Geopolitics & Sovereign Positioning
Top Line
The Atlantic Council's Commission on AI has published a comprehensive roadmap calling for systemic reform across supply chains, alliances, talent, energy infrastructure, and governance — signalling elite US foreign policy consensus that America is currently losing ground in the AI competition and requires coordinated state action to reverse that trajectory.
India's deal with UAE-based G42 to deploy Cerebras-designed supercomputers on Indian soil represents a structural challenge to US hyperscaler cloud dominance, offering a replicable model for sovereign AI infrastructure that bypasses American platforms while still depending on US-origin chip designs.
Tencent's reported imminent launch of an AI agent inside WeChat's 1.4 billion-user ecosystem illustrates China's decisive structural advantage in AI deployment scale — a gap the CFR argues Washington has no credible domestic equivalent to close in the near term.
US export control strategy faces a core tension surfaced across multiple analyses: controls on chips and models may slow Chinese frontier development at the margins, but China's platform-driven deployment engine and state coordination give it a deployment advantage that export restrictions do not address.
Key Developments
Atlantic Council Commission Declares US AI Leadership at Risk, Prescribes Systemic Overhaul
The Atlantic Council's Commission on AI has released a flagship report and a suite of accompanying issue briefs covering supply chains, energy, alliances, talent, governance, and innovation — collectively amounting to the most comprehensive US foreign policy establishment diagnosis of American AI vulnerability to date. The report's framing is explicitly competitive: the US is 'struggling' against adversaries and requires an action-oriented roadmap, not incremental adjustment. Key recommendations span diversifying semiconductor supply chains beyond Taiwan, building allied AI infrastructure through bilateral and multilateral frameworks, streamlining federal AI procurement, and expanding STEM and governance talent pipelines. These are policy proposals from an influential think tank, not enacted law, and carry no binding force — but they reflect and likely shape the consensus within which executive branch officials and Congressional staffers operate.
The supply chain issue brief identifies chokepoints across the full AI stack — from rare earth inputs through advanced packaging to model deployment infrastructure — and the degree of US dependence on single-country or single-firm suppliers. The energy brief highlights that AI's expanding power footprint creates new strategic vulnerabilities: data centre energy dependency is increasingly a national security variable, and allied coordination on energy infrastructure for AI is framed as a force multiplier. Taken together, the Commission's output signals that the US foreign policy establishment has internalised that AI competition is systemic, not reducible to chip export controls alone.
India-UAE G42 Cerebras Deal Establishes a New Sovereign AI Infrastructure Template
G42, the Abu Dhabi-based AI firm with close ties to the UAE government, will deploy Cerebras-designed supercomputers in India, giving the Indian government direct ownership of high-performance AI hardware on its own soil rather than renting compute from US hyperscalers. As reported by Rest of World, this arrangement is explicitly framed as a sovereignty play — India wants to own its AI infrastructure, not depend on AWS, Azure, or Google Cloud for sensitive workloads. Cerebras's wafer-scale chips, designed in the US, remain inside the export control framework, meaning this is not a sanctions-evasion arrangement, but it does route US-origin technology through a Gulf intermediary to an emerging market in a way that reduces direct US platform leverage over Indian AI development.
The strategic significance extends beyond the bilateral. The G42-India model — sovereign hardware ownership, US-designed chips, Gulf-state intermediary financing and deployment — is potentially replicable across Southeast Asia, the Middle East, and Africa. For the Global South, it offers a path to meaningful AI sovereignty without building domestic chip fabs, which remain economically and technically out of reach for most countries. For Washington, it raises the question of whether US influence over allied AI infrastructure is better preserved through cloud platform dominance or through proactive chip export frameworks that allow hardware ownership while maintaining design-layer leverage. The deal implicitly answers that question in favour of the latter, but the US government has not yet formalised that as a strategic posture.
China's Platform Deployment Advantage Comes Into Sharp Relief as Tencent Moves AI Into WeChat
Tencent's reported plan to embed an AI agent directly into WeChat — a super-app with 1.4 billion active users handling messaging, payments, commerce, healthcare scheduling, and government services — would constitute one of the largest single AI deployments in history measured by addressable user base. As reported by the South China Morning Post, the prototype is in testing and compliance processes for a public launch could begin as soon as this month. The compliance framing is notable: China's regulatory apparatus for AI, while strict on content, is structured to enable rapid platform-level deployment once political alignment is confirmed, not to slow it.
The CFR analysis published the same day provides the strategic context: China's platform economy — WeChat, Alipay, Douyin, Meituan — constitutes a deployment engine with no US equivalent at comparable scale or integration depth. The argument is not that Chinese frontier AI models are more capable than American ones, but that China has a structural advantage in translating model capability into population-scale behavioural change, data generation, and economic productivity gains. For the geopolitical competition, this matters because AI power is not only about who trains the best model — it is about who deploys AI at scale, generates feedback data, and embeds AI into critical economic and social infrastructure first. On that metric, China's platform architecture currently outperforms the fragmented US app ecosystem.
Signals & Trends
Export Controls Are Winning the Chip Race but Losing the Deployment War
The accumulated evidence across today's reporting points to a structural gap in US AI competition strategy: export controls are calibrated to restrict Chinese access to frontier training hardware, but the dimension on which China is currently gaining ground — deployment scale through integrated platforms — is largely unaffected by chip restrictions. Tencent's WeChat AI agent does not require access to NVIDIA H100s to reach 1.4 billion users with an AI assistant embedded in their primary communication and commerce interface. The CFR framing makes this explicit: Washington is competing on the wrong metric. A strategy centred on controlling semiconductor exports addresses one node in a multi-variable competition while leaving the deployment, data accumulation, and standard-setting dimensions largely uncontested. Senior US policymakers are beginning to recognise this — the Atlantic Council Commission's emphasis on governance, standards, and allied deployment infrastructure suggests the frame is shifting — but enacted policy has not caught up with the analysis.
Gulf States Are Emerging as Indispensable AI Infrastructure Intermediaries
The G42-India deal is not an isolated transaction — it reflects a broader pattern in which Gulf sovereign wealth and state-linked firms are positioning themselves as the preferred intermediary for deploying US-origin AI technology in markets where direct US platform presence is constrained by sovereignty concerns, regulatory friction, or political preference. The UAE, Saudi Arabia, and Qatar each have the capital, the political relationships across the Global South, and the regulatory flexibility to serve this function at scale. For Washington, this creates a dependency in reverse: the US may need Gulf intermediaries to extend its technological influence into the fastest-growing AI markets, giving those states meaningful leverage over how US AI technology spreads globally. This is a structural shift from the Gulf's prior role as an energy supplier to a potential role as a technology distribution node — a shift that carries significant implications for alliance management and export control design.
AI Talent and Institutional Capacity Are Becoming the Binding Constraint on Sovereign AI Ambitions
Across the Atlantic Council's issue briefs, and implicitly in the India-UAE deal structure, a consistent theme emerges: hardware and capital are becoming more accessible, but the governance capability and institutional readiness to deploy AI safely and effectively remain scarce. Countries that can acquire compute through sovereign deals or cloud arrangements still face a shortage of the interdisciplinary expertise needed to procure, evaluate, and integrate AI into public institutions. This creates a second-order competition — not just over who has the chips, but over who can train the civil servants, regulatory staff, and procurement officers capable of making AI work for the state. Nations that invest in this institutional layer now will have durable advantages even if their hardware access fluctuates. The US, EU, and UK have structural advantages here, but they are not exploiting them systematically through allied capacity-building programs, which represents a missed opportunity to extend influence in the Global South at lower cost than hardware subsidies.
Explore Other Categories
Read detailed analysis in other strategic domains