My core argument is straightforward: frameworks forged during the Cold War are not merely outdated—they can become actively dangerous as guides to a world being reshaped by artificial intelligence. Today’s strategists risk preparing to fight yesterday’s conflicts, while tomorrow’s power dynamics are being framed inside data centres—and, one day, potentially in outer space.
Much public discussion about AI fixates on one subject above all others: the AI financial bubble. Commentators warn of market overvaluation, speculative hype and unsustainable capital flows into generative‑AI firms. Those warnings are reasonable. But focusing only on market mania creates a collective blind spot. Financial bubbles come and go; the structural transformations driven by artificial intelligence for governance, state power and global civilisation are potentially permanent. We spend endless hours debating whether AI is overpriced, yet we rarely ask whether AI has already begun rewriting the core rules of international relations—and whether our strategic imagination has failed to keep pace.
This article explores five interlocking puzzles. None delivers definitive answers. All are meant to challenge prevailing assumptions held by policymakers, security analysts and business leaders.
Why must AI’s new power currency remain chained to planet Earth?
For generations, geopolitics was largely energy geopolitics. International competition determined oil supply routes, pipeline diplomacy and the dominance of national power grids. State strength was measured in megawatts and access to fossil fuel reserves.
Artificial intelligence has created a new currency of power: computational intelligence. Why must that intelligence be permanently confined to Earth’s surface?
Space‑based AI data centres are not imminent. Major practical obstacles remain: high launch costs, space‑debris risks, thermal limits for high‑density computing hardware, Earth‑orbit bandwidth and latency constraints, plus the vulnerability of orbital assets to anti‑satellite weaponry. If orbital compute infrastructure is built, it will itself become a new domain for great power rivalry.
Still, technical possibilities are advancing faster than our strategic policy models. Within two decades, orbital computing could become viable for specialised missions: sovereign backup data storage, high‑security isolated processing, or operations leveraging near‑unlimited in‑orbit solar energy. In space, certain terrestrial constraints disappear: local power shortages, ground‑grid collapse and geographic supply chokepoints no longer apply.
Regardless of if such a future is even plausible, nearly all of today’s mainstream energy security strategies are built exclusively around terrestrial assumptions. Energy geopolitics will not disappear; it may simply migrate into competition over orbital assets and space resources. Yet this scenario barely features in our official strategic assessments.
Can national power grids remain the unchallenged backbone of sovereignty?
Centralised national power grids rank among the great engineering achievements of the industrial age. They underpin modern state sovereignty and deliver irreplaceable public goods: cross‑regional load balancing, disaster‑emergency resilience and affordable energy access for vulnerable populations. The complete elimination of large‑scale grid backbones is highly unlikely.
But AI enables a new spectrum of options: distributed, autonomous energy systems that can, in time, be supplemented by space‑derived power. What happens if these systems become a serious complementary layer of the global energy supply?
Partially decentralised, intelligently managed energy could erode some long‑standing strategic advantages held by resource‑rich nations and energy transit states. Some legacy infrastructure assets may lose their historic centrality. This will not end resource‑driven geopolitics. New rivalries will emerge around rare‑earth minerals, battery inputs and semiconductor supply chains. Old forms of leverage decline, while new ones arise.
The critical question for strategists is whether we are planning for this full range of futures—or continuing to design policy almost solely for a central‑grid world that may not define the century ahead.
Are Cold‑War‑era buffer zones still fit for an AI‑driven security landscape?
Cold War geopolitics relied heavily on containment, buffer zones, territorial deterrence and zero‑sum territorial partitioning. International tensions were stabilised by drawing lines on maps.
AI‑enabled conflict adds powerful new security dimensions. Algorithmic influence operations, AI‑augmented cyber‑attacks and automated disinformation campaigns pay little respect to physical borders. A malicious algorithm can destabilise national markets, large‑scale data operations can reshape public opinion, and intelligent influence tools can shift strategic decision‑making—all without a single soldier crossing a frontier.
Nevertheless, digital power complements rather than replaces physical threats. Territorial disputes, conventional military risks and attacks on ground‑based critical infrastructure remain real. Buffer zones will not vanish overnight, but they risk losing much of their historical weight.
The puzzle for debate is this: why do we still treat Cold‑War‑style territorial frameworks as our primary security template? Important parts of the old territorial‑balancing toolkit face systemic strain, yet policymakers continue applying them as if they offer complete answers to 21st‑century threats.
Shared sovereignty: Are nation‑states still the unrivalled holders of power?
Traditional international relations rest on a foundational axiom: the nation‑state holds supreme sovereign authority. Is that axiom still unconditionally valid in the age of large AI platforms?
Major AI multinationals command datasets, algorithmic infrastructure and cross‑border communication channels that outmatch the resources available to many sovereign states. They shape public perceptions, influence domestic governance and move capital and information across borders faster than most state bureaucracies can respond.
We must acknowledge powerful counter‑trends. These corporations are not stateless actors. They operate under home‑jurisdiction export controls, national security laws and expanding regulatory regimes. Across the globe, we see rising AI nationalism: states are building domestic foundation‑model capacity and tightening oversight over private algorithmic systems. State power is not withering away.
Even so, leading AI firms possess degrees of strategic agency that Cold‑War‑era geopolitical frameworks cannot explain.
When one corporate foundation model underpins a large share of a country’s domestic economy, that firm can wield significant de‑facto economic leverage, without occupying a single kilometre of a nation’s territory. States do have countermeasures: open‑source public AI systems, local‑model requirements and regulatory safeguards. But deploying them requires foresight.
I describe this emerging grey zone as “shared sovereignty.” It is a descriptive analytical puzzle, not a desirable outcome. For many observers, it represents a serious democratic accountability risk requiring governance responses. Our existing international toolbox lacks agreed categories to understand this new reality.
Reading the early warning signs from today’s conflicts
Recent conflicts offer early glimpses of this shifting order. The Ukraine conflict has demonstrated how AI‑enhanced drones and satellite intelligence can reshape battlefield dynamics faster than tank formations. Cyber‑physical attacks targeting energy infrastructure, seen in incidents linked to Iran, show how adversaries can strike core national systems without crossing physical borders.
Yet in both cases, classical factors—manpower, industrial production capacity and territorial control—remain decisive. AI and cyber tools act as force multipliers, but they do not eliminate older forms of power. If analysts interpret modern conflicts purely through Cold War lenses, they will repeatedly misread events.
Multiple futures lie open. Power could diffuse toward non‑state technological actors. Alternatively, states might successfully re‑centralise control over AI. Most probable is a messy mixture of both trajectories unfolding side‑by‑side.
Conclusion: Who will write the new geopolitical map?
The old Cold War geopolitical map is under strain, not completely torn up. New rules are being coded in data centres, in corporate boardrooms, in state AI laboratories and, one day, possibly in outer space.
This article does not offer final conclusions. It puts forward questions for collective debate:
- How can we move past obsolete geopolitical paradigms, without discarding proven security instruments that still serve national interests?
- What governance rules can adapt to an AI‑transformed global landscape?
- How should states, technology corporations and multilateral institutions re‑define their roles—before external events force those re‑definitions upon us?
We have a choice. We can help shape the emerging order. Or we can simply read about it after it has already taken shape.
