Washington and Beijing want technological security, but dividing AI into rival blocs could carry a heavy cost.
When Washington asks partner countries to choose between American and Chinese AI technology systems, a primitive human instinct takes on a new form. Technology is becoming tribal.
Artificial intelligence, semiconductors, and critical minerals are no longer treated as separate commercial issues. Governments increasingly see them as parts of one strategic system. AI is therefore becoming more than a business opportunity. It is becoming a source of national power.
Why the tribes have a case
There are good reasons for countries to worry about technological dependence. A government that relies on another country for advanced chips, computing power or critical AI systems may expose itself to serious risks in an increasingly edgy and unpredictable world.
AI is already being used in defence, intelligence, cybersecurity, and other essential services. Governments naturally want to know who controls the systems on which their security and economies depend. The recent US government’s abrupt and unilateral blocking of Anthropic’s advanced AI models, Mythos and Fable, was a stark reality check for the free world. Even allied countries can become vulnerable when critical AI capabilities depend on decisions made by another government.
AI also reflects political choices. Decisions about data, privacy, censorship, access, and the role of government are built into the way these systems are developed and used. The United States and China do not always approach these questions in the same way. For many countries, choosing a technology system may therefore also mean choosing which political values and rules to accept.
More from Opinion‘Silicon Schism’ could help as well as squander resources
Competition can produce important advances for humanity. During the Cold War, rivalry between the United States and the Soviet Union helped drive the space programme. US defence research also contributed to ARPANET, an early foundation of the internet.
But competition has costs. It can waste resources, encourage dangerous assumptions and increase the risk of miscalculation. Military or technological strength does not automatically produce political wisdom. A government may become better at using technology without becoming better at deciding what that technology should be used for.
The same problem applies to AI. This emerging divide can be called the ‘Silicon Schism’: technical systems remain globally connected even as the political rules governing them diverge.
AI technology refuses to stay within the tribe
AI is difficult to divide along national lines without consequences. The Transformer architecture behind modern generative AI came from research published openly in 2017. It spread through an international research community. Software, academic papers, and technical methods move across borders faster than governments can control them.
The hardware supply chain is even more scattered. Stanford’s 2026 AI Index reports that the United States has the largest concentration of AI data centres. Taiwan Semiconductor Manufacturing Company produces almost every leading AI chip, while Dutch company ASML is the only supplier of the most advanced extreme ultraviolet lithography machines used to manufacture those chips.
This creates a basic contradiction. Countries may want technological independence, but they still depend heavily on foreign companies, workers, factories and research networks. AI infrastructure is concentrated in a few places, yet it is also overwhelmingly international. The dependence is not only physical. It is also cultural and linguistic. Large Language Models are trained on the knowledge and information of the world accumulated over centuries.
That scattered geographic interdependence is shaping the new AI order. The knowledge behind AI is global. The infrastructure is spread across several countries, and even the basic minerals required to manufacture many of its components are widely distributed. But the strategic interests surrounding it are becoming increasingly national. Governments want control over systems that no single country can fully build alone.
The big corporation complications
Large technology companies add another complication. Stanford’s 2026 AI Index reports that private companies produced more than 90 per cent of notable frontier AI models in 2025. These companies depend on international supply chains, advanced chip manufacturers, global talent, and large amounts of overseas investment. Those connections give companies reasons to resist technological fragmentation. Separate standards, restricted markets, and duplicated infrastructure would raise costs and slow development.
But companies are not geopolitically neutral. Governments can impose export controls, regulate computing capacity, and require firms to support national-security objectives. The result is a system in which corporations operate globally but remain exposed to national political demands. A company may depend on foreign suppliers and customers while still being required to follow the strategic and national priorities of the government where it is based. International agreements will become harder to negotiate when national security is involved. Treaties can create standards and make violations more costly.
The risks may grow as AI becomes part of critical infrastructure. The more deeply AI becomes embedded in finance, energy, communications, and defence, the greater the damage that could result from disrupting another country’s digital systems. The idea of Mutual Assured Digital Destruction, or MADD, remains hypothetical. But the concern behind it is real. The current US-Israel-Iran conflict has already shown that data centres can become targets in warfare: Iranian strikes damaged Amazon Web Services facilities in the UAE and Bahrain, disrupting digital services and demonstrating how physical attacks on digital infrastructure can have consequences far beyond the battlefield.
Technical convergence, political divergence
The emerging “Silicon Schism” is unlikely to create two separate AI worlds. The rival powers may ultimately have to find a workable accommodation between strategic competition and technological interdependence. The emerging pattern may be technical convergence alongside political divergence, with countries using similar underlying technologies while operating them under different rules, information systems, and strategic goals.
AI emerged from a globally connected ecosystem of research, talent, data, capital, and supply chains. It does not need a single global owner. Its global foundations cannot be wished away by geopolitical rivalry.
(Sreejith Sreedharan is a technology analyst and author of Future of Work – AI Augmented Autonomous Decentralised. He works on organizational AI readiness and created the AI Instinct Index®, a psychometric diagnostic designed to assess behavioural readiness for AI adoption and adaptive capacity in constraint-heavy environments. Views expressed in the above piece are personal and solely those of the author. They do not necessarily reflect Firstpost’s views.)
