Summary
- Deutsche Telekom is investing nearly €250m in the next expansion phase at Tucherpark.
- The facility is expanding from two to three floors, with planned power capacity increasing to 20MW.
- Nvidia B300 systems will supplement the existing B200 installation; the first deployment is reported to be sold out.
Deutsche Telekom is investing almost another €250m in its Munich Industrial AI Cloud, expanding the computing installation to a third floor and increasing its planned facility capacity to 20MW.
The investment follows strong demand for the initial deployment at Tucherpark, which the company says has sold out. Additional Nvidia Blackwell Ultra B300 processors will be installed alongside existing Blackwell B200 systems and other computing hardware.
Deutsche Telekom disclosed the expansion on 9 November, describing a further commitment to AI computing infrastructure within Germany. The company is also considering additional developments in Bavaria, although no separate investment decision has been confirmed for those possible projects.
The additional B300 equipment will double the number of processors in the relevant Blackwell deployment. The company expects a greater increase in computing performance because the newer systems offer higher processing capability.
The expansion will use an additional floor within the existing Munich facility rather than requiring a new data centre building. Its practical delivery will depend on installing the computing equipment and ensuring that the supporting power, cooling, and network infrastructure can accommodate it.
Deutsche Telekom’s Industrial AI Cloud provides specialised computing services for businesses and other organisations using advanced AI systems. The project was developed with Nvidia and infrastructure partner Polarise.
Computing growth increases supporting infrastructure demand
Adding high-density AI systems requires more than available floor space. Electrical equipment must deliver sufficient power to the racks, while cooling systems must remove the heat generated during operation.
Large clusters also need high-capacity networking, suitable storage, and management systems capable of coordinating equipment across multiple racks and computing nodes.
The new Nvidia processors have different performance characteristics from the hardware already installed. Operators must therefore account for their individual electrical, thermal, and integration requirements rather than assuming that newer equipment can be exchanged without affecting supporting systems.
Deutsche Telekom’s announced 20MW figure is described as the expanded capacity of the computing facility. The company has not published a complete breakdown distinguishing operating IT load from all supporting electrical consumption.
That limitation is important when comparing the development with other data centre projects whose published capacities may refer to grid connections, total facility demand, or power delivered directly to computing equipment.
Commissioning additional hardware within an operating facility also creates maintenance and service continuity requirements. Installation work must be coordinated with systems already serving customers, and new equipment needs to be tested before it is placed into commercial service.
Customer demand supports further investment
The reported sale of the initial deployment provides a commercial justification for expanding the installation. It indicates that the operator has already secured demand for its earlier computing capacity rather than relying solely on forecast growth.
However, selling the first deployment does not establish the revenue that will be generated by the new capacity. Deutsche Telekom has not disclosed the full terms of its customer commitments or the expected payback period for the latest investment.
AI infrastructure financing involves substantial expenditure on processors, networking, electrical equipment, and cooling. The cost structure differs from that of a conventional data centre whose owner primarily leases space and power to customers supplying their own servers.
Deutsche Telekom’s model combines substantial investment in computing hardware with the physical systems required to operate it. That creates a need to align equipment procurement, customer contracts, and infrastructure commissioning.
Additional capacity elsewhere in Bavaria remains under consideration. Any further site would require its own assessment of suitable premises, power supply, connectivity, and investment economics.
The Munich expansion represents a commitment to a larger operating installation at an existing location. The next measurable development will be the commissioning of the additional B300 equipment and the availability of the expanded computing services to customers.
