How Edge Analytics Combat AI Power Bursts in Data Centres

The explosive growth of AI workloads has created an unprecedented challenge for data centre operators: managing sudden, severe power fluctuations that threaten facility infrastructure and the wider electrical grid. 

As GPU servers execute intensive AI computations, they generate unforeseen energy spikes – a phenomenon known as AI power bursting – that can exceed the capacity limits of existing systems.

Intelligent power management company Eaton has developed what it describes as an industry-first solution to address this mounting concern. Through a firmware update for its Power Xpert quality (PXQ) event analysis system, the company enables data centre operators to detect subsynchronous oscillations (SSO) in real time, allowing them to take preventive action before equipment damage occurs.

The edge analytics capability arrives as data centres worldwide grapple with AI's soaring energy demands, which far surpass anything the industry has previously encountered. 

These large load fluctuations can cause transformer overheating, ferro resonant damage and other catastrophic impacts to equipment, making early detection essential for operational continuity.

Managing power at the edge

Eaton's approach uses sophisticated edge analytics directly within the PXQ meter, a versatile power quality device already deployed in switchgear, switchboards and power distribution units. 

The system can identify SSO bursts alongside its existing capabilities for detecting sags, swells, transients and harmonics – all without requiring data to travel to centralised processing systems.

This edge-based architecture is particularly significant in the context of AI infrastructure. By processing data locally, the system provides immediate insights that enable operators to respond to power quality events in real time, reducing latency and enhancing the overall resiliency of critical infrastructure.

"The energy demands of AI workloads surpass anything data centres and the grid have encountered before, with load fluctuations that can exceed the limits of existing infrastructure," says JP Buzzell, Vice President and Chief Data Center Architect at Eaton. 

"By enabling customers to harness their existing PXQ technology in new ways, we're delivering a market-first capability to effectively respond to AI power bursts."

The solution represents a major milestone in Eaton's grid-to-chip strategy, which aims to provide data centre operators with comprehensive tools to address the unique energy challenges posed by artificial intelligence. 

This strategy encompasses intelligent power distribution systems, backup power solutions and digital capabilities designed to optimise white space and grey space in AI-ready facilities.

The broader edge transformation

Eaton's innovation arrives amid a wider industry shift towards edge computing and distributed analytics. As AI inference applications proliferate, data centres are evolving from purely centralised facilities to hybrid architectures that place processing power closer to end users and data sources.

This decentralisation trend is reshaping infrastructure requirements. Rather than concentrating all AI workloads in hyperscale facilities, operators are increasingly deploying edge data centres that support low-latency applications whilst maintaining connections to centralised training infrastructure.

The associated architectural shift necessitates new approaches to power and cooling. As inference workloads move to the edge, operators must ensure that distributed facilities possess the same level of protection against power quality issues as their centralised counterparts. 

Edge analytics solutions like Eaton's SSO detection capability become essential components of this distributed infrastructure.

Grid-to-chip integration

Eaton's strategy extends beyond power quality monitoring. The company previously announced its collaboration with NVIDIA to support the transition to 800 VDC power infrastructure, enabling support for 1MW racks and beyond. This partnership addresses the escalating power density requirements of next-generation AI systems.

Eaton has also joined forces with Siemens Energy to develop a fast-track approach for building data centres with integrated onsite power generation. This collaboration reflects the industry's recognition that AI workloads require rethinking traditional approaches to power delivery and backup systems.

The convergence of edge analytics, intelligent power distribution and onsite generation represents a broad response to AI's infrastructure demands. 

Rather than addressing power, cooling and compute in isolation, leading providers are developing integrated solutions that span the entire stack from utility connection through to chip-level power delivery.

This integrated approach proves particularly valuable when managing AI power bursts. By combining edge analytics for real-time detection with intelligent distribution systems and robust backup power, operators can respond to SSO events with minimal disruption to ongoing workloads.

Adapting to market dynamics

The edge data centre market is experiencing remarkable expansion driven by these infrastructure requirements. 

According to research from MarketsandMarkets in 2025, the global edge data centre market is projected to grow from US$50.86bn in 2025 to US$109.2bn by 2030, a compound annual growth rate of 16.5%.

This growth is underpinned by rising demand for ultra-low-latency performance, real-time analytics and localised data processing across multiple industry verticals. Edge data centres enable organisations to manage the exponential increase in data generated by Internet of Things devices, 5G networks and advanced AI applications.

The infrastructure segment is positioned to capture the largest market share, driven by demand for resilient, scalable and high-performance digital infrastructure. This encompasses servers, storage systems, networking equipment, power generation and backup solutions, cooling technologies and security mechanisms.

With accelerated adoption of IoT, 5G networks and real-time analytics, organisations require low-latency, high-efficiency infrastructure to manage complex workloads. Edge data centres demand modular and flexible infrastructure designs to accommodate evolving technological requirements, ensuring operational continuity and seamless data processing.

Future-proofing data centre operations

For data centre operators, the ability to detect and mitigate AI power bursts through edge analytics represents both an immediate operational imperative and a long-term strategic advantage. As AI workloads continue to evolve and intensify, infrastructure must adapt accordingly.

The remote firmware upgrade capability of Eaton's PXQ system exemplifies this forward-looking approach. Rather than requiring wholesale equipment replacement, the solution enables existing infrastructure to gain new capabilities, extending the useful life of deployed assets whilst adding critical functionality.

This upgrade path is particularly valuable given the capital-intensive nature of data centre operations. By leveraging edge analytics within existing power quality meters, operators can enhance their protection against SSO without significant additional expenditure on new monitoring equipment.

Looking ahead, the integration of edge analytics with artificial intelligence and machine learning promises to further enhance power quality management. Predictive algorithms could potentially identify patterns that precede SSO events, enabling even more proactive responses to AI power bursts.

As the data centre industry navigates the unprecedented challenges posed by AI workloads, solutions that combine edge analytics with robust power management will prove essential. Eaton's SSO detection capability demonstrates how targeted innovation at the infrastructure level can address systemic challenges, protecting both individual facilities and the broader electrical grid from the impacts of AI-driven power fluctuations.

The convergence of edge computing, distributed analytics and intelligent power management is reshaping data centre design and operations. Operators who embrace these integrated approaches will be best positioned to support the next generation of AI applications whilst maintaining the reliability and resiliency that critical digital infrastructure demands.