Currently, the amount of data processed worldwide doubles about every eight months. New technologies, such as artificial intelligence and autonomous driving, are not even included in this calculation and will increase these data volumes even further. Even today, a networked car that does not drive autonomously processes data to the tune of 25 gigabytes per hour. In autonomous vehicles, this amount of data can be in the petabytes range (1 petabyte = 1,000 terabytes). This impressively illustrates the future increasing development of data volumes.
If Germany does not want to lose touch economically, it must push ahead with the expansion of its digital infrastructure, consisting of data networks and data centres. The figures prove that this has not been tackled decisively enough in recent years. Germany's share of global data center capacities has decreased: While in 2015 about 3.5 percent of servers were in data centers, this share fell to just under 3 percent by 2022. Accordingly, it is a market with great growth potential, which international operators are currently recognizing for themselves: with the ongoing and planned projects in the data center industry, capacity in Germany could double.
The biggest consumers of IT capacity are currently hyperscalers and cloud providers such as AWS, Google Cloud, Microsoft Azure, Oracle and IBM Cloud. They use the capacities for their business models – from online services to financial services to the provision of cloud storage. However, there is also a great demand from regional data centers for companies to have greater control over their own hardware. The financial sector is also demanding high capacities with its banking applications. In contrast, public administration has so far made little use of capacity – but with the increasing digitization of the sector, increasing demand is also expected here.
Different types of data centers
Data centers are specialized properties that enable the operation of IT infrastructure such as servers, networks, storage, and applications for enterprises, governments, and organizations. Various applications are operated in a data center, such as cloud computing, web hosting, e-commerce, big data analytics, and mobile applications. In terms of cost, data centers are made up of four essential components: the suitable property, the building, the technical building equipment and the IT systems. While the building and land usually represent less than a third of the investment costs, the majority of the costs (over two thirds) are accounted for by electrical, refrigeration and security technology.
Data centers can be divided into six different types based on various parameters. So-called hyperscalers are developed specifically for cloud services and large, scalable workloads. This type of data center is operated by large companies such as Google Cloud, AWS, Microsoft Azure, Oracle, and IBM Cloud.
Several companies rent space in colocation data centers . These are usually operated and marketed by an operator. Many large companies that have previously operated their own smaller data centers are increasingly switching to colocation data centers and renting as subtenants, among other things for cost reasons. Other reasons for this could also be the focus on the company's own core business or the retention of flexibility in terms of growth in the IT landscape.
The third group is made up of enterprise data centers. These are classically operated by companies themselves for their own IT as corporate data centers. However, this type is not economical for many companies and involves a lot of effort, which is why they switch to other data centers in whole or in part. The fourth category, edge data centers, are small decentralized units near end users, end devices or applications that are needed with a view to autonomous driving, smart cities, the Internet of Things and Industry 4.0. The reason for this is that they have low latency and high data processing speed due to their physical proximity, which is essential for the aforementioned areas.
Telecom data centers are operated by companies in this industry to ensure network connections, data transmission, and processing.
The sixth category is made up of high-performance computing data centers: These are specifically designed for applications such as scientific simulation, research, AI training, and financial modeling. They are relevant for companies that work with large amounts of data that need to be processed quickly.
Current market situation
Germany is a predestined location for data centers. The country is the largest economy in Europe and, with DE-CIX in Frankfurt, has the Internet hub with the world's highest data transfer rate of 14 terabytes per second (Tbit/s). In addition, Germany has high data protection requirements and well-trained specialists. There are low site risks in terms of extreme weather events, earthquakes or floods and the power supply is relatively stable.
There are about 50,000 data centers of all sizes in Germany, of which about 3,000 have a capacity of 40 kW or more and about 90 with a capacity of 5 MW or more. In the case of data centers, their size is not measured in square meters, as is the case with other properties, but in kilowatts or megawatts. This is a measure of the IT performance that can be delivered. Current forecasts assume that several hundred MW of data center capacities will be created in Germany in the next five years. The trend is towards the construction of data centers with higher data and computing capacities of 5 MW or more.
At the European level, the dominant data center clusters are the so-called FLAP-D markets (Frankfurt, London, Amsterdam, Paris and Dublin). They have historically developed at the Internet nodes. In particular, the major cloud service providers AWS, Google Cloud, Microsoft Azure, Oracle and IBM Cloud have built their hyperscaler data centers at these locations.
The Frankfurt market is the German hotspot for data centers and the second largest market in Europe after London. The combined capacity of the data centers in Frankfurt is 707 MW. In the meantime, however, the location is already quite saturated, which is why Berlin is establishing itself as a new hub in the field of hyperscalers. Along with Madrid, Milan and Zurich, Berlin is one of the secondary markets that have developed very dynamically in recent years and recorded great growth. The cities mentioned already have data center capacity above the 100 MW mark. In Berlin, there are currently applications for new data centers in the order of 1,000 MW. In the colocation data center segment, the markets in Munich, Hamburg, Stuttgart and Düsseldorf in particular are developing into dominant locations.
There are many older properties on the market in Germany, but it is difficult to adapt them to current and future sustainability requirements. The range of newer data centers is still significantly limited. As investors from abroad in particular are discovering the German market for themselves, the number of project developments is increasing and will lead to significantly higher product availability in both smaller and larger data centers in the future.
The situation on the investment market
For institutional investors, this comparatively new investment class is increasingly coming into focus for several reasons. First and foremost are the attractive returns. These are currently significantly higher than those of classic real estate investments. Annual distribution yields are about six to eight percent or more. In addition, long-term and indexed leases are concluded for data centers with high rents due to demand. This leads to a high level of income security and stable cash flow for investors. In addition, the asset can be brought into the portfolio as infrastructure or real estate and serves as an addition to the risk diversification of one's own investment portfolio. Especially since data centers have a corresponding cyclical independence.
The low level of transparency on the market is still causing a certain reluctance in some cases. Data centers are not standardized, each object is an individual building and follows an individual business case. This makes comparisons difficult. Basically, each data center should be considered individually. Compared to common real estate asset classes, the transaction market for data centers has not yet been as liquid. It is therefore important for investors to build up internal expertise or to bring in an investment manager with experience in the segment.
In general, data centers are very capital-intensive projects. In the case of large hyperscalers, the investment sums quickly run into the billions. Operators and companies cannot do this alone. Additional capital is needed here. Especially in the case of large hyperscale data centers, it is estimated that one third of the investment costs are attributable to the land and the building. However, the majority of the investments will go into technical equipment and infrastructure for power, emergency power, cooling, connectivity and security.
The investor usually pays for land and real estate, sometimes also for the technical equipment. Racks – i.e. server cabinets, IT systems, network and security infrastructure and, in some cases, maintenance – are paid for by the tenant, depending on the contractual relationship.
The transaction structures themselves are essentially divided into three types: Powered Land, Powered Shell and Fully Fitted. Powered land refers to the acquisition of a plot of land with electricity, often at a high price, for the project planning and development of a new data center. This is currently the most common transaction structure. Powered Shell is the purchase of a plot of land including a building, but without technical building equipment and IT systems for renting. The acquisition of a technically fully developed data center excluding IT systems is referred to as fully-fitted. Operation and rental are then the responsibility of the owner.
There are two variants of the rental models: the single-tenant and the multi-tenant model. In the single-tenant model , the operator/tenant usually assumes not only the space rent, but also the operational responsibility for the data center, including maintenance and repair of the technical infrastructure. He also bears the risk of failures or damage. In the multi-tenant model , leases are concluded for individual racks. The rents for such racks range from 500 to 2,000 euros per month, depending on the criteria. Maintenance and repair of the technical infrastructure is the responsibility of the operator of the data center.
Important factors in investment
The selection criteria for data center locations differ significantly from those of other types of use. The availability of energy (power supply capacity) is particularly important, and in the future especially green electricity. The connected load must also be high enough for the data center capacity. Latency/connectivity, i.e. proximity to the next node as well as to other data centers, is also important. Since data centers are critical infrastructure, high security requirements apply – for example, for site security (protection against earthquakes, flooding), which must be observed.
Sustainability requirements favour new properties
A key factor that is becoming increasingly important for operators is the sustainability of data centers. Worldwide, there are increasing activities to make the construction and operation of data centers as sustainable as possible. A key indicator for this is the energy efficiency of data centers, which is given as Power Usage Effectiveness, or PUE value for short. It describes what proportion of the total energy consumed is actually used for IT systems and how much has to be spent on the technical building infrastructure (cooling, lighting, other consumers). The value thus shows how efficiently a data center uses energy as a building. The closer the value is to 1.0, the more efficiently the data center works.
The German PUE average is 1.8. Energy efficiency has improved significantly in recent years. While the average PUE value of new data centers was still 1.98 in 2010, it fell to 1.55 in 2022. The current Energy Efficiency Act (EnEfG) provides for a maximum PUE value of 1.5 for data centers with a nominal electrical connected load of 300 KW or more that start operation before July 1, 2026. By July 1, 2030, the value of these data centers must be reduced to 1.3 on an annual average. Data centers that go online from July 1, 2026 must achieve an energy consumption efficiency of less than or equal to 1.2.
The EnEfG also sets requirements for the reuse of waste heat. When commissioning from 1 July 2026, the mandatory minimum share for the reuse of energy will be ten per cent, from 1 July 2027 it will be 15 per cent and from 1 July 2028 it will be 20 per cent – in each case two years after commissioning. It is possible to use the waste heat itself or to transfer it to the municipality by feeding it into the local heating network. However, the latter is subject to a certain degree of uncertainty, as many municipalities have not yet made much progress in the planning of their municipal heating network and are hardly able to answer corresponding inquiries about feeding waste heat into this network. For this reason, it is advisable to develop alternative variants for the use of waste heat in order to meet the specified quotas from the EnEfG.
In general, new buildings usually meet the stricter requirements. In turn, it is difficult to redesign old existing data centers so that they meet the requirements of the law. That's why investors should only consider new or like-new and efficient properties.
View
There are now numerous real estate investors who are intensively involved with data centers and want to invest in this segment for the first time. These include the entire range of insurance companies, pension funds and pension funds, both nationally and internationally. In particular, the growth forecasts for future data volume are attracting investors. The increased interest rate level, the low activity on the real estate transaction markets and the falling demand for office space are also contributing to the fact that data centres are increasingly becoming the focus of real estate investors and are seen as an alternative. It is important to note that an investment in these special properties is not comparable to a classic real estate investment. In particular, the legal and technical requirements are very different. Investors should therefore choose an investment manager with expertise in the construction and operation of data centers.
