Geopolitics & Sovereign Positioning

The use of Anthropic's Claude, deployed via Palantir's Maven Smart System, to generate and prioritise approximately 1,000 targets in the opening 24 hours of Operation Epic Fury represents a qualitative shift in AI's role in U.S. warfighting — not a demonstration or experiment, but live operational deployment at scale. According to War on the Rocks, the campaign's tempo more than doubled the opening phase of the 2003 Iraq invasion, reaching 13,000 total strikes over 38 days. Project Maven data indicates continued acceleration of targeting cycles. This is enacted capability with documented battlefield outcomes, not a pilot programme.

The strategic implications extend beyond U.S. borders. Adversaries — principally China and Russia — now have empirical evidence that the U.S. can compress the kill chain to a degree that renders traditional command-and-control timelines operationally obsolete. CSET's Foreign Affairs piece raises a countervailing concern: that embedding AI deeply into targeting and decision-making risks atrophying the human judgment and institutional knowledge the system is meant to augment. The operational advantage is real; the institutional risk is also real and will compound as systems become more autonomous.

China's supernode architecture — large assemblies of domestic chips engineered to deliver aggregate compute approximating banned Nvidia hardware — has transitioned from a workaround to a mainstream infrastructure approach. South China Morning Post reports that supernodes were ubiquitous at the World Artificial Intelligence Conference in Shanghai in July 2026, with Chinese firms treating them as a standard deployment model rather than a stopgap. This is a confirmed industrial adaptation, not aspirational rhetoric.

The strategic read here is important: U.S. export controls targeting high-end Nvidia chips have functioned partly as an industrial policy instrument for China, forcing domestic investment in chip interconnect architectures, software optimisation, and alternative compute paradigms. The irony is structural — controls designed to widen the capability gap have also accelerated the indigenous supply chain that would ultimately close it. However, the gap is not closed. A separate SCMP report confirms that domestic chips still fall short on precision-demanding inference tasks like coding, forcing Chinese firms to ration their remaining Nvidia supply for those workloads. Controls retain real bite at the high end of the stack, even if they are being engineered around at lower levels.

Alibaba's Qwen3.8-27B, a 27-billion parameter model designed to run on consumer hardware, is benchmarking on par with OpenAI's GPT-5.6 Luna — the U.S. lab's cost-efficiency flagship — according to benchmark firm Artificial Analysis, as reported by South China Morning Post. The model also closely tracks leading open-weight models from DeepSeek and Zhipu. This is benchmark performance, not deployed capability at scale, and benchmarks have known limitations — but the directional signal is consistent with a broader pattern of Chinese labs achieving frontier-adjacent performance at lower parameter counts and on constrained hardware.

The geopolitical dimension here is the accessibility vector: a model that performs at near-frontier levels on everyday hardware is a model that can be deployed across the Global South without data centre infrastructure. This changes the competitive landscape for AI influence beyond the core tech powers. Alibaba's parallel decision to divest its gaming arm Lingxi Games and its 45% cloud and AI revenue growth in Q2 2026 — reported by South China Morning Post — signals that Chinese capital is concentrating into AI at pace, with Alibaba projecting its AI compute investments to break even within two to three years.

A detailed War on the Rocks analysis surfaces a structurally significant tension in Chinese military AI integration: the PLA's official doctrine, as consistently articulated in Jiefangjun Bao (the Central Military Commission's newspaper), maintains that human commanders retain final decision authority and accountability. But Xi Jinping is actively testing that boundary through exercises and capability investments that push AI deeper into command functions. The gap between doctrinal reassurance and operational ambition is widening.

This matters for two reasons. First, it suggests China's AI-military integration is moving faster than its doctrine can absorb, creating command ambiguity that could affect crisis stability — particularly in scenarios involving Taiwan or the South China Sea where decision timelines are compressed. Second, the U.S. faces a structurally analogous problem: CSET's Emelia Probasco, writing in Foreign Affairs, warns that deep AI integration into military operations risks hollowing out the institutional judgment it is meant to support. Both major powers are navigating the same dilemma from different doctrinal starting points, with no established arms control framework to manage the interaction between their respective AI-military systems.