AI Compute Infrastructure,
Licensed as Software.
3COL™ is the orchestration layer that sits on top of your GPU infrastructure — handling student access, job scheduling, cooperative revenue distribution, and IT administration. One platform. Three access patterns. Built for education.
What 3COL™ Actually Does
Most GPU solutions hand you bare hardware and leave the rest to your IT team. 3COL™ is the software that makes GPU infrastructure manageable, accountable, and revenue-generating — without building anything from scratch.
3COL™ — the Cooperative Compute Orchestration Layer — is Civic Compute's licensed software platform. It installs on top of any NVIDIA GPU cluster and immediately provides student access portals, faculty job submission, IT administration dashboards, and the accounting engine that calculates and distributes cooperative revenue.
The platform is built on proven open-source infrastructure: SLURM (the scheduler that runs the world's largest supercomputers), NVIDIA DCGM (real-time GPU telemetry), and Prometheus + Grafana (live monitoring dashboards). What 3COL™ adds is the layer that connects these tools to your institution's identity system, your students' coursework, and your cooperative revenue agreement.
The result: a community college IT team can deploy AI compute access for an entire campus — with full audit trails, per-course usage reporting, and automatic revenue calculations — without deep GPU engineering expertise.
Every job submitted, every GPU-hour consumed, and every dollar of revenue earned is logged in a tamper-proof record accessible to your institution's administration and board at any time.
CRAM — Cooperative Revenue Accounting Module
The financial engine inside 3COL™. Reads SLURM job records monthly, applies your cooperative's revenue-share rules, and produces institution-level statements and payment files. Board-auditable. Dispute-aware.
Multi-Site Federation
3COL™ supports a network of cooperating institutions, each running their own nodes. Jobs can be routed across sites based on availability and priority, with usage accurately attributed back to each institution.
OpenAI-Compatible API Endpoint
Any tool that works with the OpenAI API — LangChain, LiteLLM, Jupyter notebooks, custom applications — works with 3COL™'s gateway out of the box. No vendor lock-in. No rewriting student code.
Failed Job Policy
Jobs that fail due to infrastructure issues are billed at $0.00. Policy is visible to users at job submission and enforced in CRAM billing.
Three Ways to Reach the Cluster
Every user role on campus — students, researchers, developers, and administrators — has a purpose-built interface to the same underlying GPU infrastructure.
Student & Faculty Portal
A browser-based interface any student can use on day one. No command line. No configuration. SSO login from your existing identity provider.
- Single Sign-On (SSO) via institutional credentials
- Curated AI model catalog — pre-loaded, ready to run
- Job submission with configurable GPU and time limits
- Course-level usage tracking and quota enforcement
- Output retrieval and download when jobs complete
- Faculty administration — enroll courses, set budgets
Direct API Access
An OpenAI-compatible REST endpoint for developers and researchers who want to integrate GPU compute directly into their code, notebooks, or applications.
- OpenAI-compatible endpoint — drop-in for existing code
- Per-department and per-course API key provisioning
- LiteLLM gateway for multi-model routing
- Rate limiting and quota enforcement per key
- Usage analytics per key, exportable monthly
- Works with LangChain, Jupyter, Hugging Face, and more
SLURM Orchestration
Direct scheduler access for advanced research workflows, large batch jobs, and HPC-style workloads that need fine-grained resource control.
- Full SLURM command-line interface for research users
- Partition configuration per institution and use type
- Priority queuing — coursework, research, commercial
- Per-institution account isolation and reporting
- Multi-site job routing across the cooperative network
- Tamper-proof job records feed directly into CRAM billing
The Technology Stack
3COL™ is assembled from proven open-source infrastructure — tools that run the world's largest GPU clusters — plus the proprietary accounting and orchestration layer Civic Compute built specifically for the cooperative model.
Browser-based interfaces for students, faculty, IT staff, and cooperative administrators. SSO-authenticated. Role-based access control. Mobile-responsive.
LiteLLM provides an OpenAI-compatible gateway that routes inference requests to the correct model and GPU partition. Per-course and per-department API keys enforce usage quotas and generate the billing events that feed CRAM.
SLURM assigns each incoming job to the correct GPU partition, enforces resource limits, maintains a permanent per-job audit record, and supports multi-site federation across the cooperative network. NVIDIA-owned. Runs most of the world's top supercomputers.
The proprietary layer that makes the cooperative model work. CRAM reads SLURM job records, matches each GPU-hour to the institution that hosted it, applies your cooperative's revenue-share agreement, and produces institution statements and payment files. Board-auditable. Tamper-proof.
Real-time GPU utilization, temperature, power consumption, and cluster health — visible to IT staff and the cooperative operations team. DCGM flags hardware issues before they become failures.
Institution-owned GPU hardware installed within your existing facility. Civic Compute owns and maintains the equipment. Zero water consumption. Operates within existing electrical infrastructure — no grid upgrades required.
What Your Institution Gets
3COL™ solves three problems simultaneously: it gives students and faculty access to real AI compute, it gives IT a manageable system to operate, and it generates recurring revenue from capacity that would otherwise sit idle.
Predictable SaaS Pricing
Annual software license. No per-job billing surprises. No exposure to GPU spot-market volatility. A fixed, budgetable cost your finance office can plan around — and grant-eligible as an operational software expense.
Revenue from Idle Capacity
Spare GPU capacity — nights, weekends, and between semesters — is sold to commercial AI workloads through the cooperative marketplace. Revenue is calculated by CRAM and paid directly to your institution monthly.
Real AI Access for Students
Students train models, run inference, and complete coursework on actual GPU infrastructure — not shared cloud credits that expire or throttle. Faculty set per-course budgets and track usage in real time.
Data Stays On Premises
Institutional data never leaves your network. 3COL™ runs on hardware inside your facility, air-gapped from Civic Compute's infrastructure. Your IT team retains full governance of what runs on your nodes.
Grant Support Included
Civic Compute provides hands-on grant application support — aligning your 3COL™ deployment with Perkins V, NSF ATE, and state workforce development funding. We write to your institution's priorities, not a generic template.
Cooperative Network Effect
Every institution that joins the cooperative strengthens every other member. Larger network = more capacity = more commercial buyers = more revenue per institution. The cooperative structure is a durable competitive moat.
Who 3COL™ Is Designed For
Every audience segment has a distinct deployment model, funding path, and value proposition — all on the same platform.
Community Colleges
Two-year institutions with existing IT infrastructure are the natural home for 3COL™. License the software, deploy on your own GPU nodes, and give every student and faculty member access to real AI compute for coursework, applied research, and workforce development programs. Spare capacity generates monthly revenue back to your institution.
CareerTech & Technical Colleges
The same subscription model as community colleges, with an employer-placement framing. Train students for real AI infrastructure roles — GPU cluster operation, model deployment, inference optimization — with hands-on access to production-equivalent compute. Certificate and credential programs built around actual infrastructure.
Research Park Landlords
License 3COL™ as a tenant amenity. Startups, research labs, and small AI companies in your innovation district get proximity-based GPU compute access without the capital cost of their own infrastructure. A genuine, differentiated competitive advantage that drives lease commitments and reduces tenant churn.
Cooperative Members
Join the cooperative network. Your institution's spare GPU capacity is pooled with other members' capacity, increasing the total commercial offering and therefore the revenue each institution earns. Cooperative governance rights give member institutions a voice in platform development, pricing policy, and network expansion decisions.
From Inquiry to Live Deployment
A 3COL™ deployment is designed to be low-friction for your institution's IT team. Most of the complexity is handled by Civic Compute — your team contributes what it already has.
Site Assessment
We review your facility — GPU hardware, available power, physical space, and network connectivity. No commitment required.
Cooperative Agreement
We execute a licensing and cooperative agreement defining revenue share, governance rights, and operational terms. You retain full space ownership.
Platform Deployment
Civic Compute installs and configures 3COL™ on your infrastructure. SLURM partitions set up per-institution. SSO integrated with your identity provider.
Campus Onboarding
Faculty and IT administrators receive training. Student access goes live through the portal. Course budgets and quotas are configured per department.
Revenue Begins
Spare capacity connects to the cooperative marketplace. CRAM calculates monthly revenue, produces institution statements, and initiates payment.
Simple, Transparent Pricing
Annual software license. No per-job fees. No GPU market exposure. Contact us for institution-specific pricing — we size every engagement to match your infrastructure and enrollment.
Starter
For two- and four-year educational institutions deploying 3COL™ on institution-owned GPU infrastructure.
- Full 3COL™ platform license
- Student & Faculty Portal
- API Gateway (LiteLLM)
- SLURM configuration and support
- CRAM revenue accounting
- Grant application support included
- Onboarding and IT training
Core
All platform features plus enrollment in the cooperative network — spare capacity is marketed commercially and revenue flows to your institution monthly.
- Everything in Community College tier
- Cooperative network enrollment
- Commercial capacity marketing
- Monthly revenue statements (CRAM)
- ACH payment to institution account
- Cooperative governance rights
- Multi-site federation access
Enterprise
For research park operators, innovation districts, and commercial deployments where tenant GPU access is a differentiating amenity.
- Full platform license
- Tenant-level API key provisioning
- Usage reporting per tenant
- Commercial SLA options
- Dedicated onboarding
- Priority support channel
- Custom cooperative terms
What Institutions Ask Us Most
Does my institution need to own GPU hardware?
Yes — 3COL™ is a software license that runs on your GPU infrastructure. Civic Compute can assist with hardware procurement, financing, and grant funding to help your institution acquire nodes if you don't already have them. Hardware cost can be substantially offset by grant funding we help you pursue.
How does the revenue share work?
Your institution's spare GPU capacity is marketed through the cooperative network to commercial AI buyers. Revenue is calculated by CRAM based on verified GPU-hours and distributed according to your cooperative agreement — typically the majority going directly to the host institution.
What does my IT team need to manage?
Day-to-day operation is minimal. The platform handles user authentication through your existing SSO, queues and routes jobs automatically, and generates revenue reports without manual input. Civic Compute provides ongoing technical support and handles SLURM configuration updates.
Can students use 3COL™ without knowing how to use a command line?
Yes. The Student Portal is a browser-based interface designed for non-technical users. Students submit jobs, select models, and retrieve outputs through a guided web UI. The SLURM command-line interface is available as a separate access pattern for research and advanced users.
How is our institutional data protected?
3COL™ runs on hardware within your facility, under your network governance. Civic Compute has zero visibility into your institution's data. The platform is air-gapped from our operational infrastructure — we can see job metadata for billing purposes, but not job content.
Is 3COL™ eligible for Perkins V or NSF ATE funding?
The software license and deployment are eligible as instructional technology expenses under Perkins V. We actively help institutions align their 3COL™ deployment with grant priorities — workforce preparation, student opportunity, and applied learning — and assist with the grant application itself.
What GPU hardware does 3COL™ support?
3COL™ is built for NVIDIA GPU infrastructure — the H100, A100, and H200 series used in serious AI workloads. DCGM telemetry runs on all modern NVIDIA data center GPUs. We conduct a site assessment to confirm hardware compatibility before any agreement is signed.
How long does a typical deployment take?
From signed agreement to live student access: typically 30–60 days depending on hardware readiness and IT integration complexity. SLURM configuration, SSO integration, and portal deployment are handled by Civic Compute's team. Your IT staff reviews and approves each step.
Ready to Bring AI Compute to Your Campus?
Schedule a 30-minute introduction. We'll walk through the platform, the licensing model, and what a 3COL™ deployment looks like at an institution your size.