The world today is locked in an Artificial Intelligence (AI) race, and Singapore is working to leverage its position as a tech hub in the Asia-Pacific to be a leader in this technology. For one, over SG$20 million (US$14.8 million) will be allocated in the next three years to increase the number of SG Digital Scholarships and overseas internships in AI roles.
The government has also pledged to invest up to SG$500 million (US$370 million) to secure high-performance computing (HPC) resources for AI innovation and capability building. These initiatives are just two examples of how the country is aiming to establish itself as a critical player in the global tech landscape.
However, meeting the demands of AI workloads will require accelerated IT and high-performance computing (HPC) architectures. That is because power consumption and heat generation from AI chips are significantly higher than conventional IT equipment. For data centres that power AI applications, traditional cooling technologies might not have the capabilities to address these workloads.
Meanwhile, the power is subject to voltage sags, spikes, and complete failure during transmission and distribution, which therefore may interrupt computer operations, cause data loss, and damage equipment. As we are gearing towards AI compute age, knowing what cooling and power methodology and solution is key to ensure we keep the AI compute operating at their optimum performance without degradation or disruption.
Singapore’s unique challenge
It is a pivotal moment for the collective effort to find solutions that will shape Singapore’s future. The demands for AI coincide with rising power challenges and decarbonisation requirements.
For instance, Singapore’s heavy reliance on natural gas for electricity — approximately 95 per cent of its energy supply — poses a barrier to the integration of renewable energy sources. This dependence limits the ability to diversify energy portfolios and reduce carbon footprints effectively.
Compounding this issue are Singapore’s limited land area and high population density, which present obstacles for large-scale renewable energy projects. With little room for expansion, finding solutions that balance energy needs with sustainability goals becomes increasingly complex.
Unlike many of its regional peers, who also have high population densities, Singapore also has to contend with limited land area. That means limited space for data centres that are pivotal to AI workloads, as there is no shortage of competing uses for land. Singapore also has to grapple with limited freshwater resources.
Also Read: GenAI’s twisted impact on the creative world: Navigating chaos to find new order
These unique circumstances imply that the country’s policymakers have had to be cautious about how to encourage data centre growth. The recent lifting of its moratorium on new data centres is proof of the selectiveness needed for the realities facing Singapore. Simply put, the country needs more data centres to continue to support technology innovations but not at the cost of resource efficiency.
Rethinking infrastructure and cooling methods
These challenges put the industry at an inflexion point. In regard to thermal management, for instance, there will need to be more deployments of liquid cooling solutions, as air-cooling technologies alone cannot efficiently support higher-density workloads. With the demands of the digital space set to escalate, it is crucial for data centres to adapt continuously, leveraging innovative solutions and operational efficiencies to meet the future head-on.
Since standard cooling systems cannot handle the heat created by GPUs based on the latest demands for AI computing, most installations will necessitate the introduction of the newest liquid-cooling technologies into the data centre and eventually enterprise server rooms. In the near future, rack densities exceeding 100 kW per rack will become widespread and on a massive scale, necessitating significant capacity improvements throughout the entire power train, from the grid to the chips in each rack.
Significant investments are required to modernise the infrastructure required to power and cool AI systems. Incorporating liquid cooling technology into uninterruptible power supply (UPS) systems is also essential for maintaining a steady power supply, which is necessary for coolant distribution’s mechanical demands. The ideal temperatures for facilities and equipment can be maintained by dependable cooling systems.
In terms of energy, there are emerging bring-your-own-power (BYOP) options to meet growing data centre demands. The role of the uninterrupted power supply (UPS), for example, can transform into a dynamic energy resource that creates an “always on” environment. Till now, data centres have largely been purely electricity consumers.
However, with recent advancements, future data centres could be a stabilising element to the local grid. The ability to shift loads, to start fuel cell systems, and to share energy from large battery energy storage (BESS) will also change how data centres move from consumers to partners in stabilising the future grid.
Also Read: Embracing AI in Southeast Asia: The strategy for avoiding cost overruns
Partnerships within the industry can also facilitate efforts to address these challenges. For instance, Vertiv is collaborating with NVIDIA to offer expertise in critical power and cooling infrastructure with advanced computing capabilities. This provides a comprehensive solution that includes high-performance power and cooling technologies to address the demands of AI applications, including high-density heat removal and energy efficiency.
Becoming future ready
With AI adoption critical to continuous innovation, the need for robust, AI-ready infrastructure solutions will be necessary. This includes pre-engineered systems that can be deployed up to twice as fast as traditional setups, allowing businesses to respond swiftly to market changes and technological advancements.
Despite significant challenges, Singapore has all the potential to become an AI hub. That will, however, depend on addressing sustainability concerns, leveraging forward-looking infrastructure, and prioritising continuous learning. The good news is that the country already has the potential to realise these imperatives. Through its continued commitment to resilience and a focus on innovation and sustainability, Singapore can realise its vision and set a precedent for others in the region and beyond.
—
Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.
Join us on Instagram, Facebook, X, and LinkedIn to stay connected.
Image credit: Canva Pro
The post The AI age is changing the data centre industry – Here’s how Singapore can pivot appeared first on e27.