Pick the right AI resource
Stony Brook researchers can use AI on campus GPUs, through SUNY's Google Cloud platform, and on New York's Empire AI supercomputer. Start from what you need to do, and from the kind of data you have.
AI resources at a glance
GPU cluster for AI and machine learning
Best for: training and running models, Jupyter notebooks, AlphaFold, imaging. GPUs: 48 NVIDIA H200 NVL and 48 RTX PRO 6000 Blackwell. Data: non-sensitive research data, not PHI. Who: Stony Brook researchers, instructors, staff and students.
HIPAA-aligned cluster for clinical research
Best for: AI on clinical and imaging data under an IRB protocol. GPUs: 8 NVIDIA A100 and 16 RTX PRO 6000 Blackwell, with high-memory nodes up to 1 TB. Data: PHI your IRB approves. Who: Health Sciences and Medicine study teams.
Generative AI tools and cloud projects
Best for: Gemini Enterprise, NotebookLM and GrantAI; Google Cloud projects with GPUs and managed AI services. Data: no PHI. Who: Stony Brook researchers, their postdocs and research graduate students.
New York's AI supercomputer
Best for: large-scale and distributed training, foundation models. GPUs: Alpha, 104 NVIDIA H100; Beta, 288 NVIDIA Blackwell GPUs in GB200 NVL72 racks. Data: non-sensitive research data. Who: Stony Brook projects in our Empire AI allocation.
Energy-efficient AI inference
Best for: large-memory AI inference. Hardware: AmpereOne M processors and Qualcomm Cloud AI inference accelerators. Who: a national system funded by the NSF (award OAC-2505376, $21.95M), led by the Institute for Advanced Computational Science with the University at Buffalo. Follow NSF data management guidelines where they apply.
Your AI path
New to AI for research
From your first AI tool to your first GPU job.
- Get SUNY AI Platform accessFree Gemini Enterprise, NotebookLM and custom agents
- Get an HPC accountOne account for SeaWulf and NVwulf
- Open Jupyter on NVwulfOr private AI chat, in your browser
- Use GPUs wellInteractive GPU sessions, step by step
Scaling up
When one GPU node is not enough.
- Size your jobsTime limits and GPU requests
- Request Empire AIFor multi-node training
- Start a Google Cloud projectOn the SUNY AI Platform
- Bring questions to office hoursMondays and Fridays at 11 AM during the semester
SUNY AI Platform
A cloud-based platform from SUNY System that gives researchers secure, scalable access to AI and machine learning tools, generative AI models and research environments on Google Cloud. In its first year, researchers at 9 SUNY campuses used it, with 82 PIs and 152 collaborators working on 82 projects.
New: AI applications at no cost to you
SUNY now provides Gemini Enterprise, NotebookLM and custom agent building through the platform, at no cost to Stony Brook or to researchers. They're for campus researchers and their student researchers, for research that could lead to publications and for current or future grant-funded projects.
- Faster literature reviews: synthesize deep research together with your own documents.
- Audio and video summaries of dense papers.
- Q&A practice for dissertation defenses, presentations and panels.
- Your codebooks, applied: use an AI research assistant to apply your own codebooks to your research data.
You also get SUNY-only office hours with experts who help you deploy and troubleshoot custom workflows, and SUNY workshops through the year.
How to get started
- Request general access with the SUNY service ticket. Researchers and their students each request their own.
- Watch for a confirmation email saying your license is active.
- Go to ai-platform.suny.edu and click Log In at the top right.
- Open Gemini Enterprise in the left menu to reach Gemini Enterprise, NotebookLM and custom agents.
Licenses are limited.
More on the platform
- Create a Google Cloud project with GPUs and managed AI services on Gemini Enterprise Agent Platform (formerly Vertex AI), and use GrantAI.
- Write it into new awards. Ask us for current SUNY-discounted GPU rates for your budget.
- Apply to SUNY System's summer call for platform resources, including HPC and managed AI services.
Get started at ai-platform.suny.eduRequest access through SUNY
Empire AI
Empire AI is a partnership of New York's leading public and private universities, including SUNY, launched in April 2024 to give researchers across the state a world-class AI computing center. It is housed at the University at Buffalo, runs on hydropower from Niagara Falls and uses liquid cooling.
104 NVIDIA H100 GPUs
13 Lenovo HGX nodes with 8 GPUs each and 4 PB of storage.
288 NVIDIA Blackwell GPUs
Four NVIDIA DGX GB200 NVL72 racks with 30 PB of storage and NVIDIA AI Enterprise software, the first academic deployment of its kind.
How Stony Brook researchers get access: submit your project through Stony Brook's Empire AI request. Requests are reviewed, prioritized and approved every two weeks. Writing a proposal that uses Empire AI? We can provide a letter of support.
Write AI into your grant
Ready-to-use Facilities, Equipment and Other Resources text for the SUNY AI Platform and Empire AI is in the proposal kit, next to the text for our own clusters. For budgets, letters of support or help choosing between campus GPUs, Empire AI and Google Cloud, book a grant consultation at least a month before your deadline.