As AI drives higher rack densities and more dynamic power demand, the role of in-rack energy storage is expanding. High-rate nickel-zinc technology brings a combination of power density, rapid response and superior safety and sustainability well suited to the evolving needs of AI infrastructure. Brandon Smith, VP of Global Sales and Product, ZincFive, tells us more.
For years, the role of the Battery Backup Unit (BBU) was relatively straightforward: provide backup power to keep the rack operating through a power interruption or until another source could take over.
AI is expanding what that job requires. Today’s high-density AI infrastructure is not only operating at significantly higher power levels, but also reshaping how power must be delivered, managed and supported. Thousands of GPUs can transition together between compute-intensive and communication-intensive tasks, creating rapid changes in demand that occur on millisecond timescales. At scale, those synchronised fluctuations can extend beyond individual servers and racks, placing additional stress on upstream power infrastructure.
That creates two distinct requirements at the rack: backup power when primary power is interrupted and dynamic power response when workloads change. As racks become increasingly populated with high-value GPU and xPU infrastructure, the ability to maintain consistent power delivery becomes increasingly important to overall system utilisation and performance. It is also forcing the industry to reconsider what in-rack energy storage should be capable of delivering.
The BBU is entering a new era
Rack-level backup is not new. Hyperscalers have deployed BBUs for years and they remain an important component of many distributed power architectures.
What’s changing is the environment around them.
As rack densities increase and AI workloads become more dynamic, designers are evaluating energy storage against a broader set of requirements. Backup runtime still matters, but so do power density, response capability, cycle performance, safety, sustainability, footprint and the ability to integrate into rapidly evolving rack architectures.
Today’s technologies tend to address different parts of that equation. Traditional BBUs are designed primarily around backup runtime. Capacitor-based systems are well suited to extremely short, high-frequency power events, but provide limited energy duration and scalability. The result is an increasingly important design question around how to put the right power capabilities closer to the compute without introducing unnecessary complexity or risk.
There may not be one answer.
Different workloads and architectures will require different approaches to backup and dynamic power support. That makes the underlying energy storage technology increasingly important.
Why nickel-zinc fits the rack
Nickel-zinc battery technology brings a combination of characteristics that align well with the evolving requirements of rack-level energy storage.
First is a balance of power density and usable energy storage. As rack power requirements increase and available space remains constrained, the ability to deliver high power while providing sufficient energy for rack-level applications becomes increasingly valuable. These characteristics can help support a range of emerging rack-level energy storage requirements within modern AI infrastructure.
Second is high-rate charge and discharge capability. AI workloads can create rapid, repeated changes in power demand. NiZn is particularly well suited to short-duration, high-power applications that require energy storage to respond quickly and repeatedly. That capability is valuable for traditional backup, but it also creates opportunities for battery-based energy storage to play a more active role in supporting dynamic AI power demands.
Third is safety. Moving batteries closer to increasingly dense and valuable compute makes battery chemistry a fundamental architectural consideration. As power infrastructure moves closer to compute, system designers must evaluate both performance and operational risk. Energy storage technologies that avoid thermal runaway may simplify deployment, help support evolving rack architectures and reduce concerns associated with placing batteries near high-value compute infrastructure. Nickel-zinc uses a nonflammable, aqueous electrolyte and has no risk of thermal runaway at the cell level, offering a compelling alternative for designers as they evaluate how to safely integrate energy storage into modern AI infrastructure. That safety profile is backed by UL 9540A testing, reinforcing NiZn’s suitability as energy storage moves closer to critical compute infrastructure.
Fourth is sustainability. As data centre operators expand AI infrastructure, the environmental impact and material lifecycle of the technologies supporting that growth are becoming increasingly important. Nickel-zinc uses abundant, highly recyclable materials and delivers 25–50% lower lifecycle greenhouse gas emissions than lead-acid or lithium-ion alternatives, giving designers a more sustainable chemistry without compromising the performance requirements of modern AI infrastructure.
Together, the combination of power capability, usable energy storage, high-rate performance, superior safety and sustainability makes nickel-zinc particularly well suited to the evolving requirements of rack-level energy storage. These are also the attributes at the core of ZincFive’s dedicated solutions for both traditional backup and AI dynamic power applications.
Bringing nickel-zinc into the rack ZincFive has spent years applying high-rate nickel-zinc technology to immediate power applications in mission-critical data centres. We are now extending that technology from the UPS layer into the rack. The development of our NiZn In-Rack Power Solutions is centred on two dedicated solutions.
Bringing nickel-zinc into the rack
ZincFive has spent years applying high-rate nickel-zinc technology to immediate power applications in mission-critical data centres. We are now extending that technology from the UPS layer into the rack. The development of our NiZn In-Rack Power Solutions is centred on two dedicated solutions.
The ZincFive Battery Backup Unit (BBU) is being developed for established rack-level backup applications, bringing NiZn’s high power capability, superior safety and high-rate performance into architectures where backup power must be available immediately when primary power is interrupted.
The ZincFive AI Dynamic Power Module (DPM) is being developed to support rapidly changing AI power demands, including transient mitigation, pulse power support and other dynamic power applications.
The distinction is intentional. We do not believe every rack, workload or customer will require the same energy storage architecture.
Instead, the goal is to give OEMs, hyperscalers and data centre operators greater flexibility to deploy the capability that fits the application, whether that means traditional backup, dynamic power support, or a combination of solutions across different portions of their infrastructure.
ZincFive is developing solutions compatible with established 48V rack-level systems while also aligning with emerging sidecar and higher-voltage architectures. This approach recognises that data centre power designs continue to diversify and allows customers and partners to evaluate the right solution based on workload, rack design and power requirements.
Safety becomes more important as storage gets closer to compute
There is another reason chemistry matters more at the rack.
AI infrastructure is concentrating extraordinary amounts of power and compute into increasingly small spaces. At the same time, emerging designs are bringing more electrical infrastructure directly alongside that equipment.
That changes the risk calculation.
Energy storage that might once have been located in a dedicated battery room or gray-space environment may now reside much closer to servers, networking equipment and liquid-cooling infrastructure. In that environment, performance alone is not enough. Safety increasingly influences not only chemistry selection, but also how and where energy storage can be deployed.
Nickel-zinc’s lack of thermal runaway gives system designers another option as they evaluate how to safely integrate energy storage into increasingly power-dense architectures. For a growing number of operators and jurisdictions, battery safety is non-negotiable, with codes, standards and site requirements increasingly limiting where certain battery chemistries can be deployed. Industry codes and standards such as NFPA 855 and IFC Section 1207 for Electrical Energy Storage Systems place increasing emphasis on thermal runaway behaviour, fire propagation testing and system-level safety considerations. Technologies with favorable safety characteristics can therefore give engineers greater flexibility in designing next-generation power architectures.
For ZincFive, that safety profile is backed by established testing and certifications: our NiZn batteries have demonstrated no thermal runaway in UL 9540A cell-level testing, while ZincFive’s existing data centre battery systems comply with multiple industry-recognised safety standards. It also moves the conversation beyond simply asking how much power can fit into a footprint, towards a more complete question like what technology can deliver the power performance, operational reliability and safety the application demands.
That question will become more important as rack architectures continue evolving.
Designing for where AI infrastructure is going
There is unlikely to be a single blueprint for the AI data centre.
Some operators will continue to rely heavily on centralised UPS architectures; others will push more energy storage capability into the rack. Some AI environments will require dedicated transient mitigation. Emerging sidecar and higher-voltage architectures will create more possibilities still.
What is becoming clearer is that energy storage can no longer be considered only in the context of an outage.
AI is expanding its role from an insurance policy that waits for something to go wrong into a more active component of the power architecture. The technologies that succeed in that environment will need to deliver more than backup runtime, they will need to balance
performance, safety, sustainability, operational flexibility and evolving infrastructure requirements as AI architectures continue to mature.
For ZincFive, bringing nickel-zinc into the rack is a natural extension of what technology already does well. With over 2 gigawatts of ZincFive power solutions delivered or contracted globally, we are applying proven experience from mission-critical data centre deployments to a new layer of power architecture and working with OEMs, hyperscalers and industry partners to determine how NiZn can best support the next generation of AI infrastructure.
Backup power isn’t disappearing, but its role within the rack is expanding. The technologies that support it must evolve accordingly.
