Artificial intelligence is changing the requirements for data centers
NDC-GARBE 03/09/2024
Infrastructure and cooling: Artificial intelligence is changing the requirements for data centers
Artificial intelligence (AI) has long since found its way into both private and professional everyday life – according to the McKinsey Global Institute (MGI) there is a lot of potential in generative AI in particular. According to the institute’s forecast, AI tools theoretically enable an annual productivity increase of 2.6 to 4.4 trillion US dollars worldwide. The prerequisite for this increase is high-performance data centers in which the individual applications and services run. The need for much higher computing power. In order to meet this requirement, the infrastructure of the buildings must adapt accordingly.
Set up more powerful cooling systems
With new high-performance processors, a lot of computing power can be quickly achieved in less space. However, this compression results in higher temperatures that exceed the absorption capacity of traditional air cooling systems. We therefore advise data center operators to look for alternative ways to cool their racks: liquid cooling. It is now considered an enabling technology for AI in data centers, because it requires less space for heat dissipation than air cooling, thus creating additional usable space for even more racks.
Due to a continuously closed circuit, Direct Liquid Cooling is also generally convincing in terms of sustainability, as liquid transfers the waste heat from the racks better. This optimises it for feeding into the district heating networks and further reduces the CO2 footprint of data centres.
Consider AI in overall data center architecture
The choice of cooling system affects the overall architecture of data centers and data center operators should take it into account as early as the building planning stage. It requires its own and thus additional water circuit for cooling. In addition, operators must prepare or convert all racks accordingly. In the future, it will therefore be necessary to take much greater account of the overall architecture of IT infrastructure, cooling and supply technology as early as the planning and construction of data centers.
Various initiatives within the industry are currently trying to move closer to a uniform standard for the construction of AI data centers. This standardization is necessary in order to accelerate the expansion of the necessary infrastructure through a common consensus and clear specifications.
Is it worth retrofitting existing data centers for Artificial Intelligence?
Of course, everyone wants their own data center to meet the new requirements within the shortest possible time. It is not only in terms of sustainability that operators should consider whether they can convert existing buildings with advice from developers . However, this decision takes into account a number of considerations regarding permits, supply and location.
Make your data center ready for AI
Artificial intelligence will continue to be one of THE defining topics in the data center industry for a long time to come. It remains to be seen what the optimal AI data center will look like. In principle, NDC-GARBE’s Green IT Cube concept focuses on developing buildings flexibly and sustainably so that both air and liquid cooling can fit in. We want to make a decisive contribution to high-performance, resource-saving and reliable infrastructure in order to create the backbone for an AI-supported digital future.
Learn more about NDC-GARBE as your partner for AI-ready data center development under: https://www.ndc-garbe.com/ndc-garbe-provider-for-data-centre-development-as-a-service-dcdaas/