Southeast Asia’s familiar predicament – being asked to choose between Washington and Beijing – is acquiring an artificial intelligence dimension. In August, Reuters reported that a draft U.S. State Department letter warned partners against combining participation in Pax Silica, the U.S.-led technology initiative, with competing arrangements. The letter didn’t include specific names, but the obvious inference was that Washington was trying to discourage participation in China’s World Artificial Intelligence Cooperation Organization (WAICO).

The draft was undated, and Reuters could not establish if it would be sent or whether it would change. Even so, its political ambition was clear: Washington wanted cooperation on AI to carry a stronger expectation of alignment.

For Southeast Asian governments, the danger extends beyond an uncomfortable diplomatic decision. Technology agreements made today can shape which choices remain affordable tomorrow. If a country’s businesses and public services become heavily dependent on one foreign supplier, resisting that supplier’s government may become more costly even without a formal demand to take sides. The region therefore needs to negotiate development benefits and political room for maneuver together, while competing powers still have reasons to court Southeast Asia.

Separate U.S. and Chinese initiatives are drawing countries into different bargains. Pax Silica, launched in December 2025, brings together countries seeking closer cooperation across technology supply chains, from critical minerals to AI systems. Its appeal lies in the prospect of investment, industrial partnerships, and advanced American technology.

WAICO began with 29 countries signing its founding agreement in Shanghai on July 16. By August 12, Foreign Minister Wang Yi reported 38 founding members. Its purpose is to “promote international cooperation and global governance on AI,” with a particular focus on making sure Global South countries have equitable access to this pivotal technology.

These parallel arrangements have not yet divided the world into exclusive camps. Kazakhstan joined both, as Reuters noted. Uzbekistan joined WAICO and is actively seeking membership in Pax Silica. Singapore described the Pax Silica declaration as non-binding and presented its participation in terms of mutual economic benefit. 

The reported U.S. letter matters precisely because it would narrow that flexibility. Whether Washington can actually get its partners to make a binary choice depends on the benefits it offers, the costs it can impose, and the willingness of partners to accept its terms.

Beijing’s offer speaks to governments worried about being left behind. WAICO presents developing countries with a place in discussions about AI governance, alongside promises of training and practical cooperation. In July, Chinese president Xi Jinping pledged 5,000 training opportunities for developing countries over five years and AI application cooperation centers with regional organizations, including ASEAN. Such promises address a concern that great power debates often neglect: how countries with limited resources can put AI to productive use.

This also serves China’s strategic purpose. Bringing more governments into a Shanghai-based institution strengthens Beijing’s claim to speak for countries poorly represented in existing technology governance. Its emphasis on access makes U.S. restrictions easier to portray as an effort to preserve privilege. Yet WAICO’s inclusive nature does not settle who will influence the agenda or capture the commercial gains. Southeast Asian participants should judge China’s offer by the capabilities it helps them acquire and the influence they gain over the cooperation itself.

Washington has a different source of bargaining power: the ability to restrict access to advanced computing equipment. U.S. guidance issued in May reaffirmed that exports of certain equipment to companies based in China and their overseas subsidiaries can require American approval. For a Chinese company operating in Malaysia, this means its access to equipment may still depend on decisions made in Washington. Kuala Lumpur’s diplomatic friendships cannot, by themselves, settle that question.

The American offer also increasingly combines access with a broader commercial relationship. The American AI Exports Program is designed to promote packages that bring hardware, software, and services together, backed by possible financing and government support. U.S. development finance officials have explicitly tied their ambitions to trusted suppliers and American strategic interests. For a recipient, an attractive investment package could therefore influence a much wider range of future technology purchases.

The political effect need not begin with coercion. A well-integrated system may work well, attract investment and win support from domestic businesses. As those benefits accumulate, however, the cost of changing suppliers rises. Governments may eventually face pressure from employers, investors, and public agencies to preserve access, even when relations with the supplier’s home government deteriorate. Commercial success can thus give a foreign partnership political staying power. That is one route through which an AI bloc could develop before countries formally agree to belong to one.

Not every foreign dependency is automatically bad, however. Southeast Asian economies can gain from American investment, Chinese technology, and partnerships with both. Rejecting those opportunities in pursuit of complete technological independence would impose substantial costs. The more real test is whether cooperation leaves local firms and institutions better able to make their own decisions, or simply better equipped to keep buying from the same partner.

Singapore shows that some choices remain open. Its participation in Pax Silica has coexisted with the use of Chinese-origin technology. AI Singapore’s SEA-LION models draw on both Alibaba’s Qwen and Google’s Gemma, adapting them to Southeast Asian languages and contexts. Local expertise allows Singapore to select and adapt foreign tools for regional purposes. 

Other governments can exercise influence through the resources and permissions investors need. Thailand’s reported September 4 pause on 166 data center projects, covering both projects under construction and those awaiting approval, came amid efforts to establish rules on resource use and economic benefits. It illustrates a development question often hidden by the geopolitical contest: how much should a host country gain in return for supplying land, electricity, water, and public support? Foreign investment can be welcome while its terms remain negotiable.

That bargaining power has limits. Investors can look elsewhere, governments face different fiscal pressures, and access to the most advanced technology is uneven. A country seeking basic, affordable AI services faces a different negotiation from one trying to become a regional technology hub. Not every country in Southeast Asia enjoys Singapore’s expertise or commands the same level of commercial interest. But those differences make clear national priorities more necessary: governments need to know what benefits they will receive for making certain commitments. 

If Washington is going to attempt to force the question of alignment, it needs to make clear what the returns will be. Southeast Asian governments should press for clarity about access, financing, and the circumstances under which cooperation could be restricted. Restrictions directed at particular security risks should not automatically become demands to abandon unrelated partnerships. If the United States asks countries to bear the economic costs of excluding Chinese suppliers, it should be prepared to offer dependable and competitive alternatives.

Beijing should face equally concrete questions. Promises of shared development need to produce sustained training, affordable services, and opportunities for local businesses. WAICO’s participants should seek a meaningful role in setting priorities, alongside clear commitments about what cooperation will deliver. An expanding membership gives China diplomatic standing; members are entitled to expect benefits beyond attending Chinese-hosted meetings.

ASEAN can help governments negotiate from better information. Its existing AI governance work provides a starting point for sharing expertise and comparing how foreign partnerships serve regional needs. Members could coordinate expectations on training, local participation, and the continuity of essential services, while retaining different diplomatic relationships. Such cooperation would make it harder for suppliers to benefit from governments negotiating in isolation.

The emerging AI rivalry will test Southeast Asia’s capacity to turn external attention into lasting advantage. Refusing to choose sides remains a defensible diplomatic position, but it needs material support: capable domestic institutions, worthwhile alternatives and agreements that preserve future choices. Washington and Beijing both want their partnerships to endure. Southeast Asian governments should seek terms that leave them better able to pursue their own priorities, including when their partners disagree.