For research and academia

Compute for the paper, the lab and the whole department

Academic pricing, grant-friendly invoicing, Slurm environments your students already know, and an option to have us build and run the cluster your department has been trying to fund.

Why labs choose us

Less time on infrastructure, more time on results

Academic pricing

20 percent off published rates, prepaid GPU-hour blocks that never expire, and fixed quotes you can drop into a grant budget.

Grant-friendly billing

Purchase orders, net-30 invoicing, usage reports by project code, and quotes formatted for NSF, NIH, ERC and Horizon proposals.

Slurm out of the box

Reserved clusters come with Slurm, Lmod modules and a shared parallel filesystem that feels like the campus HPC cluster, minus the queue.

No job time limits

Run for weeks. Automatic checkpoint scheduling to persistent storage and hot spares mean a node failure costs minutes, not the run.

Departmental clusters

Pooled grant money buys a lot of GPU when someone else handles design, procurement, cooling and operations. We build and run it.

Custom Clusters

ML engineering on demand

For groups without ML systems expertise, our engineers run the training and hand back weights, code and a citable benchmark report.

Managed Training
Open science

Reproducibility is a feature

Every instance and cluster image is versioned and archived. Cite the exact GreenAI Stack release in your paper and anyone can rerun your experiment on identical software, years later.

  • Versioned images with pinned driver, CUDA and framework versions
  • Public datasets mirrored in-region at no cost: Common Crawl, FineWeb, Wikipedia and other openly licensed corpora
  • Carbon accounting per job for the sustainability statement reviewers now ask for
  • Data sharing between collaborating institutions via shared object storage buckets
Example allocation
  • GroupNLP lab, 12 researchers
  • Compute32x H100, reserved 12 months
  • Storage200 TB parallel + 500 TB object
  • SchedulerSlurm with fair-share by PI
  • Rate$2.07 / GPU / hr after discount
  • AnnualFixed-price quote for grant
Get a quote like this
Research

Academic program

Can't find what you need? Ask our team.

What is the academic discount?

20 percent off published on-demand and reserved cluster rates for accredited universities, national labs and non-profit research institutes. Verified with an institutional email address.

Can you invoice against a grant?

Yes. We provide fixed-price quotes for grant proposals, purchase-order invoicing, and usage reports formatted for grant reporting. We can also pre-purchase blocks of GPU-hours that never expire.

Do you support long-running jobs?

Reserved clusters have no job time limits. On-demand instances can run indefinitely as long as your account is in good standing, and checkpointing to persistent storage is included.

Can students use the platform?

Yes. Teaching allocations with per-student quotas and a shared JupyterHub are available for courses. Ask about our education credits.

Apply for academic pricing

Verify with your institutional email and get a fixed-price quote for your next proposal within two business days.