USD 200bn for Poland. A massive wave of data-center investment is coming

Poland could attract more than USD 200bn in data-center investment by 2050, with spending set to exceed USD 75bn over the next decade alone, according to the latest PwC report seen by XYZ.

Rama 9, czyli kontrowersyjne centrum danych powstające w sercu mocno zaludnionej części Bangkoku. Nowe inwestycje wyrastają jak grzyby po deszczu
Rama 9, the controversial data center being built in the heart of a densely populated part of Bangkok. New developments are popping up like mushrooms after rain. Photo: TNP/Arnun Chonmahatrakool/Thai News Pix/LightRocket via Getty Images
Loading the Elevenlabs Text to Speech AudioNative Player...

Yet even investments on this scale look less impressive when set against global spending on data centers. PwC estimates that as much as USD 31.6trn could be spent worldwide on data centers by mid-century. If AI develops faster, that figure could approach USD 50trn.

How much of this capital ultimately flows to Poland will depend on more than demand for digital services and artificial intelligence. Access to energy, the ability to connect new facilities quickly to the power grid, access to processors and policies on digital sovereignty are becoming increasingly important.

Whatever happens, the scale of investment is growing. Poland will be part of that wave.

“According to the report, ‘The Economic Impact of the Data Center Sector in Poland,’ data centers generated PLN 10.6bn (EUR 2.5bn) in gross value added and PLN 4.3bn (EUR 1bn) in fiscal revenues in Poland in 2025. The sector also supports almost 41,000 jobs a year across the Polish economy. That is more than Poland’s entire aviation industry,” says Kinga Barchoń, partner at PwC Poland and real-estate sector leader.

Outlook for Poland. Data centers ride the wave

The scale of the forecasts for Poland is significant. Under the baseline scenario presented in PwC’s Global Data Centre Outlook, cumulative investment in data centers in the country is expected to reach USD 75.3bn in 2027-36. By 2050, it is projected to total USD 200.8bn. This puts Poland 11th among the 17 European countries analyzed. The figure is close to those forecast for Finland and Denmark, while exceeding the projections for Belgium, Turkey, Greece, Portugal and Czechia. Poland nevertheless ranks below Europe’s largest markets, including Germany, France, the UK and the Netherlands.

Against global spending, Poland is only a small dot on the map. PwC estimates that, under its baseline scenario, global investment in data centers will total USD 31.6trn by 2050. Annual spending is not expected to decline after the initial wave of infrastructure construction, but to continue rising – from around USD 800bn in 2026 to USD 1.1trn as early as 2030 and roughly USD 1.8trn in 2050.

“Poland’s continued development and the pace of its economic transformation will depend directly on how wisely and farsightedly the state secures the conditions for the growth of sectors that could become engines of the country’s future competitiveness. Massive investment in energy – particularly in new generation capacity, grid modernization and the development of clean, green energy – could provide another boost to the Polish economy, while also creating room for the development of modern, energy-intensive industries, including data centers,” adds Kinga Barchoń.

Interactive chart icon Interactive chart

Tens of trillions for AI infrastructure. More than railways

The USD 31.6trn figure represents the realistic scenario. If AI adoption is slower than expected, cumulative investment could amount to around USD 22trn. If, however, artificial intelligence is deployed more rapidly, investment could approach USD 50trn.

The scale of the entire cycle is also exceptional compared with previous waves of infrastructure investment. PwC calculates that railway construction in the UK and the US between 1840 and 1900 was worth the equivalent of about USD 0.72trn in 2025 prices. The initial phase of internet development between 1996 and 2001 is estimated to have cost around USD 3.29trn in today’s money.

The difference, however, is not only one of scale. Previous infrastructure cycles generally followed a similar pattern. Spending peaked during the construction phase, after which lower levels of investment were sufficient to maintain the existing infrastructure. Data centers are expected to follow a different dynamic.

Good to know

An AI gigafactory in Poland

As we reported on September 8, two Polish companies, DL Invest and Beyond.pl, are considering submitting a bid to build an AI gigafactory in Poland. “We see this segment as a natural extension of our presence in commercial real estate, particularly in the industrial and manufacturing sector. I can therefore confirm that we are working on participating in the European AI Gigafactories program and are ready to compete to bring one of these projects to Poland,” said Dominik Leszczyński, founder and CEO of DL Invest.

The project involves the construction of a vast data center expected to cost several billion euros.

XYZ

The building is built once. The equipment is replaced again and again

The most important component of future spending will not be the buildings themselves, but the servers. According to PwC, ICT equipment currently accounts for around 70% of total data-center investment. By 2050, its share is expected to rise to 93%. Servers, networking equipment, storage systems, processors and, above all, GPUs become obsolete far more quickly than buildings, power infrastructure or fiber-optic connections.

PwC assumes that equipment will be replaced, on average, every four to six years. For a data center operating for 20 years, this could mean three to five full investment cycles in ICT infrastructure. According to the report, every USD 1 spent on building the physical infrastructure of a data center could, over the long term, imply around USD 12 in subsequent spending, primarily on computing and telecommunications equipment.

Expert's perspective

Demand for AI will be strong

Poland already has strong technological capabilities and a large pool of specialists. We can also see companies moving from testing AI to deploying it in specific business processes. The next barrier will be access to adequate computing infrastructure.

And this is precisely where the biggest opportunity lies. If Polish companies, including SMEs, can access substantial computing power more easily, closer to home and on competitive terms, some projects that are currently difficult to justify economically may become commercially viable. The point, then, is not simply to build a huge data center in Poland, but to use it to enable more Polish companies to develop products, automate processes and compete effectively with businesses that today have much easier access to AI infrastructure.
Servers alone will not create innovation. But a lack of them can effectively stop those who are ready to innovate.

The US will capture nearly half of the entire market

Investment in data centers will not be distributed evenly. Under the baseline scenario, the Americas are expected to attract a combined USD 16.5trn by 2050. That represents more than half of global spending. The United States alone accounts for USD 15.1trn—around 48% of the entire global market.

The second-largest region will be Asia-Pacific, with around USD 8.2trn in cumulative investment. Europe is expected to attract about USD 5.6trn. The Middle East will account for around USD 1.1trn, while Africa will attract about USD 255bn.

In practice, this means Europe will account for less than one-fifth of global spending, even though its share of global GDP is higher. The reverse is true in the Americas. The US advantage stems from the concentration of the largest cloud providers, AI model developers, technology companies and the advanced-semiconductor ecosystem. If the pace of AI adoption accelerates, cumulative investment in the Americas could rise from the baseline USD 16.5trn to USD 27.1trn.

In Asia-Pacific, sensitivity to the pace of AI development is even greater in percentage terms. With faster adoption of artificial intelligence, spending could rise by as much as 69% compared with the baseline scenario. With slower adoption, it could be 34% lower. Europe is much less responsive. Even under the accelerated AI-development scenario, its investment is expected to be only 23% above the baseline.

Europe has the demand. The problem is its ability to build

Europe’s relatively modest growth, given the global scale of the market, is not, according to PwC, the result of weak demand for data centers, but of supply constraints. The report says that even if AI adoption were to accelerate, Europe would not be able to increase investment to the same extent as Asia or the Americas. That is because the baseline scenario already factors in infrastructure constraints.

One example is the mature data-center hubs of Western Europe, which are running into problems with access to grid connections, land and energy. Nordic countries, by contrast, benefit from factors including a higher share of renewable energy, a cooler climate and lower electricity costs. According to PwC, energy prices in Nordic markets are 40-50% lower than in other parts of Europe.

Poland is therefore competing not only with the largest established hubs, but also with markets seeking to turn access to energy into an advantage over traditional locations.

For Central and Eastern Europe, however, another factor could create an opportunity: digital-sovereignty policy.

“The data in the report confirm this. Under the digital-sovereignty scenario, established hubs such as Ireland, the Netherlands and Germany lose some of the cross-border flows they currently handle, while Poland records a 34% increase in cumulative capital expenditure compared with the baseline scenario. This places Poland eighth globally among the countries that benefit most from this scenario. Within Europe, only Turkey and Greece perform better under the digital-sovereignty scenario. Further market growth will, however, depend on investment in energy infrastructure and on the ability to meet the rising demand for power generated by AI and GPUs,” says Kinga Barchoń, partner at PwC Poland and real-estate sector leader.

Digital sovereignty does not stop investment

The report also examines a scenario in which public administrations and regulated sectors become less willing to rely on foreign infrastructure to handle critical data. This includes government systems, financial services, health care and some AI applications.

Under the digital-sovereignty scenario, total investment through 2050 falls from USD 31.6trn to USD 29.5trn, a decline of 6.7%. The shift, however, is mainly regional. Africa gains around 12% compared with the baseline scenario, Asia-Pacific 7% and Europe 2%. The Middle East loses around 9%, while the Americas fall by as much as 17%.

For Africa, this translates into an increase in cumulative investment from around USD 255bn to USD 284bn. In the Americas, the opposite occurs: investment falls from USD 16.5trn to USD 13.7trn. The US market alone is expected to lose around USD 2.9trn in investment under this scenario. Some workloads currently handled from the United States would be relocated to the countries where the demand originates. The Middle East falls from around USD 1.1trn to USD 1trn.

Expert's perspective

The demand is there

For us, AI is the foundation on which we are building further solutions to automate companies’ finance and accounting functions. Investments such as an AI gigafactory are needed.

The benefits of this project, however, would extend far beyond access to infrastructure itself.

A gigafactory could attract capital, specialists and further technology investment to Poland, while also creating demand for solutions developed by local startups and technology companies. This could create a flywheel effect: building the infrastructure would help AI companies grow; the next step should bring new skills and capital, which in turn would encourage more AI-related projects to be located in Poland.

Europe faces challenges

Europe as a whole changes only marginally, but beneath the surface there is a redistribution of investment. The UK gains thanks to the scale of its public and financial sectors, while Poland and Turkey benefit from a greater need to handle sovereign workloads locally. Ireland, the Netherlands and Germany, by contrast, lose out because under the baseline scenario they handle a larger share of traffic originating in other countries.

“It is worth noting that Europe as a whole shows only a marginal change compared with the baseline scenario. The aggregate regional figure, however, masks a significant redistribution within Europe. With its large economy, membership of the European Union and NATO, and rapidly expanding digital sector, Poland is one of the few European markets to benefit from this shift,” adds Kinga Barchoń.

For Poland, the shift amounts to as much as 34%.

“Poland also offers some of the lowest data-center construction costs in Europe – lower than in mature Western European markets such as Frankfurt, Dublin or Amsterdam,” Barchoń says.

As she points out, Poland is undergoing an energy transition, with heavy investment in new sources of power and in expanding the transmission grid. In addition, the national grid operator is actively preparing infrastructure to meet future demand from data centers.

“Let us not forget that the cooler climate is also an advantage. Poland has some of the lowest average temperatures in Europe outside the Nordic region. This significantly reduces the amount of energy required for cooling and lowers operating costs. Taken together, these factors create an additional, distinctive value proposition that is difficult to replicate elsewhere on the continent,” adds Kinga Barchoń.

The biggest threat? Restricted access to chips

A further escalation of the trade war over the most advanced semiconductors would have very different consequences. PwC analyzed a scenario in which export restrictions between the US and China are expanded. Access to advanced GPUs also deteriorates in a larger number of countries, while restrictions are extended to some critical raw materials.

The initial impact would be severe. By 2030, annual global investment in data centers would fall to around half the level projected under the baseline scenario. Only later, as supply chains adjusted, would investment begin to make up the lost ground. Interestingly, by 2050 annual spending would already be 8% higher than under the central scenario. This would reflect a catch-up effect as previously postponed projects were finally implemented.

That would not, however, be enough to recover the lost years. Cumulative CAPEX through 2050 would amount to USD 25.5trn rather than USD 31.6trn, a shortfall of around $6trn. That represents a decline of roughly one-fifth of the value of the entire market.

No region would be immune to this scenario. The Middle East would suffer the largest percentage decline, at 29%. Africa would lose 24%, Europe 23%, Asia-Pacific 22% and the Americas 16%. In absolute terms, investment in the Americas would fall from USD 16.5trn to USD 13.8trn. Asia-Pacific would decline from USD 8.2trn to USD 6.4trn, Europe from USD 5.6trn to USD 4.3trn, and Africa from USD 255bn to USD 193bn.

The most resilient markets would be those with strong positions in semiconductor supply chains. In Asia, PwC points to Taiwan, Japan and Singapore, among others. In Europe, Germany would remain relatively resilient, mainly because of the importance of domestic companies in supplying precision equipment and materials used in the production of advanced semiconductors.

Energy is becoming more important than capital itself

PwC identifies five main factors that will determine where new investment is located: access to energy; connectivity and transmission latency; security and the ability to host data in a trusted jurisdiction; access to GPUs and a developed technology ecosystem; and regulatory predictability and public acceptance.

The first of these is the most important. In many locations, the constraint is not a lack of financing or available land, but access to transmission infrastructure, substations and transformers. PwC notes that some grid equipment comes with lead times of several years. This can determine whether a project gets under way on schedule.

AI makes the problem more acute. GPUs require more energy than infrastructure used for traditional cloud computing, while increasing server-rack density also requires additional investment in cooling.

“AI infrastructure is becoming one of the most important capital-allocation challenges for the next generation. It brings together technology, energy, real estate, supply chains, regulation and financing. This changes the way infrastructure investors should think about capital requirements, risk and returns. Poland benefits from rapid digitalization, the growth of cloud computing and increasing investor interest,” adds Kinga Barchoń.

Examples cited by PwC show that energy is already creating competitive advantages between markets. Nordic countries benefit from energy prices that are 40-50% lower than in some other European countries. Outside Europe, the report points, among others, to Kenya, where around 95% of electricity supplied to the grid comes from renewable sources.

Key Takeaways

  1. The global data-center boom will be one of the biggest investment cycles of the coming decades. PwC estimates that global spending could reach USD 31.6trn by 2050 under its baseline scenario, and nearly USD 50trn if artificial intelligence is adopted more rapidly. What makes this market distinctive is that investment does not end once a facility has been built. ICT equipment accounts for an ever larger share of total costs, with its share expected to rise from around 70% today to 93% in 2050. Servers, processors and GPUs need to be replaced regularly, meaning that a single data center can generate successive waves of investment for many years.
  2. Poland could attract a significant share of European investment, but access to energy and grid infrastructure will be essential. Under PwC’s baseline scenario, data-center investment in Poland is expected to reach USD 75.3bn in 2027-36 and USD 200.8bn by 2050. Poland benefits from relatively low construction costs, a cool climate, a growing digital market and investment in energy infrastructure. A shift toward greater digital sovereignty could be particularly favorable, with investment in Poland rising by 34% compared with the baseline scenario. At the same time, market growth could be held back by constraints on connecting new facilities to the grid and by the growing energy demands of AI infrastructure.
  3. The location of future data centers will increasingly be determined by energy, access to chips and technology policy, rather than capital alone. Europe has strong demand for digital services, but less capacity than the US or Asia to expand supply rapidly because of constraints on energy, land and grid connections. An even more serious threat would be a tightening of restrictions on trade in advanced semiconductors. Under such a scenario, global investment through 2050 would fall from USD 31.6trn to USD 25.5trn. This shows that the success of investment in AI infrastructure will depend on a combination of energy policy, secure access to GPUs, regulation and governments’ ability to create stable conditions for investors.