Geopolitics & Sovereign Positioning

China's 'AI+ Education' action plan, issued jointly by the Ministry of Education and four other ministerial bodies, mandates AI integration across every stage of formal and continuing education. This is an enacted policy directive with implementation obligations attached to the education bureaucracy, not a voluntary framework. The scope — from primary schools through lifelong learning — signals that Beijing is treating AI literacy as a national infrastructure investment comparable to its earlier pushes on STEM and manufacturing competency. South China Morning Post

The geopolitical read is straightforward: the US-China AI competition is ultimately a talent competition, and China is moving to ensure its workforce is not a bottleneck. While US export controls target chips and model weights, they cannot restrict the cultivation of domestic AI engineering capacity. The plan compounds China's existing advantages in data volume and industrial deployment speed — highlighted separately by GigaAI executive Wang Xiaofeng, who argued that China's world model development is outpacing US counterparts due to faster integration with its industrial base. South China Morning Post The education mandate is the upstream investment that sustains that deployment advantage over a 10-to-20-year horizon.

The US-Israeli conflict with Iran, now in an unstable ceasefire, has surfaced two distinct supply chain chokepoints that bear directly on global AI infrastructure capacity. The helium story — Qatar's Ras Laffan facility going offline and triggering a 45-day inventory clock with spot prices doubling — has received coverage. The bromine story has not, and analysts at War on the Rocks assess it as potentially more severe: bromine is the feedstock for flame retardants used in memory chip production, and Israel is a primary global supplier. A sustained disruption to Israeli bromine output could constrain memory chip production globally, directly throttling AI data centre expansion. War on the Rocks

Separately, the Strait of Hormuz closure following Trump's announcement compounds the pressure on Asia's technology sector. Roughly a quarter of global seaborne crude and 20 percent of LNG flows through the strait, and AI data centres are among the most energy-intensive infrastructure investments on the planet. Analysts cited by the South China Morning Post warn that prolonged hostilities could throttle both semiconductor production and data centre buildout across the region. South China Morning Post The Gulf AI investment boom — including major sovereign data centre and cloud projects — faces a compounding risk: security costs for infrastructure protection are rising precisely as energy and component costs spike.

Oxford Economics research quantifies what has been suspected but inadequately documented: China is materially benefiting from the approximately $2 trillion in US data centre investment through Asian technology supply chains, with Chinese component and equipment suppliers embedded in the upstream manufacturing networks serving American hyperscalers. The research estimates that up to three-quarters of data centre project costs flow to equipment, and the supply chain for that equipment runs through jurisdictions with significant Chinese industrial participation. South China Morning Post This is a second-order consequence of export controls that Washington has not resolved: controls on leading-edge chips do not sever Chinese participation in the broader AI hardware economy.

Simultaneously, DeepSeek is expanding its own compute infrastructure, recruiting server engineers and data centre delivery managers in Ulanqab, Inner Mongolia — its first publicly advertised on-site operational roles outside Hangzhou. The Inner Mongolia location is significant for energy costs and proximity to renewable power, suggesting DeepSeek is building out physical infrastructure ahead of its V4 launch rather than relying solely on third-party cloud capacity. Questions about how DeepSeek is acquiring the compute it needs — given restrictions on high-end Nvidia chips — remain unresolved, and this expansion does not answer them. South China Morning Post Separately, reports of Shanghai-based Dishan Technology reaching prototype verification stage on a 2nm AI GPU — if confirmed — would mark a meaningful advance in China's domestic frontier chip capability, though the company has not provided independent verification and the gap to TSMC's production-scale 2nm remains substantial. South China Morning Post

Two distinct developments signal accelerating military AI integration among US and allied forces, with different maturity profiles. Foreign Policy published concurrent pieces examining Pentagon AI risk management frameworks, with analysis from Anthropic's engagement with the defence sector. The pieces argue that warfighter trust is the binding variable for AI adoption — if operators don't trust AI-generated targeting or situational awareness outputs, adoption stalls regardless of capability. The 'humans in the loop' debate, prompted in part by reported AI involvement in targeting decisions during the Iran conflict, is now a live doctrinal question rather than a theoretical one. Foreign Policy Foreign Policy

Japan's move is more concrete: the Ground Self-Defense Force has established two new offices dedicated to institutionalising drone warfare, explicitly linked to manpower shortfalls from demographic decline. This is an enacted organisational change, not a proposal. The strategic logic is that Japan's shrinking military-age population makes unmanned and AI-enabled systems a structural necessity rather than an optional capability enhancement. The Diplomat For the US alliance architecture in the Indo-Pacific, Japan's move matters: it increases Japanese autonomous warfighting capacity while creating new interoperability requirements — and opportunities — with US AI-enabled military systems.

At the inaugural Hong Kong Global AI Governance Conference, Alibaba's policy lead Fu Hongyu framed China as 'at the front lines of global efforts to introduce guardrails' on AI, deploying the language of 'common ignorance' to argue that no government — including the US — has superior regulatory insight, making international collaboration the necessary frame. This is a diplomatic positioning move, not an enacted governance framework, but it represents a sustained Chinese effort to occupy multilateral AI governance space that the US has intermittently vacated. South China Morning Post

The conference venue — Hong Kong, positioned as a bridge between mainland China and international institutions — is deliberate. Beijing is using Hong Kong's residual international connectivity to host governance conversations that it cannot host on the mainland without triggering Western participation concerns. The 'common ignorance' framing is also strategically useful: it neutralises US claims to governance leadership by asserting epistemic equivalence across all actors, implicitly delegitimising export control regimes premised on the assumption that restricting Chinese access serves safety goals.