# Xinity > Sovereign AI infrastructure software for European enterprises. Run any generative AI model fully on-premise, on your own servers, with an OpenAI-compatible API. GDPR-compliant by architecture. No cloud. No data leaves your infrastructure. Headquartered in Vienna, Austria. Xinity lets European enterprises replace public LLM APIs with a sovereign endpoint running inside their own walls. Teams switch from cloud AI to self-hosted AI by changing three environment variables; existing apps, SDKs, and prompts keep working. The platform ships in two layers: an open-source engine for developers, and an enterprise platform for organisations that must answer to regulators. ## What Xinity does Xinity is a software layer that turns an organisation's existing servers into a private, production-grade AI platform. Instead of sending prompts and data to a public cloud AI provider such as OpenAI, Anthropic, or Google, companies run large language models on hardware they own and control, so no data ever leaves their building or their legal jurisdiction. Concretely, Xinity provides: - An OpenAI-compatible API. Existing applications that already call a cloud AI provider switch to Xinity by changing the base URL and API key. No code rewrites, no new SDK, same request and response format (chat, JSON mode, function calling, streaming). - On-premise model hosting. Run open-source models such as Llama, Mistral, Gemma, and Qwen, or proprietary models, on local GPUs. Model weights are deployed to the customer's hardware and verified on load. - Production orchestration. Intelligent model routing across GPUs, automatic load balancing, GPU-aware scheduling, fault tolerance, and multi-node scaling. - Governance and observability. Per-request logging, cost tracking, monitoring, role-based access control, SSO, and a full audit trail mapped to EU AI Act articles, all stored on the customer's own infrastructure. - Fast deployment. The engine, platform, and (optionally) hardware ship together; first model typically runs in hours to days, not months. In short: the AI comes to the data, not the other way around. This keeps sensitive information inside the organisation's control and changes the cost model from unpredictable per-token pricing to forecastable, capacity-based infrastructure costs. ## Built for regulated enterprises with sensitive data Xinity is designed for organisations that cannot send their data to a public cloud, because doing so would breach GDPR, the EU AI Act, NIS2, banking secrecy, journalistic source protection, attorney-client privilege, medical confidentiality, or trade-secret obligations. These organisations are not cloud-sceptical; they are sovereignty-blocked. Xinity gives them a way to adopt generative AI without their prompts, documents, customer records, or proprietary knowledge ever leaving their own infrastructure or their jurisdiction. Typical sectors include finance, healthcare, legal, public sector, media, and industrial manufacturing. ## Who it is for - European enterprises with strict data sovereignty and residency requirements. - Regulated organisations handling sensitive or confidential data under GDPR, the EU AI Act, NIS2, or sector-specific rules. - IT and platform teams running open-source models on their own hardware. - Developers who want an OpenAI-compatible local inference engine with no hyperscaler dependency. ## Pilot and pricing - 30-day pilot: an ASUS Ascent GX10 AI supercomputer with the complete Xinity software, models, and support, installed on the customer's premises by a Xinity technician. Free for the trial month and cancellable at no charge within it. Open to companies with 15 or more employees. - Ongoing pricing is capacity-based (pay for the GPUs you run, not per token), billed as a software subscription on the customer's own infrastructure, across Community (free), SME, and Enterprise tiers. See the pricing page for current figures. - [Pilot Program](https://xinity.ai/pilot): apply for the 30-day pilot. - [Full pricing](https://xinity.ai/sovereign-ai-pricing): complete pricing across all tiers. ## Key facts - Founders: Alexander Zehetmaier (CEO) and Jonas Vander (CTO), each with 10+ years of AI infrastructure experience, who studied AI together at Radboud University. - Two-layer product: Xinity Engine (open-source core) and Xinity Platform (enterprise layer with hardware provisioning, compliance tooling, RBAC, SSO, and support). - Core inference engine: open source, Apache License 2.0, free forever. - Dashboard: source-available; free tier covers one organisation and one inference node, paid tiers unlock multi-node and multi-org orchestration. - Compliance: GDPR-aligned by architecture, EU AI Act audit trail, NIS2 controls, mTLS, role-based access control, data residency control. Protected from the US CLOUD Act because no data leaves the customer's jurisdiction. - Deployment: on-premise, Docker, Kubernetes, bare metal; typical go-live in hours to days, not months. - Headquarters: Am Gestade 5/2, 1010 Vienna, Austria. ## Community and events - [Sovereign AI Lab](https://xinity.ai/sovereignailab): a monthly Vienna meetup community hosted by Xinity for practitioners and decision-makers working on sovereign and on-premise AI. - Sovereign AI Roundtable: an invitation-based event series bringing together European enterprise and public-sector leaders to discuss data sovereignty, compliance, and self-hosted AI. ## Pages - [Homepage](https://xinity.ai): Xinity overview and entry point. - [What is Sovereign AI](https://xinity.ai/what-is-sovereign-ai): explainer on sovereign AI and why it matters. - [Platform / Features](https://xinity.ai/sovereign-ai-platform): the Xinity platform and its capabilities. - [How it works](https://xinity.ai/how-xinity-works): deployment process and architecture in three steps. - [Pricing](https://xinity.ai/sovereign-ai-pricing): tiers and capacity-based pricing. - [Pilot Program](https://xinity.ai/pilot): 30-day on-premise pilot with hardware, software, and support included. - [ROI Calculator](https://xinity.ai/roi-calculator): estimate cost savings versus cloud AI. - [Open Source](https://xinity.ai/xinity-opensource): the open-source engine and how to use it. - [Compliance](https://xinity.ai/compliance): GDPR, EU AI Act, NIS2, and security controls. - [Shared Responsibility Model](https://xinity.ai/shared-responsibility-model-on-premise-ai): who owns what when running AI on-premise, across nine areas: customer duties, Xinity duties, and what is genuinely shared. - [Customers](https://xinity.ai/who-uses-xinity): who uses Xinity and for what. - [Partners](https://xinity.ai/xinity-partnership): partner program and ecosystem. - [Become a Partner](https://xinity.ai/become-a-partner): apply to join the partner program. - [Advocates](https://xinity.ai/advocates): the Xinity advocates program. - [Referral Program](https://xinity.ai/referral): refer companies to Xinity. - [Careers](https://xinity.ai/careers): open roles at Xinity. - [Blog](https://xinity.ai/sovereign-ai-blog): articles on sovereign AI, compliance, and cost. - [Press](https://xinity.ai/press): press coverage and media resources. - [About Us](https://xinity.ai/aboutus): team, story, and mission. - [Contact](https://xinity.ai/contact): get in touch or book a demo. ## Industry solutions - [Sovereign AI for Banks](https://xinity.ai/sovereign-ai-for-banks): on-premise AI for financial institutions under banking secrecy and EU regulation. - [Sovereign AI for Hospitals](https://xinity.ai/sovereign-ai-for-hospitals): on-premise AI for healthcare providers handling patient data under medical confidentiality. - [Sovereign AI for Law Firms](https://xinity.ai/sovereign-ai-for-law-firms): on-premise AI for legal work under attorney-client privilege. - [Sovereign AI for Media](https://xinity.ai/sovereign-ai-for-media): on-premise AI for newsrooms and journalistic source protection. - [Sovereign AI for Government](https://xinity.ai/sovereign-ai-for-government): on-premise AI for public administrations and government bodies. ## Blog: featured posts - [Europe's Answer to AI Sovereignty](https://xinity.ai/sovereign-ai-blog/xinity-europe-answer-to-ai-sovereignty): why Europe needs sovereign AI infrastructure and where Xinity fits. - [How Xinity Builds Genuinely Private AI Infrastructure](https://xinity.ai/sovereign-ai-blog/how-xinity-builds-genuinely-private-ai-infrastructure): the architecture behind Xinity's privacy claims. - [The Utilization Inversion](https://xinity.ai/sovereign-ai-blog/the-utilization-inversion): why cloud economics broke when AI agents arrived. - [Xinity vs Ollama, vLLM, and LocalAI](https://xinity.ai/sovereign-ai-blog/xinity-vs-ollama-vllm-and-localai): how Xinity compares to other self-hosted inference tools. - [AI Agents Need Sovereign Infrastructure](https://xinity.ai/sovereign-ai-blog/ai-agents-sovereign-infrastructure): the case for running AI agents on infrastructure you control. - [How to Migrate from Cloud AI to On-Premise with Xinity](https://xinity.ai/sovereign-ai-blog/how-to-migrate-from-cloud-ai-to-on-premise-with-xinity): a practical guide for moving from cloud LLM APIs to on-premise. ## Blog: all posts - [Build or Buy the LLM Control Plane?](https://xinity.ai/sovereign-ai-blog/build-or-buy-the-llm-control-plane) - [The Three Lives of a Server](https://xinity.ai/sovereign-ai-blog/the-three-lives-of-a-server) - [Xinity Runtime v0.23.0](https://xinity.ai/sovereign-ai-blog/xinity-runtime-v0.23.0) - [How to run LLM inference in an air-gapped environment](https://xinity.ai/sovereign-ai-blog/air-gapped-llm-inference) - [How Does On-Prem Infrastructure Scale?](https://xinity.ai/sovereign-ai-blog/how-does-on-prem-infrastructure-scale) - [Shadow AI is a supply problem, not a discipline problem](https://xinity.ai/sovereign-ai-blog/shadow-ai-supply-problem) - [AI video for companies whose material can't touch the cloud](https://xinity.ai/sovereign-ai-blog/ai-video-for-companies-whose-material-can-t-touch-the-cloud) - [How Xinity Builds Genuinely Private AI Infrastructure](https://xinity.ai/sovereign-ai-blog/how-xinity-builds-genuinely-private-ai-infrastructure) - [Why Not Just vLLM? An Honest Answer](https://xinity.ai/sovereign-ai-blog/why-not-just-vllm-an-honest-answer) - [Does Xinity see your data?](https://xinity.ai/sovereign-ai-blog/does-xinity-see-your-data) - [EU AI Act Transparency Rules Are Now Live](https://xinity.ai/sovereign-ai-blog/eu-ai-act-transparency-rules-are-now-live) - [Build Your Own AI Data Center, Part 1](https://xinity.ai/sovereign-ai-blog/build-your-own-ai-data-center-part-1) - [NOVARA auf Xinity Runtime: KI-Kollege ohne Cloud-Anbindung (German)](https://xinity.ai/sovereign-ai-blog/an-ai-coworker-that-never-phones-home-running-novara-on-xinity-runtime) - [Xinity Has Moved Into AI Factory Austria](https://xinity.ai/sovereign-ai-blog/xinity-has-moved-into-ai-factory-austria) - [The Utilization Inversion](https://xinity.ai/sovereign-ai-blog/the-utilization-inversion) - [Can AI be trusted? How to Build AI You Can Verify](https://xinity.ai/sovereign-ai-blog/can-ai-be-trusted) - [Does a Sovereign AI Gateway Cost You Speed?](https://xinity.ai/sovereign-ai-blog/does-a-sovereign-ai-gateway-cost-you-speed) - [The questions to ask before you believe the word "sovereign"](https://xinity.ai/sovereign-ai-blog/sovereign-ai-what-it-really-means) - [Warum ein KI-Motor das richtige Chassis braucht (German)](https://xinity.ai/sovereign-ai-blog/warum-ein-ki-motor-das-richtige-chassis-braucht) - [Eight New AI Models, One Sovereign Machine](https://xinity.ai/sovereign-ai-blog/eight-new-ai-models-one-sovereign-machine) - [Run Local AI Models with Xinity and keinsaas Navigator](https://xinity.ai/sovereign-ai-blog/local-ai-models-xinity-keinsaas) - [The EU just made sovereign, open source AI official policy.](https://xinity.ai/sovereign-ai-blog/the-eu-just-made-sovereign-open-source-ai-official-policy.) - [Case Study: MeinDienstplan Did the Math](https://xinity.ai/sovereign-ai-blog/case-study-meindienstplan-did-the-math) - [Sovereign AI Roundtable](https://xinity.ai/sovereign-ai-blog/sovereign-ai-roundtable) - [Xinity vs Ollama, vLLM, and LocalAI](https://xinity.ai/sovereign-ai-blog/xinity-vs-ollama-vllm-and-localai) - [AI Agents Need Sovereign Infrastructure](https://xinity.ai/sovereign-ai-blog/ai-agents-sovereign-infrastructure) - [Xinity × SNCS Partnership](https://xinity.ai/sovereign-ai-blog/xinity-×-sncs-bridging-the-gap-between-ai-innovation-and-data-sovereignty) - [How to Migrate from Cloud AI to On-Premise with Xinity](https://xinity.ai/sovereign-ai-blog/how-to-migrate-from-cloud-ai-to-on-premise-with-xinity) - [Sovereign Cloud or Sovereign Not Cloud?](https://xinity.ai/sovereign-ai-blog/sovereign-cloud-vs-architectural-sovereignty) - [Is open source safe enough for regulated industries?](https://xinity.ai/sovereign-ai-blog/open-source-safe-for-regulated-industries) - [Why We Open Sourced Xinity](https://xinity.ai/sovereign-ai-blog/why-we-open-sourced-xinity) - [Sovereignty-Washing: How to Spot Fake Sovereign AI Claims](https://xinity.ai/sovereign-ai-blog/sovereignty-washing) - [Your AI Is Probably Not EU AI Act Ready.](https://xinity.ai/sovereign-ai-blog/eu-ai-act-readiness-what-it-means) - [Understanding the AI Stack](https://xinity.ai/sovereign-ai-blog/understanding-the-ai-stack-where-xinity-fits) - [Data Residency for On-Premise AI](https://xinity.ai/sovereign-ai-blog/data-residency-on-premise-ai) - [Europe's Answer to AI Sovereignty](https://xinity.ai/sovereign-ai-blog/xinity-europe-answer-to-ai-sovereignty) - [GDPR Article 9 and AI](https://xinity.ai/sovereign-ai-blog/gdpr-article-9-ai-sensitive-data) - [Sovereign AI vs Cloud AI](https://xinity.ai/sovereign-ai-blog/sovereign-ai-vs-cloud-ai) ## Open Source - [GitHub](https://github.com/xinity-ai/xinity-ai): the open-source repository. - License: Apache License 2.0 (core engine). Dashboard is source-available with a free single-org, single-node tier. - Includes: inference engine, OpenAI-compatible REST API, model routing and multi-node orchestration, management dashboard, fine-tuning pipelines, audit logging, and Docker / Kubernetes / bare-metal deployment. - Architecture: gateway (OpenAI-compatible proxy), daemon (GPU node runtime), dashboard, model registry; uses engines such as Ollama and vLLM under the hood. ## Press - [Forbes Austria — Europe's AI independence](https://www.forbes.at/artikel/europas-ki-unabhaengigkeit): coverage of Europe's push for AI sovereignty. - [Forbes Austria — Xinity's software goes open source](https://www.forbes.at/artikel/under-30-update-xinitys-software-wird-open-source): on the open-source launch. - [Brutkasten — Mostly AI founders invest in Xinity](https://brutkasten.com/artikel/xinity-mostly-ai-founder-steigen-bei-neuem-wiener-startup-ein): investment announcement. ## Contact - [Website](https://xinity.ai): xinity.ai - [Email](mailto:contact@xinity.ai): contact@xinity.ai - [GitHub](https://github.com/xinity-ai/xinity-ai): source code and issues. - [LinkedIn](https://www.linkedin.com/company/xinity-ai): company updates. - [Instagram](https://www.instagram.com/xinity_ai/): visual updates and events. - [Mastodon](https://mastodon.social/@xinityai): fediverse presence. ## Optional - [Whitepaper](https://xinity.ai/whitepaper): in-depth technical and strategic overview. - [Downloads](https://xinity.ai/downloads): downloadable resources. - [Media Centre](https://xinity.ai/mediacentre_xinity): brand assets and media kit. - [Impressum](https://xinity.ai/impressum): legal entity information. - [Privacy Policy](https://xinity.ai/privacy-policy): data protection notice. - [Terms and Conditions](https://xinity.ai/termsandconditions): terms of service.