Chindata Group, a leading carrier-neutral hyperscale data centre solution provider in emerging Asia-Pacific markets, launched the full-stack solution for data centre power and cooling issues of high-computational servers at the Artificial Intelligence Generated Content (AIGC) era.
Hyperscale data centres face the dual challenge of meeting substantial computing demands while ensuring energy efficiency. The rapid proliferation of AI technologies has amplified the industry’s pursuit of enhanced computing capabilities, faster delivery, and an environmentally-conscious expansion. This growth brings increased power demands, necessitating innovative power supply, distribution, and cooling solutions.
To address the current power and cooling issues, Chindata, in collaboration with its partners, unveiled four comprehensive hyperscale data centre solutions that can meet diverse customer computing needs amid the current surge in AI technology usage:
X-PrePower 3.0 innovatively integrates the equipment of the power supply link, simplifying power pathways and reaching an impressive overall efficiency of 98.5%. The X-PrePower 3.0 realises prefabrication and factory integration. Compared with the traditional power distribution mode, it saves 20%+ floor space and reduces the on-site delivery time from 2 months to 1 week, effectively solving the challenges of long deployment cycle, low energy efficiency, and high operation and maintenance costs of data centre power supply and distribution systems under the surge of computing power, and ensuring the safety and reliability of the whole life cycle of power supply.
X-Cooling Maglev Two-phase Cooling System combines magnetic levitation(maglev) technology, refrigerant pump technology, distributed multi-connection technology, evaporative condensation technology, thin-fanwall technology, phase-change heat transfer technology, DC fan technology, and AI intelligent group control technology, organically solving the technical challenge of high power density air cooling in single racks ranging from 10 to 35kW.It also solves the defect of centralised cold water system low reliability, long construction cycle, and complex debugging. At the same time, it can break through the constraints of building conditions, climate, and water sources, and organically combine with cold plate liquid cooling to achieve flexible adaptation. Distributed system with matrix redundant architecture patented is more stable and reliable; The system Cooling Load Factor(CLF) can achieve an average annual rate of 0.08 in North China and 0.15 in Malaysia.
Large-Scale Fully Prefabricated Data Centres have pioneered a new swift, quality-centric delivery mode, combining domestic factory prefabrication and on-site steel structure construction oversea, which substantially reduces the construction timeline of cross-broader projects. The data centre project currently under construction in Malaysia, adopting the overall modular design and modular assembly , saw its timeline reduced to just eight months, from foundation building, comprehensive testing, delivery to installation, setting a new industry benchmark.
Kunpeng IDC Operations Management Platform offers an all-in-one platform for data centre operations, management, monitoring, and control with the goal of promoting the implementation of a standardised operation and maintenance system.
Binghua Zhang, The CTO of Chindata Group, said, “In AIGC-era, our comprehensive full-stack solutions provide a new technical framework and evolution route for data centres in the era of LLM (Large Language Model). With technology innovation and partners’ agile collaboration, Chindata breaks the innovation boundary of data centres by leading design concepts. Leveraging multiple technologies integration and innovation, we enable the rapid development of intelligent computing industry. In the future, Chindata will continue to embrace the industry ecosystem, collaborate sincerely with industry partners, engage in joint innovation, to drive the continuous implementation of cutting-edge data centre technologies , provide more classic cases for the industry’s rapid iteration and build a better future.”