Infrastructure as the first act of Empire
We said that we would create a series of essays around the topics of work, power and the future of both so here it is. The idea is to pull from bits and pieces of our history and classical theory to understand the vulnerabilities of future societies.
Firstly, empires rarely begin with conquest, they begin with connection.
Long before borders are fixed or subjects named, power lays down pathways: routes along which people, goods, messages, and authority can move with increasing ease. Only later does domination harden into law or violence. The first act of empire is infrastructural, not ideological.
This was as true for the Roman Empire as it is for the networked powers of the present. Roman expansion did not proceed by marching blindly into the unknown. It advanced by paving it.
Our global road networks are often remembered as feats of engineering. However, their deeper power was what they made possible. Roads enable us to connect faster, transport everything and allow new flows of information. This is happening today at an unprecented scale with AI infrastructure.
In the past connection preceded control. Control followed naturally.
Two thousand years later, the most consequential infrastructure of the modern world is not visible from the ground. It lies beneath oceans, threads through cities, and terminates in anonymous buildings humming at the edges of industrial parks. Submarine fiber-optic cables now carry the flows that once travelled by road: orders, wealth, intelligence, culture.
These cables are the new imperial substrate.
They do not simply enable communication; they determine whose communication is fast, whose is slow, whose is reliable, whose can be interrupted. They define the topology of the modern world not as geography, but as access. Just this week we see users on X begging Elon Musk for starlink access in Iran. Whoever holds the power to connect holds the flow of information.
In this landscape, distance matters less than latency. Power no longer depends on how far something is, but on how long it takes to reach. A region today becomes marginal when its data routes are slow, congested, or indirect.
Latency is the new frontier.
To reduce latency is to pull the periphery closer to the centre. To control chokepoints, landing stations, exchange hubs, cloud regions, is to exercise a form of sovereignty without territory. Empires once fought over passes and ports. Today, they worry about cable routes and redundancy.
What appears as technical optimisation is in fact political ordering. Many believe that Elon Musk and others who hold the keys to the cloud only recently became interested in politics but when you see the vast networks of infrastructure these billionaires have built and accumulated it could only naturally lead to their influence in global political spheres.
Neutral infrastructure is misleading. Roads were never neutral. They privileged certain movements over others, certain cities over others, certain futures over others. Roman roads did not simply allow travel; they directed it. They made some paths inevitable and others irrelevant.
Fiber-optic networks do the same.
In 2026, the global fiber-optic landscape has reached a historic fever pitch, with nearly 40 new submarine cables expected to go live this year alone. This “great plumbing upgrade” is no longer driven by traditional telecom companies but by the insatiable hunger of hyperscalers like Google, Meta, and Amazon, who now control over 70% of the market to fuel the explosive growth of generative AI. This shifting ownership has transformed digital infrastructure from a public utility into a potent geopolitical weapon; routes are being aggressively diversified to bypass volatile chokepoints like the Red Sea and South China Sea, while “AI corridors” now dictate global power as network connectivity “chases” the massive electrical grids required for data centers.
Decisions about where cables run, which regions host data centres, whose traffic is prioritised, and whose platforms become default are not merely economic. They determine whose labour is visible, whose speech circulates, whose markets integrate, and whose lives are synchronised with the dominant system.
Infrastructure does not follow empire. Empire follows infrastructure.
This inversion is easy to miss because connectivity feels benign. Roads promise trade. Cables promise communication. Platforms promise participation. Yet each promise comes with an asymmetry: movement becomes easier in one direction than another; dependence grows faster than autonomy.
Rome did not need to conquer everyone at once. It needed only to make non-participation increasingly costly.
The same pattern defines contemporary power. Regions and populations are drawn into global systems not by force, but by necessity. To work, to trade, to speak, to be counted, one must connect. To connect is to accept protocols, standards, and dependencies set elsewhere.
Movement becomes inevitable. Exit becomes unthinkable.
Empires collapse when they can no longer maintain their connective tissue. Roman roads fell into disrepair not because stone decayed, but because coordination did. The routes still existed, but they no longer led reliably to power.
The future of empire will be decided less by armies or ideologies than by infrastructure: by who builds it, who maintains it, and who depends on it to live ordinary life.
All roads once led to Rome.
Today, they lead to the cloud.
And the cloud, like Rome, does not announce itself as empire only as the way things work.
