Where infrastructure enables an ecosystem
In 2020, while many countries were still debating national AI strategies, Linköping built.
In that mid-sized Swedish city, the Berzelius supercomputer was installed at Linköping University. Privately financed and powered by advanced NVIDIA systems, it strengthened Sweden’s AI research capacity almost overnight.
Within a few years, Berzelius had become more than a research machine. It became a place where academic research translated directly into practical use.
Infrastructure, when designed with intent, does more than provide compute. It concentrates capability.
A model that already works
Berzelius was built with a mission: give researchers access to advanced, local AI compute. It avoided complex ownership structures and long decision cycles. Instead, infrastructure and community were aligned from the start.
In 2024, the European Commission launched its AI Factories program, selecting nineteen facilities across Europe and backing them with billions in public funding. The Netherlands joined with plans for a national facility in Groningen, expected to become operational in late 2027 or early 2028, roughly eight years after Sweden activated Berzelius.
National AI factories signal scale and ambition. By their nature and ownership structure, they come with significant governance layers and bureaucratic complexity. That is inherent to national initiatives.
Regional AI factories serve a different role. They sit closer to users, closer to domain expertise, and closer to the companies that turn research into value. Within a shared domain, they align research, industry and entrepreneurship around a common purpose. They are structurally more agile and able to move faster.
The two models are not competitors. They complement each other. National capacity provides scale and geopolitical weight. Regional factories provide speed, focus and applied impact.
The Lorentz position
Lorentz builds on that insight and takes it one step further.
Rather than focusing only on research capacity, Lorentz AI Centers are designed as sector-focused ecosystems from day one. Each center combines a purpose-built AI factory with an anchor tenant, a leading university partner, and a group of scale-ups and corporates working within a clearly defined domain.
The first Lorentz Regional AI Factory focuses on Health. In the Netherlands alone, this sector exceeds €120 billion annually, yet operates on fragmented IT systems and legacy data structures. The opportunity for applied AI is significant, but it requires structured compute access, modelling support and strong coordination.
Lorentz is not a political megaproject. It is an operational model designed to become active within months, starting in the Netherlands, and the opportunity to scale in the Nordics, for sustainability and energy efficiency.
Cloud lowers the threshold. It does not create ecosystems.
Cloud infrastructure has made AI experimentation easier. That matters.
But experimentation alone does not build ecosystems.
When compute, data, governance and domain expertise are aligned within a regional setting, something changes. Intellectual property stays closer to its source. Anchor tenants help set direction. Universities connect research directly to business use. Scale-ups grow within shared infrastructure instead of rebuilding everything themselves.
Sovereignty in this context is not a slogan. It is clarity. It means knowing where workloads run, under which jurisdiction data is stored, how costs evolve and who controls the infrastructure.
The real bottleneck is not GPUs
Access to compute is necessary. It is rarely enough.
Most organisations run on decades-old IT systems. Hundreds of applications are tied together through complex integrations and inconsistent data. When AI systems connect to this environment, structural weaknesses surface quickly.
AI does not fix fragmented data foundations. It exposes them.
That is why a Lorentz AI Centre includes modelling support and an AI Centre of Excellence from the start. Infrastructure alone does not create impact. Coordinated capability does.
AI is not an IT experiment
For years, IT spending was seen as a cost of doing business. Today, it determines competitive position.
AI is not a technical side initiative. It is a strategic investment decision that shapes operating models and capital allocation.
Boards decide whether AI becomes embedded infrastructure that strengthens over time or remains dependent on external platforms and short-term pilots. In AI time, one year can equal a strategic decade.
Regions that align infrastructure, governance, talent and capital move forward. When purpose-built AI infrastructure is organised within a shared domain, the effect is structural.
Research connects faster to use. Industry builds on shared capability. Talent and compute reinforce each other instead of dispersing.
If this resonates, listen to the conversation with Ken van Ierlant and Viktor Mirovic in the Your AI Your Way Podcast, episode 5. The real question is not whether AI infrastructure matters. It is whether you are prepared to organise around it.
If you want to explore what a Regional AI Factory could look like for your sector, get in touch with one of our MDCS.AI experts.
