KB008: Using NAIRR Pilot for AI Research

Scope: Artificial Intelligence / Machine Learning Audience: Researchers needing specialized AI hardware (H100s, Wafer-scale) or Model Access Last Updated: Feb 9, 2026

1. What is NAIRR?

The National Artificial Intelligence Research Resource (NAIRR) Pilot is a federal initiative connecting U.S. researchers to computational, data, and software resources for AI research at no cost.

Best For:

• Exotic Hardware: Access to Cerebras, SambaNova, Graphcore, and Groq accelerators.
• Massive Scale: Training jobs requiring hundreds of H100/A100 GPUs.
• Model Access: API credits for OpenAI, Anthropic, and Hugging Face.

2. Allocation Types

A. Startup Projects (The “On-Ramp”)

• Purpose: Benchmarking, code testing, and preparing for full proposals.
• Duration: Up to 3 months.
• Review: Fast-tracked (~2 weeks).
• Strategy: Use this to prove your code scales before asking for a massive allocation.

B. Research Resources

• Purpose: Full-scale scientific campaigns.
• Duration: 12 months (typically).
• Review: Standard NSF-style peer review (Monthly/Quarterly cycles).

3. How to Apply

1. Visit: nairrpilot.org
2. Browse Catalog: Identify the specific resource you need (e.g., “Delta at NCSA” or “Cerebras CS-2”).
3. Submit Request:
   • For Startup: Brief 1-page description.
   • For Research: Full proposal (~5-10 pages).

Research Computing Support: We can assist with technical justification and benchmarking data for your proposal. Contact us at research-computing@ucr.edu.