Servers · NL-CS2
Core S2 AI
Run serious AI models entirely inside your walls.
$3,999
- 24GB GPU for local LLMs and image models
- 12-core CPU, 64GB ECC RAM
- Local AI stack pre-installed (chat UI + API)
- Also does everything the Core S1 does
Free US shipping over $500 · 30-day returns · 3 years warranty
Overview
What it does
The Core S2 AI pairs the S1's server foundation with a 24GB GPU, enough to run popular open-weight chat, transcription, and image models locally. Customer data, contracts, and prompts never leave your network, and there is no per-token bill. Ships with a local AI stack pre-installed and ready to use.
In the box
- Core S2 AI server
- Power cable
- Ethernet cable
- Setup documentation
Specifications
The numbers
| GPU | 24GB VRAM workstation GPU |
|---|---|
| CPU | 12-core / 24-thread |
| Memory | 64GB ECC (max 192GB) |
| Storage | 2TB NVMe + 2x 4TB NAS drives, 4 bays |
| Network | Dual 2.5GbE, 10GbE option |
| Software | Local AI stack pre-installed, OpenAI-compatible API |
| Warranty | 3 years |
FAQ
Core S2 AI questions
Which AI models can the Core S2 AI run?
The 24GB GPU comfortably runs the popular open-weight model families for chat, coding assistance, transcription, and image generation at production speed for a small team. The pre-installed stack exposes an OpenAI-compatible API, so existing tools and scripts point at your server with a one-line change.
Why run AI locally instead of using cloud APIs?
Three reasons: privacy (data never leaves your building), cost (no per-token fees, which matters at volume), and control (models keep working regardless of vendor pricing or availability changes). Cloud APIs still win for frontier-scale models; many businesses use both.
Also in servers
Deploying Core S2 AI at scale?
Multi-location rollouts and orders above ten units get volume pricing and a dedicated contact. Tell us your counts and timeline.