The Emerging Geopolitics of AI

Abstract

The US–China AI rivalry is evolving into a broader contest over strategic resources, technology ecosystems and the rules governing their international use. This article examines the competition across four dimensions: supply chain control, innovation, technology dispersion and governance, focusing on the contrasting approaches represented by the US-led Pax Silica initiative and China’s World Artificial Intelligence Cooperation Organization (WAICO).

Pax Silica prioritises secure access to critical minerals, energy, infrastructure, semiconductors, manufacturing capacity and advanced AI technologies, while building a network of trusted partners. Its limited emphasis on technology dispersion and multilateral governance may, however, constrain its appeal to countries in the Global South.

WAICO takes a different approach, emphasising multilateral governance, international cooperation and AI diffusion, particularly among developing countries, while anchoring its activities in the UN framework.

The article first examines the objectives and priorities of these two initiatives before comparing their approaches and considering their implications for the emerging global AI order. It argues that their divergence could either create opportunities for cooperation or contribute to competing AI ecosystems, leaving middle powers increasingly required to navigate between them.

The AI rivalry in supply chain control, dispersion, innovation and governance

Most Americans see China as the main competitor, or even enemy, of the US. This rivalry plays out in all sources of national power, military, economy, and perhaps most important of all, technology. As US and Chinese firms battle over creating the most powerful frontier models, the state apparatus of both nations are also preparing to bring the AI rivalry into international relations.

For the purpose of this article, the AI race has been divided into four dimensions. Supply chain control, dispersion, innovation and governance. Supply chain control refers to the idea of having direct and secure access to the other resources needed for creating the microchips that power AI development. Dispersion is having one country’s AI model adopted by other countries and firms. Innovation refers to the development of the most advanced AI models. Lastly, governance refers to taking the lead in developing international rules and regulations concerning AI and encouraging other countries to adopt and implement them. It is worth noting that as innovation is primarily a bilateral competition between private-sector companies, the two initiatives discussed in this article mainly focus on supply chain control, dispersion and governance.

It is in the backdrop of this new tech rivalry that the US and China have both announced their respective international AI initiatives, Pax Silica for the US and World Artificial Intelligence Cooperation Organisation (WAICO) for China. However, rather than being copies of each other, these initiatives are strikingly different when it comes to their respective focus areas, and not only reflect Chinese and US AI policy, but also priorities and methods used in international politics by both super powers in general.

What is Pax Silica?

According to the US State Department’s website, Pax Silica is “ the (US) Department of State’s flagship effort on AI and supply chain security, advancing new economic security consensus among allies and trusted partners .” More specifically, it is an initiative by US Under Secretary for Economic Affairs, Jacob Helberg, who prefaces the initiative with the following words “ This historic declaration hails a new economic security consensus ensuring aligned partners build the AI ecosystem of tomorrow—from energy and critical minerals to high-end manufacturing and models .“

However what exactly Pax Silica is meant to achieve is still far from clear at this point. Notably, the initiative is prefaced by a declaration, namely the Pax Silica Declaration, also available on the US State Department’s website.

The declaration further clarifies some of the concepts underpinning the initiative, with a heavy focus on security: economic security, investment security, protecting critical technologies and infrastructure etc. Furthermore, the declaration makes it clear that the initiative has a rather broad scope, encompassing the entire AI supply and value chain including frontier foundation models, infrastructure, advanced manufacturing, logistics, energy etc. The declaration even contains references to trading practices which are not necessarily related to AI supply chain management, such as “ overcapacity and unfair dumping practices ”, a clear reference to China.

Given China’s position as the United States’ principal strategic rival, the composition of the initiative suggests that it may also serve as a hedging effort against the possibility of greater Chinese influence over AI development. It is certainly not a coincidence that all members of the US security architecture in Asia-Pacific, Australia, Japan, South Korea, the Philippines and India, are signatories to Pax Silica . Even Taiwan is listed as the only ‘non-signatory participant’.

As it stands, the US-led partnership is a framework for cooperation seemingly meant to counteract China’s growing role in all stages of AI development. The details of what exactly this cooperation entails remain vague, as even Pax Silica’s own website at this moment offers very little details on the precise nature of the initiative, or the companies involved.

The US-led framework is in line with the current US strategy of focusing on cutting-edge innovation, as well as securing dominance over strategic supply chains. So far, Pax Silica makes very few references to any advancements in AI governance and dispersion, an opening that Beijing may be able to take advantage of. The Pax Silica approach is also broadly consistent with the current US administration’s preference for working through trusted partners rather than relying on traditional multilateral institutions. This approach is reflected in the initiative’s emphasis on cooperation among a select group of allies and partners, rather than establishing a broadly inclusive global framework. To this stance even clearer, U.S. Under Secretary of State for Economic Affairs and Pax Silica architect Jacob Helberg even wrote an article in June criticising the idea of digital sovereignty and of the UN regulating AI.

For the developed Western powers, aligning with the US on AI development does make sense. The EU has also collectively signed the declaration, likely to ensure future access to supply chain security, and so has Australia as well as several EU member states separately. For Global South nations however, the initiative seemingly offers very little concrete value so far, and with no mention of dispersion and a very weak cooperation angle, developing nations may see the initiative as yet another “club” for rich countries to give themselves an unfair advantage in future economic ecosystems.

The WAICO, an empty promise or the beginning of a Chinese lead in governance?

The World AI Conference & High-Level Meeting on Global AI Governance, or simply WAIC, is an AI conference held in Shanghai annually since 2018. The conference has a broad scale, and in 2026 it had over 1400 international guests (including academics, researchers and business people), 140 themed forums and over 3000 exhibits. An impressive figure, underscoring China’s efforts to establish Shanghai as a global centre for technological innovation. The scale and prominence of the event also demonstrate the importance the Chinese government attaches to the AI race, as President Xi Jinping himself delivered some of his most substantial public remarks on artificial intelligence to date at the latest WAIC.

Just as the Pax Silica declaration highlights the priorities of US AI policy, President Xi Jinping’s speech similarly sets out a clear direction for Chinese AI policy. However the concepts mentioned in the speech differ significantly from the Pax Silica declaration.

Where the Pax Silica declaration is heavily focused on supply chain security, the WAIC speech remains firmly within the framework of China’s AI policy, namely focusing on AI governance and dispersion. The speech mentions the word “governance” 10 times (the Pax Silica declaration has 0 mentions, for comparison), and even anchors China’s AI governance in the UN multilateral framework by direct mention “We should practice true multilateralism and recognise the important role of the United Nations.“

Xi also explicitly mentions dispersion to the global south as a key priority for China in his speech, a policy that is perfectly consistent with China’s focus on open source models. During his speech, chairman Xi also made tangible commitments to AI dispersion, announcing that China will provide developing countries with 5,000 opportunities in AI training and seminar programs, as well as developing international AI application cooperation centres with the Association of Southeast Asian Nations, the League of Arab States, the African Union, the Community of Latin American and Caribbean States, the Shanghai Cooperation Organisation, and BRICS. Finally, the speech also hailed the announcement of what could tentatively be considered China’s answer to US AI-framework, the World Artificial Intelligence Cooperation Organisation, or simply, WAICO.

Although relatively little is currently known about WAICO, the information available nevertheless offers some indications of its likely objectives and priorities. Firstly, all of the 29 signing countries are from the global south and none of them include G7 countries, only Kazakhstan has signed both the Pax Silica declaration and the agreement on establishing WAICO. Other signatories include Russia, Malaysia, Kenya, Pakistan and Serbia. Secondly, WAICO will be an intergovernmental international organisation headquartered in Shanghai, and will commit to upholding the UN charter. Finally, the organisation will focus on international cooperation and global AI governance.

This puts WAICO in an interesting position vis-à-vis Pax Silica . Instead of responding by making its own copy of the US initiative, a block organisation focusing on supply chain security outside of the UN multilateral framework, WAICO is an intergovernmental organisation focusing on the other tracks in the AI race, namely governance and dispersion, while remaining firmly anchored in the UN charter.

These two organisations not only show each country’s core AI policies, but are the representation of broader policy trends in China and the US. While the Trump administration is generally focused on securing resources and staying ahead in cutting-edge technology, it is now China that is seemingly championing multilateralism while focusing on global governance and cooperation with the global south.

Looking to the future, possible convergence on AI supply chain management or destined to clash?

Pax Silica and WAICO are an interesting dynamic. On one hand, they can be considered to be the US and China’s competing flagship global AI initiatives. On the other hand, by analysing the content of these initiatives, it has been shown that they actually aim to dominate completely different aspects of AI governance, with Pax Silica focusing on supply chains and WAICO focusing on governance and dispersion.

These focus areas also reflect broader trends in Chinese and US foreign policy, with the Trump administration notably distrusting UN-style multilateral governance, preferring unilateral initiatives or blocks of trusted allies. Pax Silica is very much a product of this style of foreign policy. In addition, the focus on securing strategic resources and tight control over advanced technology all echo previous actions by the Trump administration, such as the resource-focused pressure campaign in Venezuela, explicitly aimed at 'recovering' oil reserves and redirecting flows away from China and Cuba and attempting to block Anthropic’s most advanced AI models for use by non-Americans.

Likewise, WAICO is also extremely coherent with general Chinese foreign policy, focusing on global governance anchored in UN multilateralism and dispersion of technology to the Global South. It is particularly interesting to observe how it is now China that consistently tries to position itself as the defender of multilateralism against one of its major architects, the US.

In an ideal world the fact that China and the US have different focuses when it comes to the AI race should mean that there is no reason for conflict. In fact, one could argue that this provides ample space for a convergence of AI governance and development rather than a point of conflict. However, as strategic rivalries between the two countries intensifies, middle powers should prepare for the brewing conflict. Unless the US and China are able to agree on common standards and rules regarding supply chain accessibility, middle powers might have to navigate two parallel but fundamentally incompatible systems, with access to supply chains and AI development opportunities contingent on alignment with one block or the other. However, whether it’s a state actor in the Global South or a developed nation, all governments must prepare for a potential AI race between the US and China in the areas discussed above, as AI and advanced microchips will undoubtedly be considered as strategic resources in the future on par with oil, minerals and energy.