The deeper engineering and data-science capability targets cooling optimisation and critical-asset analytics across mixed-vendor estates, reinforcing an advisory-first model that keeps human operators in control of every recommendation.
CETA System Co., Limited, a Hong Kong-incorporated technology company, announced in September 2023 that it has strengthened its data-centre artificial-intelligence engineering and data-science capability to support deeper cooling optimisation and critical-asset analytics across mixed-vendor environments. With cooling accounting for as much as 40% of total energy consumption at a typical data centre, the analytical quality of any optimisation platform carries direct weight on operating cost and resilience, and the company has deliberately prioritised capability depth over headcount expansion.
Data centres remain operationally unforgiving, with human error implicated in a substantial share of major outages that industry estimates place at two-thirds to four-fifths of incidents. Against that reality, domain expertise and engineering judgement carry material weight in any AI deployment, and the expanded capability is designed to reinforce the advisory-first model that operators expect when they evaluate AI-driven platforms for live production environments, where analytical depth, transparent reasoning and operator oversight matter more than autonomy. Engaging that expertise early, rather than retrofitting design after construction, avoids the cost overruns and compromised outcomes that late changes invite.
Critical IT infrastructure cannot tolerate downtime in the way conventional systems can, because the facilities concerned underpin national security, healthcare, finance and emergency services, where a single failure can disrupt several services at once. Resilience on this footing depends less on redundancy counts than on containment, the ability to isolate a failing zone without compromising core functions, achieved through layered protection, zoning and compartmentalisation, independent detection paths and fail-safe behaviours that preserve operational integrity under degraded conditions. Viewed in those terms, resilience measures how intelligently risk is distributed, not how many redundant units are counted.
Modern data centres depend on interconnected chains spanning electrical, cooling and IT systems, where proprietary software from individual device vendors cannot fully address cross-system coordination and non-integrated platforms raise operational risk through the absence of consolidated data. The strengthened team is aimed squarely at that complexity, and as Lee Tsz-Hin, Chief Executive Officer of CETA System Co., Limited, frames it, the discipline now demands “engineers who understand how power, cooling, security and network systems interact, not simply how each functions in isolation.”
The newly deepened disciplines reflect where reliability in live operation is actually won and lost. More major cooling failures originate from controls design and implementation than from mechanical equipment, so controls and reliability engineering now sits alongside the electromechanical backbone of motors, pumps, drives and control systems that sustains cooling and airflow. Telemetry and data engineering across mixed-vendor estates supplies the data foundation, normalising signals that each manufacturer structures differently before any meaningful correlation becomes possible.
On that foundation, data science replaces fixed-threshold monitoring with models that learn the signature of normal operation across thousands of correlated variables and flag deviations early, supporting both thermal stability and anomaly detection for critical assets such as UPS systems, generators and chillers. Many failure modes develop through slow, statistically small changes that stay within acceptable limits until a fault escalates, which is why Lee describes early pattern recognition as “the difference between planned intervention and reactive response.”
Field engineers translate these analytical outputs into guidance operators can act on with confidence, completing a life cycle that runs from design inception through site improvement and ongoing maintenance. CETA System ties the work together through vendor-agnostic integration with existing building-management and data-centre infrastructure-management environments, using open-protocol connectivity across BACnet, Modbus, SNMP and API sources, with units, naming conventions and timestamps normalised so that cause and effect can be correlated across IT demand, power distribution and cooling response.
Across these capabilities the platform operates on a human-in-the-loop basis, with the operator retained in supervision and decision-making at each stage while the system surfaces patterns, flags anomalies and presents recommendations that alerts, reviews and failsafe checks verify before any action is taken. High availability is treated as a software-level responsibility rather than a hardware assumption, built on failover logic and active failsafe operating modes rather than passive redundancy alone, and Lee characterises that boundary as a design principle rather than a concession, noting that on this model “subject-matter expertise sits with the operator.”
The case for that boundary is borne out by operator sentiment, which remains well documented: across the latest industry readings, only 14% of operators trust AI to change equipment configurations and around one in three trust it to control data-centre equipment directly, while more than 70% trust AI to interpret sensor data or predict maintenance requirements. CETA System has structured its model around exactly that distinction, delivering analytical capability through recommendation and explanation while operator oversight is retained at every decision point, so that confidence is built through transparent reasoning, vendor-agnostic integration and failsafe operating modes rather than assumed.
About CETA System
Founded in 2017 and incorporated in Hong Kong, CETA System develops artificial-intelligence solutions built specifically for data-centre infrastructure. A single vendor-agnostic platform brings together HVAC and chiller-plant energy optimisation with predictive maintenance for critical assets including UPS systems, generators and chillers, integrates with existing building-management and DCIM environments, and follows an advisory-first deployment model for colocation, enterprise and hyperscale operators across the Asia-Pacific region and beyond.
- Website: https://cetasystem.com
