Prem seeks $100M Series A as export bans boost sovereign AI demand

2026-06-18 · Source: AI – SiliconANGLE · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Cybersecurity & Data Privacy · Depth: Fundamental Awareness, quick

Summary

Swiss artificial intelligence startup Prem SA is raising \$100 million in Series A funding at a valuation of at least \$500 million, with the round expected to close in Q3 2026. Founded in 2023, Prem provides software for running AI models on a company's own infrastructure, ensuring data privacy and control for regulated firms like hedge funds and law firms. This fundraising coincides with the launch of Fluso, an encrypted workspace for AI agents that uses open-weight models on customer data. Demand for "sovereign AI" solutions is surging due to recent U.S. export bans. For instance, a June 12 order forced Anthropic PBC to disable foreign access to its Fable 5 and Mythos 5 models. Gartner Inc. forecasts \$80 billion in global sovereign cloud IaaS spending for 2026. European spending is projected to triple from \$6.7 billion in 2025 to \$23.1 billion in 2027. Prem's Swiss base, with its EU adequacy decision, positions it against rivals like Mistral AI and Aleph Alpha.

Key takeaway

For Directors of AI/ML or Legal Professionals managing sensitive data, you should prioritize sovereign AI solutions. Recent export bans on models like Anthropic's Fable 5 and Mythos 5 highlight the risks of third-party dependencies. Deploying models on your own infrastructure, as Prem SA offers, ensures data residency and auditability, mitigating compliance risks and maintaining operational control. Evaluate providers offering private, auditable environments, especially those with favorable jurisdictional standing like Switzerland, to secure your AI operations.

Key insights

Sovereign AI solutions are gaining traction due to data control needs and export restrictions on cloud-based models.

Principles

• Data residency is critical for regulated industries.
• Open-weight models enable auditability and control.
• Geopolitical factors influence AI model accessibility.

Method

Prem's offering involves running AI models, fine-tuning, document analysis, and inference in private, customer-controlled environments, ensuring data never leaves the customer's infrastructure.

In practice

• Deploy AI models on private infrastructure.
• Evaluate open-weight models for auditability.
• Consider Swiss-based AI providers for EU data.

Topics

• Sovereign AI
• Data Residency
• AI Export Controls
• On-Premise AI
• Open-Weight Models
• AI Startup Funding

Best for: CTO, VP of Engineering/Data, Executive, Investor, Director of AI/ML, Legal Professional

Related on AIssential

• India’s Sovereign AI Push Runs Into US Model Access Limits — India's sovereign AI push is challenged by foreign model access limits, revealing critical infrastructure and governance dependencies.
• Establishing AI and data sovereignty in the age of autonomous systems — Enterprises are prioritizing AI and data sovereignty to mitigate IP risks associated with third-party generative AI models.
• Mistral Pioneers Sovereign AI in Europe — Mistral offers a European, open-source, sovereign AI stack as an alternative to U.S. proprietary models.
• Announcing strategic MOUs with Indra Group and Multiverse Computing - Cohere — Digital sovereignty in AI is achieved through collaborative partnerships and flexible deployment options across diverse environments.
• IBM Report: Most EMEA Executives Don’t Fully Understand Their AI Dependencies — Many EMEA executives lack critical visibility into AI dependencies, risking costs and operational stability.
• BREAKING: The Claude Ban Shook India's Developers - Now They're Rethinking Their Entire AI Stack — India's AI strategy is shifting towards Small Language Models and open-weight alternatives to mitigate dependency and address unique local challenges.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI – SiliconANGLE.