What the Data Centre Boom Can Teach Us About Systems Leadership
The media is flooded with stories about AI and the data centres required to power it.
Like any radical change at this scale, it is tempting to reach for the narrative of ‘unprecedented times’. That is partly true, but it is also a convenient story. If something is unprecedented, we can tell ourselves it is impossible to understand, let alone anticipate.
Systems leadership requires the opposite. It means paying attention to dispersed signals and asking what they reveal when viewed together. Right now, capital, workplaces and communities are being pulled into the same AI investment cycle at once. What I’m observing is that most leaders are still reading the resulting pressure points as separate stories.
Over the past two weeks, Macquarie Asset Management and Singapore's sovereign wealth fund GIC agreed to finance a new venture called Theseus Infrastructure. It has been established to construct billions of dollars of data centres in the US for Anthropic, owner of the Claude AI model, ahead of a reported $1.4 trillion public listing. The memo of understanding didn't specify how many data centres would be built, where, or at what total cost. It is understood to involve thousands of megawatts of new capacity. The structure exists to relieve Anthropic of the financial risk of development. Macquarie ran a similar, although certainly more robust, process with AirTrunk, which it built from a single site into Australia's biggest data centre operator before selling its stake at a $24 billion valuation.
The Theseus deal shows capital committing before the detail is explored, let alone settled. Billions of dollars, thousands of megawatts, no confirmed sites, no total cost. At the same time, the RBA has warned that the scale and speed of data centre investment is becoming an inflation risk. Investment is moving faster than the workforce, materials and grid capacity required to absorb it. Demand is arriving in months while much of the supply response takes years.
The same pattern is playing out at a community level. Elena Schlossberg, a Virginia activist who has spent a decade tracking the costs of data centre development, was invited to a Washington roundtable with a visiting Australian premier, alongside data centre developers and energy lobbyists. A day later she was quietly uninvited, via her status changing from "going" to "not going", with capacity constraints given as the reason.
The three separate arenas of capital markets, monetary policy and a community meeting on the other side of the world are trying to tell us something. I think it’s a story about an unconscious shift in the way we are working. Moving at the pace the opportunity demands and conveniently downplaying the risk and scrutiny that would normally slow a decision down for interdependency evolution.
Three years working inside energy organisations has given us a particular vantage point on signals like these. We are not energy experts. Our work at Interchange is understanding how people and organisations navigate complex change, and the energy transition has provided a front-row seat to what happens when technical, economic and human systems begin colliding. You start to see the human patterns underneath the technical ones. The way leadership teams respond when the ground shifts beneath them. The conversations that happen in the room and the ones that don't.
The AI and data centre boom is making that collision visible in plain sight. For a long time, we've trained leaders as though leadership were an individual sport. In reality, the system is much stronger than any one individual. We readily accept this when it comes to ecosystems. The strength of a tree is inseparable from the environmental conditions in which it grows. Inside organisations however, we continue to attribute behaviour overwhelmingly to the qualities of individual leaders.
The average leader has five to eight direct reports, but their decision-making is shaped by far more than their own judgement. It is primarily shaped by what the organisation around them rewards, punishes, permits and most importantly, refuses to discuss. That's why the same person can behave very differently from one organisation to the next. Their character hasn't necessarily changed, the system they are a part of has. A leader's job, then, is not simply to manage the task. Rather, it is to create enough containment around the group's anxiety that people remain capable of thinking while they do it.
Systems thinking & Psychodynamics
This is where systems thinking and systems psychodynamics offer related but importantly different lenses. Systems thinking helps us see how the parts interact. Systems psychodynamics asks what happens to the humans inside that system when those interactions generate anxiety. Systems psychodynamics emerged from the foundational work of the Tavistock Institute and scholars such as Wilfred Bion. It proposes that when anxiety rises, groups can default to what Bion called basic assumption behaviour. These are collective, largely unconscious responses that can feel like purposeful action but function as forms of avoidance, with fight, flight, dependency and pairing among the most common. They are visible in boardrooms, leadership teams and, if you know what to look for across entire industries. My own research on social defences extends this thinking into a modern-day context whereby the desire for pace is leaving discipline in the dust.
The physical system gives us another way of seeing the same problem. The tech sector invested a record $12.3 billion in equipment and machinery over the past year. Anthropic alone plans to buy at least 1.4 gigawatts of capacity from Australian data centres that will cost up to $21.6 billion to build. The scale of that investment is wild, but the speed at which it is arriving is more interesting. Billions of dollars of new demand is being committed in months to physical systems whose workforce, energy and infrastructure capacity takes years to expand. Economist Nik Dawson captured the physical constraint very nicely with "You can import the chips, but you can't import the wiring." Every imported server still needs Australian electricians, plumbers and technicians to install and cool it, and that workforce takes years to develop. That gap between demand and supply has reach far beyond data centres, it is an economy and society wide problem.
The sparky workforce alone can act as a systems indicator. Greater Western Sydney has acquired approximately 6,000 electricians in recent years, while planned data centre projects will require at least another 2,500 of them. The issue to consider is what happens when billions of dollars of private investment into data centres competes with other industries for that skillset. The risk is displacement when sparkies attracted to higher-paying infrastructure projects disappear from housing, hospitals, schools, grid upgrades and regional communities. The data centre solves its workforce problem while another part of the system inherits it.
Each individual decision is rational. The electrician takes the better-paid job. The developer pays what it needs to secure scarce skills. The hospital, school or housing project tries to remain within budget. Collectively those rational decisions can shift capability away from essential infrastructure and increase labour costs across the economy enhancing our existing productivity problem.
This is what Jarrett and Vince describe in their research about the regressive dimension of radical change. They share that the forces a transformation generates in the surrounding system are rarely visible to the people driving it. The system consequences are the product of collective action, not individual intent. That distinction matters enormously for leaders. Systems leadership therefore asks a different question from the one most boardrooms are currently asking. The conventional question is: how do we adopt AI fast enough to stay competitive? The systems question is: what is the full environment inside which that adoption is occurring, and what pressures are we generating or inheriting that we have not yet considered or accounted for?
The data centre boom offers at least three lessons.
Lesson one: Read beyond the dominant frame
Victoria provides an interesting example. The state was recently ranked last in the nation as a place to do business. The Business Council of Australia's latest Regulation Rumble report put it eighth of eight jurisdictions, with BCA chief executive Bran Black stating plainly, as reported by Sumeyya Ilanbey in the AFR: "Victoria is at the back of the pack for business tax and regulatory settings." At the same time, global capital is flowing into Victoria for data centre development at extraordinary scale.
The logic driving regional Victoria's appeal is pretty straightforward, land, water, evolving infrastructure and connectivity via subsea cables. The Latrobe Valley has existing high-voltage infrastructure created through decades of coal power generation that can now potentially be repurposed. It has land at scale. It has connectivity. It sits within an energy system transitioning toward the renewable capacity that large-scale compute increasingly requires.
Victoria looks uncompetitive through one analytical frame while appearing unusually attractive through another. Lawlor and Sher write about the tendency of organisations to ‘mistake the map for the territory’, becoming so invested in their existing metrics and models that they lose the capacity to perceive what those models cannot capture.
The systems leadership lesson is not that one analysis is right and the other wrong. It is that both can be true. The work is resisting the urge to resolve the contradiction too quickly and asking what becomes visible when you hold both versions of the system at once.
Lesson two: Notice where the pressure goes
The second lesson is already visible in the electrician example. The data centre solves its workforce problem while another part of the system inherits it. Jarrett and Vince's idea of the regressive dimension of radical change helps explain why this matters. Transformation creates pressures beyond the boundaries being considered by the people driving it. Those pressures do not disappear simply because they sit outside the project plan or strategy. Systems psychodynamics adds another dimension. Fitzsimons, Petriglieri and Petriglieri describe how anxiety can migrate through organisations, becoming embedded in structures, processes and decisions far from where it originated. The same is true of pressure more broadly. A system will find somewhere for it to go.
This is where pace becomes particularly interesting. Moving quickly can look like progress from inside the part of the system doing the moving, while capability gaps, risk, cost or anxiety are simply being transferred somewhere less visible. For leaders, the question is therefore not only "How quickly can we move?" It is "Where is the pressure created by our movement going?"
That question applies just as readily to organisational transformation as it does to data centres. What looks like successful acceleration in one part of an organisation may be creating overload, workarounds, resistance or risk somewhere else. Seeing those consequences before they become someone else's problem is part of the work of systems leadership.
Lesson three: Pay attention to what the system cannot tolerate
The third lesson takes us back to Elena Schlossberg being invited into the room and then removed from it. Systems psychodynamics encourages us to pay attention not only to what a group is discussing, but to what it is defending itself against. Bion's basic assumptions describe some of the ways groups unconsciously manage anxiety rather than engage with the task in front of them. Schein similarly describes how shared assumptions can protect groups from anxiety and become increasingly difficult to see precisely because they are so deeply held.
Christopher Bollas offers another useful idea: the "unthought known", the things we know at some level but have not yet been able or willing to think about consciously. In organisations, these dynamics can show up in what isn't said, which questions repeatedly get deferred, which information is rationalised away, or which voices become inconvenient when the pressure rises. Systems leadership requires more than technical competence. It requires enough psychological presence to remain curious when the information arriving is uncomfortable, contradictory or threatening to the direction already chosen.
One of the most useful diagnostic questions a systems leader can ask is therefore not simply, "Who is in the room?" The more courageous and uncomfortable question is "Whose presence, perspective or information is the system finding difficult to tolerate?" That’s because the information a system works hardest not to see may be the information its leaders most need.
The data centre boom is not, at its core, a technology story. It is a story about what happens when one part of a system changes faster than the institutions, infrastructure and people around it can adapt. That same pattern is playing out inside organisations everywhere and it’s worth paying attention to.
The leadership challenge is to remain capable of thinking while everything around you is accelerating. It’s the ability to hold contradictory information without rushing to resolve it, to notice where pressure is being displaced, and to remain curious about the things the system would rather not discuss.
Systems leadership begins with a deceptively simple premise. The thing demanding your attention is not always the thing you most need to understand.
What is your system finding difficult to tolerate?
References
Wiggins, J., & Rapaport, E. (2026, August 11). Macquarie, GIC to power Anthropic expansion with data centre rollout. Australian Financial Review.
Kinsella, L., & Mizen, R. (2026, July 8). RBA flags data centre boom as new threat in inflation fight. Australian Financial Review.
Ilanbey, S. (2026). Victoria ranked worst place in the nation to do business. Australian Financial Review.
Stuchbery, M. (2026, February 5). What are data centres and what do they mean for regional Victoria? ABC Gippsland.
Koley, B. (2026). Australians deserve straight answers about doing data centres the right way. Australian Financial Review.
Gardner, J. (2026, August 12). An AI activist was about to enter the lion's den. Then this happened. Australian Financial Review.
Bazalgette, J. (2018). Leadership: The impact of the full human being in role. In E. Aram, R. Baxter, & A. Nutkevitch (Eds.), Adaptation and innovation (pp. 31–50). Routledge.
Bion, W. R. (1961). Experiences in groups and other papers. Tavistock Publications.
Fitzsimons, D., Petriglieri, G., & Petriglieri, J. L. (2011). Defensive organizing: How leaders' unregulated anxiety shapes collective defences in organisations. INSEAD Working Paper.
Fotaki, M. (2021). The psychoanalytic contribution. In D. Lawlor & M. Sher (Eds.), An introduction to systems psychodynamics: Consultancy, research and training (1st ed.). Routledge.
Jarrett, M., & Vince, R. (2017). Psychodynamics of strategic leadership. In M. Jarrett & R. Vince (Eds.), The Oxford handbook of organisational change and innovation. Oxford University Press.
Kets de Vries, M. F. R. (2006). The leader on the couch: A clinical approach to changing people and organisations. Jossey-Bass.
Lawlor, D., & Sher, M. (2022). An introduction to systems psychodynamics: Consultancy, research and training. Routledge.
Petriglieri, G., & Petriglieri, J. L. (2020). The return of the oppressed: A systems psychodynamic approach to organisation studies. Academy of Management Annals, 14(1), 411–449.
Schein, E. H. (2004). Organisational culture and leadership (3rd ed.). Jossey-Bass.
Wells, L. (1985). The group as a whole: A systemic socioanalytic perspective on interpersonal and group relations. In A. D. Colman & M. H. Geller (Eds.), Group relations reader 2. A. K. Rice Institute.
