Skip to main content



News Release


Contact: Paul Redfern
Cell: (607) 227-1865

June 10, 2026

New CAC billing model for Red Cloud and Seneca

The Cornell Center for Advanced Computing is transitioning to a flexible Compute Unit Bundle billing model. This single, unified currency allows PIs to seamlessly allocate budgets across diverse computing profiles. It retains the "prepaid card" functionality of Red Cloud subscriptions to prevent over-billing.

Starting June 25, 2026, both Red Cloud and the new Seneca HPC cluster will use this shared billing system.

Key Takeaways:

  • Changeover Date: June 25, 2026.
  • Bundle Pricing: $350 per bundle of 10,000 compute units ($0.035 per unit).
  • Red Cloud Balances: Current balances automatically convert to compute units on a 1:1 basis on June 25.
  • Seneca Pilot Balances: The last day of Seneca pilot access is June 30.
  • Bulk Discounts: None available.

Compute Unit Consumption Rates

Usage is calculated per hour based on your chosen hardware tier:

ResourceHardware TypeUnits per HourEffective Cost per Hour
Red Cloud1 CPU core1 unit$0.035
Red Cloud1 NVIDIA T4 GPU12 units$0.42
Red Cloud1 NVIDIA V100 GPU32 units$1.12
Seneca1 CPU core1 unit$0.035
Seneca1 NVIDIA A100 GPU57 units$2.00
Seneca1 NVIDIA H100 PCIe GPU70 units$2.45
Seneca1 NVIDIA H100 NVL GPU100 units$3.50

Seneca Access Options

Project PIs can choose between two options to access the Seneca cluster:

  • Option 1 (Compute Unit Bundles): Enable Seneca usage and purchase $350 bundles through the Project Management Portal.
  • Option 2 (Seneca Condo): Contribute your own Seneca-compatible hardware to the cluster. You pay zero compute unit fees on your own nodes, receive a dedicated Slurm queue, and pay a predictable per-node professional maintenance fee. Bursting into general cluster capacity utilizes your purchased compute unit bundles. Contact help@cac.cornell.edu to discuss.

Key Links

If you need assistance or have questions about the new billing model for Red Cloud and Seneca, or the Seneca Condo option, contact CAC Help.