Last week, we attended Data Centre World (DCW) Singapore. As well as allowing us to meet with customers and prospects, the event brought together industry leaders, operators, and vendors to shine a spotlight on one of the industry’s most critical and evolving challenges: power resilience.
As data centres scale to meet AI-driven demand, the pressure to deliver reliable and sustainable power has never been greater. Across two days of seminars and on-stand discussions, one message was clear: power resilience must be tested and proven.
Here, Paul Brickman, Commercial Director at Crestchic, shares his key takeaways from the event:
High-density racks make thermal testing even more critical
The rise of AI, HPC, and accelerated workloads is pushing rack power densities upward, in many cases exceeding 100 kW per rack. It’s no longer enough to design for high capacity on paper; facilities must prove they can deliver it safely and efficiently under real conditions.
By simulating the electrical and heat output of high-density racks before live deployment – using a loadbank – operators can validate power distribution, cooling performance, and redundancy strategies without risking downtime or hardware damage.
Liquid cooling takes centre stage
Liquid cooling was also a hot topic – and for good reason. As rack density continues to climb, direct-to-chip cooling is an effective way to manage the extreme thermal loads generated by AI and high-performance environments.
To ensure these systems perform as intended, facilities are turning to water-cooled loadbanks during commissioning. These loadbanks simulate the electrical and thermal output of high-density racks, allowing engineers to test power delivery, heat removal, and system response safely and accurately.
Resilience requires testing, not just redundancy
A strong theme at DCW Singapore was that redundancy alone doesn’t equal resilience. Power resilience must be verified, not assumed.
Loadbank testing allows operators to validate UPS systems, generators, transfer switches, and entire backup circuits under controlled load conditions, ensuring they behave as expected if power fails. Regular testing identifies issues long before they cause an outage.
Beyond mechanical resilience, many operators are embedding energy storage systems and microgrids to provide independent fallback and grid support. Together, these approaches create a layered energy safety net. However, while resilience is increasingly built into the initial infrastructure, it still must be proven through structured, periodic testing.
Energy Storage and microgrids are the new normal
With grids across Asia-Pacific under increasing pressure, the conversation is shifting from “backup power” to “energy ecosystems.” Battery Energy Storage Systems (BESS), hybrid genset solutions, and on-site renewables are redefining how data centres stay online during grid fluctuations or outages.
Loadbanks play an essential role in testing and commissioning BESS installations – verifying that a system can respond instantly to grid frequency changes, deliver the required discharge rates, and meet its operational specifications.
Beyond commissioning, loadbanks also support revenue optimisation for operators participating in fast frequency response (FFR) services. By discharging safely into a loadbank, BESS units can maintain optimal charge levels and avoid degradation, ensuring both reliability and profitability.
As energy storage becomes central to data centre strategy, rigorous loadbank testing will be key to validating both performance and resilience in the face of growing grid volatility.
Collaboration, skills, and a culture of testing
DCW Singapore’s power sessions echoed a growing realisation: even the best hardware fails without well-trained teams and disciplined testing routines. Business continuity planning, including crisis simulations and recovery protocols, must be understood by engineers and operational teams.
Our team can play a key role in building that expertise by delivering training to in-house teams. This collaboration ensures that testing is not just a tick-box exercise, but a genuine transfer of skills that help support ongoing uptime goals. If required, experienced engineers can attend site to run controlled load tests and commission systems, helping to fill skills gaps. By adopting a culture of continuous validation, and building partnerships with skilled partners, data centres can reduce the potentially catastrophic impact of a power failure.
Final thoughts: Making resilience routine
Our conversations at DCW Singapore 2025 suggested a shift in mindset: resilience is not a one-time design goal; it’s an ongoing practice. Ultimately, power resilience is about having the confidence that, when the power stumbles, business systems, people, and plans will keep the lights on. Chatting with people from across the data centre sector, reinforced what we already know: true confidence only comes from testing, proving, and improving – again and again.