™ credits for the platform that runs them.
Who is the offer for?
Startups and teams with active GPU credits from one of these clouds.
Do we have to switch clouds?
No. Nucleaton™ deploys where your GPUs already are, in your own account.
Is C-Gen.AI affiliated with these clouds?
C-Gen.AI is an official Oracle partner. It is not affiliated with or endorsed by Nebius, Google, Microsoft, DigitalOcean or Amazon Web Services.

*Up to $10,000 in Nucleaton™ credits. We’ll confirm the amount for your team when we talk.

C-Gen.AI is an official Oracle partner. Google Cloud, Azure, DigitalOcean, AWS, Nebius and OCI are trademarks of their respective owners. C-Gen.AI is not affiliated with or endorsed by Nebius, Google, Microsoft, DigitalOcean or Amazon Web Services.

Your GPU credits, fully put to work.

Claim up to $10,000 in Nucleaton™ credits and run a complete GPU platform on the cloud you already use.

Claim up to $10,000 in Nucleaton™ creditsNo commitment required.