Over the past year, we have seen growing adoption of automation and early movement towards increasingly autonomous data center infrastructure. Systems can be optimised in real time, using predictive analytics to unlock new operational efficiencies, improve uptime and reduce waste. It is worth distinguishing between automation, predictive optimisation and true autonomy: most data centers today sit somewhere along that spectrum, rather than having reached full autonomy.
Alongside the increasing sophistication of AI models, the self-managing capabilities of data center infrastructure will be a major driver of growth moving forward. According to one market estimate from Marketintelo, in 2025 the autonomous AI data center infrastructure market was valued at $18.7 billion. By 2034, it’s expected to $163.4 billion, representing a forecast compound annual growth rate of 26.3%. Greater automation will be an important part of meeting rising demand while improving resource efficiency.
The shift towards more autonomous infrastructure could significantly reshape data center operations. But that doesn’t mean that humans are out of the loop. They still play a key role in providing checks and balances in the system and, more significantly, using the latest advances in technology to make strategic decisions that support resilient, efficient data center operations.
Moving forward, human-machine collaboration has the potential to be the driving force behind data centers. But operators must strike the right balance between human input and autonomous systems to make the most of this collaborative dynamic.
The (reverse) centaur
There are two contrasting models for human-machine collaboration, drawn from automation theory: the centaur and the reverse centaur, representing two ends of a spectrum rather than the only possible outcomes. These categories are derived from the mythological creature, which is part human, part horse. In this metaphor, we substitute the horse for machine intelligence.
So what is a centaur in this context? The centaur has the head and torso of a human and the body of a horse. This means the human is in control, guiding its machine body based on operational insights and ensuring autonomous systems function safely and effectively. The machine assists the human in achieving their desired outcomes.
In a reverse-centaur model, the relationship is inverted: the machine effectively determines the course of action, while people are relegated to executing or supervising its decisions without sufficient authority, context or control. AI agents could act independently without checks and balances.
For example, a data center system optimising cooling for energy efficiency should operate within agreed temperature limits and refer exceptions to an operator. Human oversight is needed to ensure that an efficiency gain does not compromise equipment reliability.
As this example shows, the potential implications of autonomous systems acting independently are significant. It is incumbent on operators to adopt a human-led operating model: one where autonomous systems are directed and moderated by humans, rather than the other way around. In practice, this means defined decision rights and permissions, continuous monitoring, clear escalation thresholds, human override capability, and safe fallback mechanisms built into every layer of the operation.
It is incumbent on operators to adopt a human-led operating model: one where autonomous systems are directed and moderated by humans.
Human-first operations
The best kinds of collaboration play to the different parties’ strengths, finding the synergies between them and unlocking new opportunities in the process. The same goes for autonomous data center operations.
Human operators are good at joining the dots, understanding context, and dealing with the unpredictable nature of physical infrastructure. However, they are also limited and fallible. They can’t be everywhere at once, covering all the aspects of a data center, or multiple data centers that are networked across the globe. And they do make mistakes, especially when they are spread too thin or have to work for long periods of time.
Machines in contrast are designed for continuous operation. When Information Technology and Operational Technology systems are properly connected, they can provide a more comprehensive, near-real-time view across distributed operations, whether infrastructure is local or overseas, and analyse volumes of data beyond human capacity, spotting patterns in that data to make actionable predictions. This can help address human limitations.
A centaur is so much more than a human in the loop. It is a sophisticated combination of organic and digital intelligence, acting seamlessly to deliver the best possible outcomes for data center operators. Getting the balance right starts with a clear set of principles: what can be safely automated, what requires human approval, how exceptions are escalated, how override and fallback mechanisms work, and how teams are equipped and trained to act on machine-generated insight. Done right, there are huge benefits associated with harnessing machine intelligence for human decision-making.
Preparing for the future
Autonomous data center operations have yet to fully mature. Systems will become more advanced, unlocking ever-greater efficiencies and new capabilities. For Hitachi, this means connected intelligence across IT, OT, power, cooling, building systems, security and service operations: bringing these domains together to improve resilience, uptime, efficiency and predictability. The operators who continually embrace these new advances across IT and OT will be the ones that thrive in a market where scale and performance are paramount.
As autonomy increases, poorly governed decisions or system errors can carry greater financial, operational and reputational consequences.
Looking ahead, ensuring data center infrastructure is operated as a centaur (not a reverse centaur) is essential. The organisations that succeed will be those that combine machine speed and scale with human judgement, accountability and operational expertise: human-led autonomy, underpinned by the ability to connect intelligence across IT and OT.
The organisations that succeed will be those that combine machine speed and scale with human judgement.