Key specifications
Illustrations are for reference. Service scope is defined by the plan you choose.
Save 5% with a 2-year term or 10% with a 3-year term.
A dedicated GPU server running open-weight AI models for your company only: private chat, document Q&A and an internal API.
Some data should never be pasted into a public AI service: client files, source code, contracts, health or financial records. A Private AI Server gives you large language model capabilities on hardware dedicated to your company, with models, prompts, documents and logs kept on that server.
We install and tune open-weight models (for example from the Llama, Mistral or Qwen families, subject to each model’s license), a browser-based chat interface with single sign-on, and retrieval-augmented generation (RAG) so users can ask questions about your own documents and see the sources behind each answer. Developers get a chat-completions style API endpoint for internal tools.
TechSpire Solutions manages the hardware, operating system, GPU drivers, model updates and security patching. You decide which models run, who can use them and how long conversation logs are kept.
All prices in USD, per server per year at the 1-year rate. Swipe the table sideways to see every plan.
| Feature | Business | Professional Recommended | Enterprise |
|---|---|---|---|
| Price | $18,000.00 / year per server | $36,000.00 / year per server | $84,000.00 / year per server |
| With a 3-year term | $16,200.00 / year Save 10% | $32,400.00 / year Save 10% | $75,600.00 / year Save 10% |
| Best for | For a first private AI deployment serving one team. | For company-wide use with larger models and several document sources. | For heavy workloads, fine-tuning and around-the-clock coverage. |
| What’s included |
|
|
|
| Order | Customize term & quantity for the Business plan | Customize term & quantity for the Professional plan | Customize term & quantity for the Enterprise plan |
| SKU | TS-AI-104 |
|---|---|
| Category | AI & Productivity |
| Pricing | Per server, per year |
| Contract | Annual subscription; 1-, 2- or 3-year term |
| GPU | 1× NVIDIA L40S 48 GB / 2× NVIDIA L40S 48 GB / 2× NVIDIA H100 80 GB |
| Memory & storage | From 128 GB RAM + 2 TB NVMe up to 512 GB RAM + 8 TB NVMe |
| Models | Open-weight LLMs and embedding models, license-checked before deployment |
| Access | Web chat with SSO, internal API, optional site-to-site VPN |
| Data handling | Prompts, documents and logs stay on your dedicated server |
| Support | Response times by tier; 24/7 critical response on Enterprise |
| Lead time | Deployed in 10–15 business days |
The standard offer is a dedicated server that we host and manage. On-premises installation in your own rack can be quoted separately after we review power, cooling and network requirements.
We deploy open-weight models whose licenses permit your intended commercial use and that fit the GPU memory of your tier. We review license terms with you before deployment and can swap models during the term at no extra charge.
No. Inference runs on your dedicated server. Outbound connections are limited to operating-system and model updates that we schedule, and can be restricted further on request.
Custom application development, integrations beyond the connected sources in your tier, and purchases of on-premises hardware are quoted separately.
Have another question? Ask a solutions architect
