The rash - the energy hunger of intelligent data processing

Rapid technological advances are leading to a rapidly growing demand for intelligent, powerful and secure data processing: from recognizing patterns in private music streaming, to highly specialized AIs to detect fraud in the financial sector, to deep learning image recognition software that detects cancer and other diseases at an early stage. However, what is economically a boom in the cloud market is ecologically a heavy burden on the environment. This is because the growing demand for computing power in research, business and everyday life also increases the consumption of resources and energy, including the electricity required to operate the necessary IT infrastructures. Added to this is the significant increase in waste heat from air-cooled computing capacities, most of which is released into the air - potential that has not yet been fully exploited.

Requirements - Efficient cloud infrastructures for a sustainable digital future

In order to resolve the conflict between high computing power and sustainability, innovative technologies are needed that improve the climate balance across all sectors. This can be achieved by setting up a decentralized, green cloud infrastructure, converting existing IT infrastructure and developing innovative green IT concepts.

Sustainable technologies such as direct hot water cooling and the reuse of server waste heat can enable green cloud operation. Energy-efficient server cooling enables the reuse of waste heat from data centers, e.g. for heating buildings or via connection to local and district heating networks.

By setting up a decentralized infrastructure, energy optimization beyond the boundaries of a data center can also be achieved in the future. Computing-intensive AI loads can be distributed and processed according to global energy efficiency parameters at locations where, for example, renewable energy is available or where there is demand for waste heat. Realizing this vision requires automated, scalable and robust software solutions, which we are continuously working on.

The role of partnerships & the ecosystem

With the help of strong partnerships, synergies can be optimally exploited to increase the energy efficiency of digital infrastructures. For example, it is possible to connect to heat sinks and use waste heat efficiently. In the Cloud&Heat data center in the Eurotheum (Frankfurt), waste heat is used directly on site to heat the local office and conference rooms, hotels and restaurants. Another example is the cooperation with energy companies to enable the use of surplus heat from modular AI data centers in the adjacent district heating system.

The development of cloud software for the operation of a virtual, distributed data center benefits from sound skills and experience. One example is the open source lifecycle management Yaook, the development of which was initiated by Cloud&Heat and STACKIT and is now part of the newly founded ALASCA association. The project is an important building block for realizing the vision of distributed infrastructures and the development of open source technologies to strengthen digital sovereignty in Europe. Cooperative projects and initiatives that aim to standardize cloud components (e.g. Gaia-X or the Sovereign Cloud Stack) are also essential for the efficient management of heterogeneous IT resources.

Process for implementing the solution

The energy optimization of digital infrastructures is a continuous process. As already mentioned, Cloud&Heat currently operates data centers that are cooled efficiently and in which the waste heat is also used sensibly. The model calculation at the 111-metre-high Eurotheum, the former headquarters of the European Central Bank, in Frankfurt's city center shows the CO2 and cost-saving potential of using server waste heat to heat buildings. Companies from the tech sector and AI users use an IT infrastructure here that is tailored to their individual needs and meets the highest security standards. Every computing operation supplies the building with heating energy and hot water thanks to the water cooling system. The calculation shows that an average of 40% or €255,000 is saved in energy costs. This means a saving of 710 tons of CO2 - which would require 56,800 deciduous trees to compensate for an area equivalent to 90 soccer pitches in Germany.

A conclusion - Sovereign and sustainable cloud for the development of AI applications

If we want to exploit the potential of AI, we need to find common ways to minimize the negative ecological impact on our environment. Through many years of experience, Cloud&Heat has developed a broad expertise in various areas related to the planning, construction and operation of energy-efficient, open cloud infrastructures that create the optimal conditions for the operation and development of AI applications.

By continuously developing our technologies in a strong partner ecosystem, we are getting one step closer to the vision of a green and digitally sovereign cloud every day.

More information: www.cloudandheat.com / https://thinkgreen.cloudandheat.com/

Whitepaper:
https://www.cloudandheat.com/wp-content/uploads/2023/03/2019-12-16_Whitepaper-Einsparpotenzial.pdf
https://www.cloudandheat.com/wp-content/uploads/2023/03/2021-CloudHeat-White-Paper-2-de-rev10.pdf

 

Picture: Cloud&Heat Technologies GmbH