National Institute of Telecommunications

The discussion was moderated by Anna Wittenberg — journalist and deputy editor-in-chief of zero.pl — who set a distinctly accountability-focused tone from the outset. She reminded participants that just a year earlier, “AI factories” functioned mainly as a slogan — somewhere between a strategy and a political declaration. Today, however, the focus is no longer on announcements, but on results: what has actually been built, where we are, and whether we can keep up with the pace of global change.

The panel included: Mikołaj Budzanowski (President for Central and Eastern Europe at InnoEnergy CE), Michał Goszczyński (Director of the Department of International Cooperation at the Ministry of Science and Higher Education), Wirginia Leszczyńska (CEO of DL Invest Group), Marek Magryś (Director of the Academic Computer Centre Cyfronet), Michał Pieprzny (Partner and Consulting Market Leader for Poland at Deloitte), Wojciech Stramski (CEO of Beyond.pl), Krzysztof Szubert (Director’s Plenipotentiary for Commercialisation and International Cooperation at NIT), and Michał Ziółkowski (Board Member for Technology at Play).

The most structured and multi-layered picture of the situation was presented by Krzysztof Szubert, referring to the Baltic AI Gigafactory project. His statement reflected more than just the scale of the investment — it told the story of a project that had already gained momentum but was then drawn into a new, far more complex European mechanism.

In December, the conditions changed quite significantly (…) which complicated the project for us.

This sentence captures the essence of the problem. We are talking about a project worth billions of euros, requiring not only enormous computing power but also a stable energy base and public-private partnerships. Meanwhile, the change in rules — shifting key decisions to the level of the European Commission and EuroHPC — meant reorganising the entire concept. Poland, which was already at an advanced stage of preparation, had to adapt to the new decision-making architecture, which in practice meant delays.

It was in this context that one of the most striking — and at the same time most accurate — comments of the panel emerged. Marek Magryś, referring to the pace of European action, did not hide his frustration:

It’s April, almost May, and nothing has happened.

His statement was not merely a critique of institutions, but an indication of a systemic problem. Europe — accustomed to processes, consultations, and regulations — operates at a pace that increasingly fails to match the dynamics of the technology market. Comparing this with the United States, where new infrastructure facilities are being built at the same time, was not rhetorical — it was a real comparison of two models of action.

Against this backdrop, the voice of business resonated particularly strongly. Wojciech Stramski brought something often missing from such discussions — the experience of acting here and now, without waiting for systemic decisions.

We cannot wait (…) we are investing in this infrastructure.

Even more important was what followed. Demand for computing power has not only emerged — it has quickly exceeded available resources. This completely changes the perspective. Contrary to fears often associated with large infrastructure investments, the problem is not a lack of customers, but a lack of capacity to serve them.

One of the most valuable parts of the panel was the shift from large-scale projects to the everyday functioning of companies. Participants pointed out that despite enormous interest in AI, many initiatives remain stuck at the testing and pilot stage.

We don’t see large-scale, scalable solutions (…) these are island innovations.

The reasons are complex. On the one hand, organisations are not yet ready for the deep transformation of processes required by AI. On the other hand, there is a very practical barrier — cost.

The cost of tokens (…) kills most use cases.

This is where infrastructure ceases to be an abstraction and becomes a condition for real implementation of technology.

In the third part of the panel, the discussion clearly shifted towards a strategic perspective. A key statement was made:

The game is not about infrastructure (…) the game is about our IP.

At this point, AI factories cease to be an end in themselves. They become a tool for building a technological economy based on knowledge, data, and competencies. Without this, Poland risks a scenario in which it educates specialists for other markets and then buys back the solutions they create.

An interesting perspective on the role of the state also emerged. Instead of direct investment financing, there was increasing emphasis on building stable demand.

Off-take (…) would be sufficient for the financial model.

This approach assumes that the state does not need to be the main investor, but can play a key role as a client guaranteeing the use of infrastructure, thereby attracting private capital.

Towards the end, the discussion took on a surprisingly self-reflective tone. A statement was made that goes beyond technology:

We don’t buy Polish solutions (…) others buy their own.

This diagnosis concerns not infrastructure, but how the market functions and decisions are made. It shows that even when competencies exist, barriers may lie in a lack of trust and consistency in building domestic advantages.

The conclusion of the panel was clear and left no illusions. Poland has reached a point where further delays in decision-making are beginning to have real consequences.

Either we seize this moment and become a hub (…) or we fall behind.

At the same time, it was emphasised that the development model based on low costs is coming to an end. This means a transition to an economy based on knowledge, technology, and its own added value.

The discussion in Katowice was, in essence, about the present, not the future. Artificial intelligence is ceasing to be an innovation and is becoming infrastructure — the foundation of a modern economy.

And that is why the most important question is no longer whether we will build it.

It is whether we will manage to do it on our own terms.