How much does it cost to run AI yourself?
Hardware, license, and operations: predictable costs instead of token pricing. Run your scenario in the cost calculator.
Questions and answers
Answers about hardware, models, costs and getting started.
Does on-premise AI make sense for SMEs?
Yes, especially with compliance requirements, Swiss data sovereignty, and strict data protection.
Which data should never go to the cloud?
Sensitive business data, personal and compliance-relevant data. On-premise keeps everything under your control.
For which company sizes does on-premise AI make sense?
From small teams to large enterprises. Often economical from about 5 users.
Can ERP, CRM and document data be integrated?
Yes, interface-compliant and with minimal integration effort.
Can I evaluate on-premise AI before buying?
Yes: every Wednesday is Test Wednesday. Request an API key and test realistically.
How can I get a quote?
Via price request or contact form. We are happy to help with a demo and decision support.
How performant is on-premise AI?
Preliminary S2 figures (August 2026): 1,300 tokens/s under parallel requests, 44 concurrent office users. Verification ongoing.
How do maintenance and updates work?
Planned updates, maintenance windows, optionally zero downtime. Security patches promptly.
What does the license include?
Updates, new features, and always current AI models. Yearly per GPU, no token fees.
Which operational model is used?
Managed On-Premises, co-management, or self-service. Depending on your needs.
How does the purchase process work?
After the purchase, we build. Time to market: about 3 weeks.
Can installation and maintenance be done remotely, or is on-site required?
Usually remote. On-site for installation or when needed.
How does the on-premise AI solution scale?
Scale with additional servers and custom cluster planning. Dell D1 and D2 are rack alternatives with one or two GPUs.
Can I start small and scale later?
Start with S1 or S2, choose Dell D1 or D2 for rack installations, and plan larger deployments individually.
Is your team based in Switzerland?
Yes. Our team is fully based in Switzerland.
What are typical use cases for SMEs?
From documents to code to internal assistants: privacy-compliant and high-volume.
Which hardware does onprem.ai use?
NVIDIA GPUs in S1 and S2, with Dell D1 and D2 as 2U rack alternatives. Larger installations are planned individually.
No matching questions found.
Try another search term.