The United States is preparing to tell dozens of partners that the emerging artificial intelligence contest with China may leave little room for playing both sides.
According to an internal State Department draft reviewed by Reuters, Washington intends to warn countries aligned with its AI initiatives that participation in competing Chinese-led frameworks could exclude them from the U.S.-led coalition.
- The document is addressed to the 35 signatories of the Joint Statement on AI Opportunity, a group that includes members of Pax Silica and other governments that have signalled their willingness to align AI cooperation with Washington. Italy is among them.
- “To be part of everything is to be part of nothing,” the draft says. Membership in Pax Silica, it adds, is “not merely a membership subscription, but a commitment”.
What to watch: The document could still be amended, and Reuters could not determine when it might be sent. But its political direction is difficult to miss. Washington appears to be moving from building a competing technology ecosystem to defining the conditions for belonging to it.
- At the centre of the immediate dispute is Kazakhstan. Washington had celebrated its accession to Pax Silica in June, not least because of the country’s significant critical-mineral reserves. But Astana subsequently also joined the World Artificial Intelligence Cooperation Organization launched by Chinese President Xi Jinping in July.
- It is so far the only country known to have joined both. For Washington, that appears to have turned Kazakhstan into a test case for a much broader question: can a country be considered a trusted partner in one technology ecosystem while simultaneously participating in an initiative designed to advance a competing one? The answer emerging from the State Department is increasingly “no”.
That matters because Pax Silica reaches far beyond AI models. Launched by Washington last year, the initiative connects semiconductors, critical minerals, artificial intelligence, investment and supply-chain security.
- Its underlying proposition is that technological sovereignty does not require countries to replicate the entire AI stack domestically. Instead, trusted partners contribute complementary capabilities to a common architecture.
China is developing an alternative proposition. At the World Artificial Intelligence Conference (WAIC) in Shanghai in July, Beijing launched its World AI Cooperation Organization (WAICO), backed by 29 countries, alongside an offer built around open-weight technology, capacity-building, technical standards and access for developing economies.
- The competition is therefore moving beyond who produces the most capable model. It is increasingly about who defines the networks through which compute, minerals, semiconductors, standards, investment and AI systems move.
Those networks depend on trust. The implications were already becoming visible before the latest U.S. draft emerged. Asked by Decode39 about the shrinking room for ambiguity in AI governance, Hudson Institute Research Fellow Zineb Riboua argued that the space governments believed they retained between the American and Chinese ecosystems was largely illusory.
- “The room governments believe they retain is an illusion sustained by the interval between procurement and consequence, and that interval is closing,” she said.
- Governance frameworks can prove even more durable than temporary advantages in compute because they determine how systems interoperate, how models are evaluated and how public authorities procure and deploy technology.
- The draft reported by Reuters suggests that Washington may now be preparing to turn that strategic logic into a more explicit diplomatic requirement.
For Italy, this distinction matters. On July 31, Italy formally joined Pax Silica, following its earlier signature of the Joint Statement on AI Opportunity. The decision sits alongside Italy’s participation in the U.S.-led Call to Action for 6G Leadership and Security, its engagement with Washington on critical minerals and the bilateral dialogue on Trusted Technologies.
- Foreign Minister Antonio Tajani described the Pax Silica accession as confirmation of Italy’s “political will to work with the United States and other partners” on production technologies, strengthening both security and “the unity of the West”.
Yes but… The strategic direction is therefore relatively clear. The more difficult question is implementation.
- That question had already surfaced after the presence of Italy’s Agency for Digital Italy, AgID, at the WAIC in Shanghai, shortly after Rome had joined the U.S.-led Joint Statement on AI Opportunity. The episode did not prevent Italy’s subsequent accession to Pax Silica. But the message now emerging from Washington suggests that similar ambiguities may become harder to sustain as the two AI architectures consolidate.
There is an important distinction. Participation in the Shanghai conference is not equivalent to joining China’s new organization, and Italy is not in Kazakhstan’s position. But the direction of U.S. policy matters precisely because it may progressively raise the political significance of interactions that governments previously regarded as technical, commercial or institutionally routine.
- Martijn Rasser, Vice President for Tech Leadership at the Special Competitive Studies Project, had already identified implementation as the next challenge when Italy joined Pax Silica.
- “The test will be whether its members translate political alignment into coordinated action,” he told Decode39, while arguing that Italy was moving in the right direction but still had more to do in identifying unacceptable dependencies, protecting critical technologies and diversifying supply chains.
- Reuters’ disclosure adds another dimension to that test.
The bottom line: Italy has picked its side in the emerging AI architecture. If Washington follows through on the approach contained in the draft, the question for Rome will increasingly be whether that choice is reflected consistently across the institutions of the Italian state.
- The age of AI hedging may not be over yet. But Washington is making increasingly clear that, for members of its technology coalition, the room for it is closing.
