AI is quickly shifting into new domains. Image: Shutterstock
Doug Specht examines how artificial intelligence is shifting from a shared commons to a strategic, controlled territory
Earlier this month in Las Vegas, the world’s largest technology exposition (CES 2026), became the stage for something more consequential than any new gadget: the formal end of global AI cooperation. While the conference halls showcased the latest robotics and edge computing systems, the real story was written in boardrooms and government buildings across three continents.
In rapid succession, the Trump administration revised US chip export policies, the UK activated its £500 million Sovereign AI Unit, France announced €109 billion in AI investment, and the UAE consolidated a $40 billion data centre portfolio. These announcements, simultaneous and coordinated, signal a fundamental geopolitical realignment. The world’s great powers have decided: artificial intelligence will not be a shared commons, but a strategic territory to be controlled, walled off, and defended.
Enjoying this article? Check out our related reads…
This is not competition as we understand it in technology markets. It is enclosure, the transformation of commons into a controlled state and corporate property, wrapped in the language of ‘sovereignty’ and ‘security’.
Just as 19th-century land enclosure movements privatised shared agricultural commons, today’s techno-nationalism is fragmenting digital infrastructure into regional fortresses. And just as historical enclosure devastated those without land or power to resist, the geopolitical fragmentation of AI risks embedding technological inequality into the world’s digital infrastructure for decades.
From ‘AI for all’ to ‘AI for us’
The mythology surrounding artificial intelligence has always been one of democratisation. For the past five years, technology leaders spoke of AI as a public good, a transformative tool that would lift all nations, democratise decision-making, and solve humanity’s collective problems. This rhetoric framed AI development as a shared endeavour with universal benefits. The United Nations, World Bank, and countless tech conferences echoed the vision: ‘AI for good,’ ‘AI for all’, ‘responsible AI’.
But CES 2026 revealed the shift beneath the rhetoric. Jensen Huang of Nvidia and Lisa Su of AMD shared stages with White House officials and military strategists, and the conversation was not about innovation for humanity. It was about national control. The language had changed. ‘Sovereignty’ replaced ‘democratisation’. ‘Strategic autonomy’ replaced ‘access’. The subtext became the text: this is about power, territory, and exclusion.
AI technology has boomed in recent years, with the rise of apps such as ChatGPT, Claude and Perplexity. Image: Shutterstock
Every major power abandoned the pretence of global cooperation. Instead, they have pivoted to what might be termed ‘techno-nationalism’, elevating AI to the status of strategic infrastructure equivalent to military capability or energy security. The Trump administration’s January 2026 decision to ease some H200 chip exports to China was not a retreat but a recalibration: allowing limited commerce while simultaneously building high walls elsewhere. France, the UK, and the UAE recognised the same imperative. Build walls. Control territory. Ensure insiders benefit; ensure outsiders cannot compete.
The infrastructure frontiers
Geography still matters in the digital age, perhaps more than ever. AI requires vast computational capacity, which requires electricity, which is tied to place. Data centres must be built somewhere, consuming electricity from somewhere, processing data about somewhere. These locations are not economically neutral; they are geopolitical assertions.
The UAE’s $40 billion acquisition of Aligned Data Centers, completed in October 2025, exemplifies this new geography. Abu Dhabi is not building data centre capacity for innovation’s sake. It is constructing a non-aligned computer archipelago spanning the Middle East, Asia-Pacific, North Africa, and Turkey. This positions the Gulf as an AI infrastructure intermediary, able to serve neither the US bloc nor the Chinese sphere exclusively, but both on its own terms. It is geopolitical leverage disguised as infrastructure investment.
Europe’s approach is more defensive. The UK’s Sovereign AI Unit, France’s €109 billion investment plan, and Belgium’s partnerships with Shield AI represent a coordinated effort to prevent total dependence on either Washington or Beijing. Yet the EU cannot manufacture cutting-edge semiconductors; those capabilities remain concentrated in Taiwan and South Korea, politically aligned with the US.
As Doug Specht argues, data centre locations are geopolitical assertions. Image: Shutterstock
Europe can build data centres and develop software, but the foundational hardware comes from abroad. ‘Sovereignty’ under these conditions is an illusion, a political claim masking actual subordination. You do not have genuine sovereignty over AI infrastructure when your chips are imported, and your energy grids are vulnerable to external pressure.
China’s approach is explicit: total vertical integration. Since Nvidia exports are restricted, China is building every layer, chip design, manufacturing, packaging and testing, data centres, and energy infrastructure. The goal is not technological parity with the West; it is technological independence from the West. Every segment of the value chain must be Chinese-owned and operated, insulated from foreign restrictions.
This fragmentation is not accidental or merely strategic convenience. It is the logical outcome of treating AI as territory rather than knowledge. When governments classify AI as infrastructure equivalent to ports, railways, and power grids, which they now do, they inevitably fence it. And when infrastructure becomes walled, the geography of power shifts violently.
Who gets left behind?
Behind the spectacle of sovereign AI initiatives and geopolitical posturing lies a quieter catastrophe: the systematic exclusion of the Global South from any potential benefits of AI and the deepening embedding of its dependence.
The International Monetary Fund estimates that AI could double income inequality between advanced and low-income countries. This is not hyperbole. The mechanism is straightforward. Advanced economies can invest billions in domestic AI infrastructure, retrain workforces, and capture value creation.
Low-income countries lack computing capacity, energy grids, and trained personnel to build domestic AI systems. They will import AI tools from wealthy nations’ companies, generating zero local benefit except employment in low-wage roles: data annotation, content moderation, customer service.
The disparity in governance makes this worse. According to the UN Conference on Trade and Development, 118 nations, mostly from the Global South, are not meaningfully represented in AI governance forums. The World Bank, IMF, and informal “AI councils” where standards are set remain dominated by advanced economy representatives. The Global South is consulted, not heard. They are subjects of AI policy, not authors of it.
AI trained predominantly on Western data performs poorly on non-Western populations, diagnostic accuracy drops for conditions more common in the tropics, agricultural models fail on smallholder farming systems, language models generate poor outputs for non-European languages. Yet the Global South cannot develop its own AI systems without computing capacity they cannot afford to build. They must adopt Western-trained models and accept their inadequacy.
Why this moment?
The simultaneity of policy announcements at CES 2026 was not a coincidence. It signals the moment consensus shifted from ‘global AI system’ to ‘regional AI systems’. This is not a temporary phase of competition. It is an acknowledgement, finally made explicit, that the geopolitical bifurcation of technology is permanent.
The question is whether this enclosure can be resisted. History suggests enclosure is difficult to reverse, but not impossible. The answer lies not in technology but in politics, and at present that political will appears lacking.