Poland is starting to appear more frequently on the European data center map, but its opportunity doesn’t necessarily lie in replicating the Frankfurt, London, Amsterdam, Paris, or Dublin model. The growth of artificial intelligence is changing the needs of these facilities: having land and filling server rooms is no longer enough. Energy, power grids, cooling, high-performance computing, and the capacity to deploy hardware at scale are increasingly influencing where AI infrastructure can expand.
The keys to AI infrastructure in Poland in 20 seconds
- The Polish market currently hovers around 200 MW of capacity, and CBRE estimates it could reach 500 MW by 2030.
- Microsoft announced an additional 2.8 billion zloty investment to expand its cloud and AI infrastructure in the country.
- Poland is developing the PIAST AI AI factory and aims to participate in the European gigafactory program for AI.
- The main constraint will be electricity: PSE plans to invest more than 64 billion zloty in the grid by 2034.
Europe faces a problem that could turn into an opportunity for markets like Poland. In 2025, 859 MW of new data center capacity was added across the continent, according to CBRE. However, demand continues to outpace supply. By 2026, the consultancy estimates needs of around 848 MW compared to approximately 710 MW of new capacity coming online.
The five main European markets, known as FLAPD (Frankfurt, London, Amsterdam, Paris, and Dublin), still host much of the activity. But they also face challenges that are hard to resolve quickly: shortage of available electrical power, land restrictions, lengthy connection processes, and facilities not originally designed for high-density AI racks.
Poland arrives later. For that reason, it can do some things differently.
From the current 200 MW to a potential 500 MW market
CBRE estimates the Polish data center market at around 200 MW and projects it could reach 500 MW by 2030. Warsaw is likely to remain the main concentration of infrastructure, although opportunities are starting to emerge elsewhere in the country.
The consultancy explicitly mentions locations such as Bełchatów, Szczecin, Bielsko-Biała, and Choczewo. Electricity availability will be decisive in determining which of these projects come to fruition.
This movement has already attracted major technology providers. Microsoft launched its Polish cloud region in 2023 after announcing an initial investment of $1 billion. In February 2025, it announced a second phase involving 2.8 billion zloty, roughly $700 million at the time of announcement, aimed at expanding cloud and AI infrastructure through mid-2026.
Google also strengthened its cooperation with Poland through a memorandum signed with the Polish Development Fund and Operator Chmury Krajowej to extend AI adoption among Polish companies and institutions.
But building a national AI platform involves more than attracting hyperscale investments.
Poland is also developing PIAST AI, an AI manufacturing plant located in Poznań, and in 2025 formally announced its intention to participate in the European gigafactory program for AI. The scale difference is significant: Europe envisions these future facilities to house large numbers of accelerators and provide capacity for training and developing advanced models.
This requires simultaneous planning for computing and energy.
Electricity could determine how much Poland can grow
A GPU can be purchased in months. A new high-voltage line, substation, or multi-hundred-megawatt electrical connection takes years of planning.
This is likely Poland’s primary challenge.
Polskie Sieci Elektroenergetyczne (PSE), the operator of the transmission network, has an approved development plan for 2025-2034 that includes more than 64 billion zloty in investments. It covers around 4,700 kilometers of new 400 kV circuits, 28 new substations, and the modernization of another 110.
The plan isn’t solely designed for data centers. Poland needs to transform its power system for many other reasons: electrification, renewables, storage, offshore wind, and future nuclear generation. However, PSE explicitly recognizes that new large electricity consumers—such as data centers—will increase demand.
AI data centers fit perfectly into that category.
The planning aims to create capacity to integrate approximately 18 GW of offshore wind, 45 GW of photovoltaic solar, and over 19 GW of onshore wind, as well as facilitate the connection of the upcoming Pomeranian nuclear plant and potential small modular reactors.
This makes power grid planning inseparable from digital policy. A country may have land, fiber, operators, funding, and demand for AI, but if obtaining 100, 200, or 500 MW of power takes many years, the project will relocate elsewhere.
Poland could avoid some of the issues faced by mature European markets if it aligns decisions on new data center zones with the development of generation, transmission, and distribution infrastructure.
AI data centers are also cooling differently from traditional ones
There’s another physical difference. Facilities built for conventional business loads were designed around much lower rack densities than those introduced by AI accelerators.
CBRE notes that during 2025, more European operators incorporated liquid cooling to support high-density AI loads.
Therefore, building the facility first and deciding later on the computing hardware may be a flawed strategy.
A design-ready AI facility can jointly plan electrical distribution, liquid cooling, high-speed networks, and high densities. It can also consider heat recovery, energy storage, or mechanisms to adjust computing loads based on grid availability from the outset.
The less visible element is the supply chain.
Large clusters demand more than GPUs. Servers, CPUs, high-speed memory, storage systems, Ethernet or InfiniBand switches, optical components, transformers, UPS systems, generators, and cooling equipment are all part of the same project. Lead times for any of these components can determine the project’s start date.
Poland’s membership in the European Union, its geographic position, and its pool of technical professionals are advantages, but they don’t replace the need for industrial coordination.
The European market sends a clear signal. Traditional large centers continue to grow, but finding electricity and space there is becoming more difficult. CBRE forecasts that European availability rates could fall to 5.7% in 2026, while demand is expected to surpass the new capacity coming online.
Poland doesn’t need to become another Frankfurt to benefit from this situation.
It can develop centers closer to areas with new electricity generation, design liquid cooling and high-density facilities from the start, expand infrastructure outside Warsaw where sensible, and link IT capacity planning with national electric plans.
The challenge is synchronizing these traditionally separate planning timelines. A data center operator may focus on three to five years; an electric utility, ten to fifteen; server manufacturers work in shorter cycles; and a GPU-dependent company may want capacity in months.
If operators, utilities, administrations, manufacturers, integrators, and capital only start planning once demand appears, they will likely come late.
Poland’s opportunity is precisely in proactive planning. The roughly 500 MW the market could reach by 2030 doesn’t have to be just more data center capacity. It can be designed as infrastructure specifically tailored for a new generation of computing that consumes, cools, and connects differently.
Frequently Asked Questions
How much data center capacity does Poland have?
CBRE estimates Poland currently has around 200 MW of capacity. It forecasts that this could grow to approximately 500 MW by 2030.
Why can Poland attract new AI data centers?
The country combines EU membership, growing digital demand, technological investment, and space for new facilities. Additionally, issues with electricity availability and land in more mature European markets are encouraging the search for alternative locations.
What is the main obstacle for AI data centers in Poland?
Electricity availability and connection times will be key factors. PSE plans to invest over 64 billion zloty until 2034 to expand and modernize the transmission network, but these investments also need to address other system requirements.
What is PIAST AI?
PIAST AI is an AI manufacturing plant Poland is developing in Poznań as part of Europe’s initiative to expand supercomputing capacity dedicated to AI. The country has also expressed its intention to participate in the European gigafactory program for AI.
