The Smart Sovereign AI Factory
As intelligence becomes general, trust becomes the infrastructure.
Samara unites sovereign compute, behavioural intelligence and agentic alignment on European ground. Enterprises deploy AI they can trust today. General intelligence is researched, prototyped, and governed inclusively and responsibly, for what comes next.
Horizon oneThe human-agentic enterprise
Deploy AI your people and clients can trust, on sovereign infrastructure you trust.
Horizon twoBeneficial general intelligence
Research and enable beneficial general intelligence responsibly and inclusively, under shared governance, in a collaboratory built for it.
Every wave of AI has moved the scarcity. The next one moves it to trust.
First the scarce resource was compute, then intelligence. Now intelligence is becoming abundant, and general intelligence may already be arriving. What stays scarce is the context that makes AI valuable and the trust that makes it safe to deploy.
Wave 1
Compute
Computing power
Whoever could build or buy the most compute.
Compute is now being built at unprecedented scale. It remains essential, and leadership now depends on what runs on it.
Wave 2
Models
Intelligence
Whoever trained the most capable model.
Open models now reach around 90% of frontier performance at release, and close most of the rest within months at a fraction of the cost.
MIT open- versus closed-model performance study, 2025.
Wave 3
Trust infrastructure
Trust
Whoever can deploy and evolve intelligence responsibly, and protect the human-agentic systems that make it valuable.
When every organisation has access to advanced AI, the advantage lies in what is uniquely yours: your people, your agents, your data and the institutional knowledge that ties them together. Trust is the infrastructure that holds them.
Horizon
General intelligence
Governance
Whoever earns the trust of the people and institutions it serves.
Some of the world’s leading AI labs now say the era of general intelligence has begun. Whether it arrived this year or arrives in the next few, the question that matters is the same: under whose governance, and with whose trust?
What is scarce
Who leads
What it means
AI capability is accelerating. Deployment is falling behind.
Vendors hand over the technology and leave organisations to close three gaps alone. As intelligence becomes general, the same three gaps grow from business risks into questions for whole societies.
Human readiness
Agentic alignment
Sovereign compute
Samara closes all three gaps at once, guided by European standards of ethics, governance and innovation.
Enterprise today — Human readiness
Are your people ready to work differently?
Employees and customers live the change directly, yet their stress and trust usually surface late, as attrition, failed adoption or reputational risk.
Enterprise today — Agentic alignment
Can your agents be trusted to act?
Agents make decisions with minimal governance for trust, ethics or accountability. Risk compounds wherever data and compute sovereignty are weak.
Enterprise today — Sovereign compute
Is the infrastructure underneath truly yours?
Sovereignty is widely claimed and rarely held: nearly every provider answers to a foreign jurisdiction, a shareholder's agenda or a government's chequebook.
General intelligence — Human readiness
Can societies and institutions adapt?
When intelligence becomes general, readiness grows from a change-management question into a question for whole workforces, public institutions and democracies.
General intelligence — Agentic alignment
Can general intelligence stay aligned under pressure?
Capable systems can reason well about the common good and still take the short-term win. At general capability, that gap grows from an operational risk into a societal one.
General intelligence — Sovereign compute
Who governs the ground intelligence grows on?
If general intelligence evolves on infrastructure answerable to one company or one state, it will carry their values with it.
95%
of enterprise AI pilots fail to deliver financial value
MIT NANDA, 2025, cited in Roland Berger, The AI-First Organization, 2026
40%+
of agentic AI projects are expected to be cancelled by 2027
Gartner, 2025
The factory integrates the full AI value chain in one place: the compute underneath, the intelligence that runs on it, the governance that supervises it, and the people and institutions who shape it. The team behind Lorentz built the AI factory at TU Eindhoven in 2025. Samara takes that proven format and scales it.
Foundation: the sovereign base
GPU infrastructure
Enterprise-grade NVIDIA compute with guaranteed hours, SLA-backed uptime and full operational support, on Grace Blackwell GB300 in Finland and next-generation Vera Rubin architecture in Eindhoven.
Lorentz
Your AI runs on sovereign ground you can trust.
Foundation: the sovereign base
Compute intelligence
Smarter compute that lowers energy intensity and carbon footprint, starting from a turnkey hydropower facility in Finland that is already built and permitted.
Lorentz
Sovereign capacity, years ahead of new European builds.
Intelligence: agentic deployment
Agentic systems
The active intelligence layer for creating and deploying agentic enterprise solutions, with multi-modal architectures and behavioural supervision built in.
EARTHwise, KeenCorp and TasteIndex
Agents are built with the people and customers they serve in view.
Intelligence: human and agentic governance
Governance and trust
Evaluation, accountability and supervision of agent reasoning and behaviour against the 13 criteria of the EARTHwise Alignment Benchmark, mapped to the EU AI Act. Combined with KeenCorp's emotion AI, which reads tension and trust across teams from existing written communication, identity-protected by design, to guide human trust and culture alignment.
EARTHwise and KeenCorp
Trust that can be measured, governed, and enhanced.
Peak: market reach
Ecosystem and deployment
The market interface where trusted advisers such as KPMG and Roland Berger carry integrated solutions to enterprises and governments.
Samara with its distribution partners
One diagnostic unlocks the whole ecosystem.
Peak: capability
Samara Academy
Human capability and interdisciplinary excellence, built with universities, frontier AI labs, AI ethics and safety programmes, and company-funded PhD positions in client organisations.
The Samara Collaboratory
People grow alongside the technology.
Peak: leadership
Centre of Excellence
Shaping standards, policy and adoption pathways for competitive and responsible AI, in Europe and beyond.
The Samara Collaboratory
Europe helps set the terms for general intelligence.
- Delivered by
- Why it matters
One governance thread. Production agents today and sandboxed general-intelligence R&D tomorrow are supervised and deployed under the same benchmarks.
A sovereign base. Layers 1 and 2 provide full tenant data isolation, on sovereign infrastructure that can be trusted.
Regenerative energy. Our Finnish site runs on hydropower, and our Eindhoven deployment follows the TU/e AI factory model, built by the same MDCS.ai team on 100% renewable energy. We are committed to regenerative, low-carbon solutions, including hydrogen, for the energy demands of sovereign compute.
Your advantage is your context.
When every organisation can reach advanced AI, value comes from what is uniquely yours: your data, your people’s expertise, your relationships, your culture and the behavioural intelligence that makes you distinct.
Samara protects that context and lets it compound for its owner.
- Full tenant data isolation
- GDPR and EU AI Act native by design
- Entry through a single, qualified access point
Trust is measured, earned and kept.
Samara works the same way with human teams and AI agents: establish a baseline, improve alignment, then keep watching. Three live signals feed the loop, one from each of the people and systems AI touches.
Assess
A Leadership Tension Assessment for teams, or an Arena pilot on live agents, with evidence produced inside the client’s own systems. The baseline comes first, so every later change can be attributed.
Improve
Alignment tuning for human teams and AI agents. Teams see the tensions that block change. Agents gain the experience of consequence in simulation and self-adjust. The result is measurable behaviour change.
Supervise
Real-time tracking of tension and drift, for people and agents alike, as a continuous intelligence layer. When drift appears, the loop returns to improvement.
The same loop will govern sandboxed general-intelligence research inside the collaboratory.
Founded by four companies. Built as one ecosystem.
Each co-founder brings a distinct intelligence. Together they close the loop from compute to human and agentic behaviour to governance.
Lorentz
Sovereign compute
The sovereign compute trunk of the factory: NVIDIA-backed European AI infrastructure with SLA-backed uptime, operated on sovereign ground.
EARTHwise
Agentic alignment and AGI research
The EARTHwise Arena creates, tests, tunes, certifies and supervises AI agents, and powers the Samara AGI Lab’s research toward beneficial general intelligence.
KeenCorp
Human readiness intelligence
Reads tension and potential in organisations from the written communication that already exists. Identity-protected by design, at team and organisation level.
TasteIndex
Customer desire intelligence
Decodes the sensory signal behind what customers truly want, enabling hyper-accurate product customisation for leading food and beverage brands.
The Samara Collaboratory
Building the conditions for beneficial general intelligence.
The future of intelligence will be shaped by more than capability: by the ecosystems, values and cultures within which it evolves and is deployed.
The Samara Collaboratory brings together researchers, enterprises, governments and civil society to advance AI and AGI on sovereign European infrastructure, guided by shared benchmarks, transparent governance and the AGI Constitution.
Powered by Lorentz, KeenCorp, TasteIndex and EARTHwise, the Collaboratory creates a trusted environment where human, collective and artificial intelligence can evolve together in service of a thriving future.
Samara AGI Lab
Research lab for general intelligence
Samara’s R&D lab for general intelligence utilises the EARTHwise Arena and the EARTHwise Alignment Benchmark for the prototype development of novel multi-modal AGI architectures, in collaboration with our AGI partners.
AGI Sandbox
With TU Eindhoven, TU Delft, and SingularityNET.
A sandboxed R&D programme progressing toward general intelligence on European sovereign terms, under the same supervision and alignment benchmarks that govern production agents.
AGI Constitution
World Academy of Art and Science, 2024
A governance framework for general intelligence, led by Dr. Anneloes Smitsman and shared with the UN AGI Forum. It guides how research inside the collaboratory is conducted and held to account.
Centre of Excellence
Standards and policy
Where standards, policy and adoption pathways are shaped with governments and institutions, so Europe helps set the terms for general intelligence.
Samara Academy
Education and capability
Building human capability alongside the technology, with universities, frontier AI labs, AI ethics and safety programmes, and company-funded PhD positions.
EARTHwise
Co-founder: alignment research lead
EARTHwise brings the EARTHwise Arena and the EARTHwise Alignment Benchmark to the Samara AGI Lab, and leads its alignment research.
Lorentz
Co-founder: sovereign compute
Lorentz provides the secure, sovereign compute on which sandboxed research runs, isolated from production tenants.
TasteIndex
Co-founder: human preference signal
TasteIndex contributes its sensory and preference signal, helping research understand what people value and desire, beyond what they say.
KeenCorp
Co-founder: human trust signal
KeenCorp contributes identity-protected insight into tension and trust in organisations, so research stays grounded in how people actually experience AI.
The Elowyn Tree
The Elowyn Tree is being developed by EARTHwise within Samara as the first AGI prototype architecture, powered by more than 69,000 people playing Elowyn, where humans and AI agents develop win-win logic through the consequences of play.
Samara is the winged seed of the elm. In Cornish, an ancient Celtic language, the elm is elowen, the root of Elowyn’s name.
Responsible. Inclusive. Collaborative.
Peer-reviewed and published
Frameworks are published and peer-reviewed, so progress can be evaluated, challenged and improved.
Inclusive and shared
Researchers, enterprises, governments and civil society shape the work together, with access tiered by role.
Governed and transparent
Sandboxed research runs under the benchmarks that govern production agents, guided by the AGI Constitution.
Built with partners who know what is at stake.
Samara brings together infrastructure, distribution and research partners, a Governance Board with deep experience in technology, law and finance, and advisers who helped define the field of general intelligence.
Compute sponsor of the Lorentz trunk
AI factory partner and AGI Sandbox research partner
AGI Sandbox research partner
Trusted-adviser distribution partner
Trusted-adviser distribution partner
Trusted-adviser distribution partner
AGI Sandbox research partner
AGI research partner of EARTHwise
Games partner of EARTHwise
Clients and pilot partners of Samara’s ventures
KeenCorpASML, Schiphol Group, Nubank, Baker Tilly, Kemin, Covenant Health, Veldsink Groep, Saints & Stars and CTSN
TasteIndexLindt, Aviko, FrieslandCampina, Juan Valdez Café, Henri Willig, Royal Club, Delta Wines and Gusteau
Peer-reviewed
EARTHwise alignment research published in the AGI-26 proceedings, Springer Nature, Lecture Notes in Artificial Intelligence 16855.
Governance
The AGI Constitution, led by Dr. Anneloes Smitsman and published by the World Academy of Art and Science in 2024.
Governance Board
Advisers
Dr. Ben Goertzel, SingularityNET. Dr. David Hanson, Hanson Robotics. Dr. Mariana Bozesan, AQAL Capital. Dr. Rachel StClair, Servamind. Dr. Trista Patterson, formerly Microsoft and Xbox. Liv Boeree. Virendra Kumar Sahu. Kees van der Vleuten. Roeland Reinders. David Horlock. Cindy Olson.
2027
Launch: Finland and the Netherlands
Sovereign compute goes live in Finland, then Eindhoven, with healthcare anchor tenants proving the model. KeenCorp, TasteIndex and EARTHwise run as tenants from day one.
2028
Scale: three factories in Europe
The same architecture is replicated across European sites. Sandboxed general-intelligence research begins, under the benchmarks that govern production agents.
2029
Global: ecosystems of factories
Samara ecosystems across Europe and beyond drive large-scale AI innovation, with trust proven at every step toward beneficial general intelligence.
Partner with Samara.
Samara is built as an ecosystem, with four ways in.
Deploy
Run AI on sovereign ground, with readiness, alignment and supervision built in.
For enterprises and public institutions
Build
Develop and test agentic solutions inside the factory, on the full Samara stack.
For builders and venture partners
Collaborate
Join the collaboratory: research, standards and sandboxed general-intelligence R&D.
Distribute
Carry Samara to your clients as a trusted-adviser partner.
For advisers and integrators
Investors, enterprises, research institutions and public bodies: let’s talk about what we can build on trusted ground.
Start a conversationThe age of general intelligence needs trusted ground.
Samara is building it in Europe: sovereign, collaborative, inclusive and in service of life.
Join the Samara Collaboratory.
Get in touch with us for enquiries about Samara and our Collaboratory. Let us know about your interest as a researcher, enterprise, public institution or partner. Our team reviews every submission.