War, Politics and Compute Power
War is politics by other means – and today, supercomputers, AI and platforms are the new “means” in US–China tech rivalry, compute sovereignty and AI governance.
How AI and microchips are redefining the US-China tech rivalry
War, as Clausewitz reminded us, is the continuation of politics by other means. In the 21st century, those “other means” now decisively include supercomputers, platforms, semiconductors - and increasingly, artificial intelligence.
From Clausewitz to Compute Power
Clausewitz’s core insight was that war is not an aberration from politics but an extension of it, shaped by strategic calculation, economic interests, and social forces. This grammar of power is now being rewritten in the language of chips, cloud, EV platforms, AI models, and national‑scale compute infrastructure.
The recent decision by the US Department of Defense to add Chinese tech champions such as Alibaba, BYD and Baidu to its “Chinese military companies” list is a textbook illustration of that evolution. The designation is not just a legal classification; it is a political act that constrains capital flows, reshapes supply chains, and signals to allies and markets that key technologies are now understood as instruments of state power.
Tech Firms as Strategic Actors
The expanded US Section 1260H list spans a broad swathe of China’s technology ecosystem: e‑commerce and cloud (Alibaba), search and AI (Baidu), electric vehicles (BYD and Nio), battery manufacturers, lidar providers, and display‑panel producers. These are the connective tissues of China’s digital and industrial economy, touching everything from logistics and payments to mobility and sensing.
On paper, the list blocks designated firms and their controlled entities from securing US defence contracts and complicates their access to US capital markets. In practice, it does something more profound: it formally recodes these companies from “commercial entities” to “strategic assets” embedded in a military‑civil fusion architecture. Once that recoding happens, every subsequent policy decision - export controls, procurement rules, sanctions, investment screening - is filtered through a national security lens.
Supercomputers as Geopolitical Infrastructure
Against this backdrop, China’s new LineShine supercomputer (HPC) is not just a scientific instrument; it is geopolitical infrastructure. In the recent post I wrote about LineShine, I explored how its debut at number one on the June 2026 TOP500 list, overtaking the US El Capitan system, signals a new phase in compute sovereignty and AI capability.
LineShine delivers verified exascale performance using domestically produced CPUs, a proprietary interconnect and a sovereign operating system stack. That combination matters: it demonstrates resilience under export controls, maturation of China’s domestic chip ecosystem, and a commitment to sovereign control over high‑end compute for workloads such as climate modelling, weapons simulation and AI training.
When you put LineShine alongside Beijing’s broader distributed AI computing initiatives, you see a coherent strategy: build a national compute backbone that can support brain‑scale models, industrial optimisation and defence applications without depending on foreign GPUs or software stacks. In Clausewitzian terms, compute power itself is becoming one of the “other means” through which politics continues.
For readers who want to go deeper into LineShine specifically – its architecture, performance profile and strategic implications – I unpack that in more detail in my earlier post: China’s LineShine supercomputer and the geopolitics of compute.
Technology, AI and Geopolitics: One System
None of this is historically unprecedented. Railways and telegraphs underwrote 19th‑century empire; oil, nuclear reactors and satellite constellations defined much of 20th‑century geopolitics. Control over key technologies has always meant leverage over logistics, intelligence, economic growth and military reach.
What is different now is the density and pervasiveness of the AI‑compute nexus. Logistics platforms, digital payments, EV ecosystems, cloud services and social media are no longer peripheral to statecraft; they are the infrastructure through which trade, influence and coercion flow. Supercomputers like LineShine sit behind this surface layer, providing the raw calculation and model‑training capacity that turns data into operational advantage across civilian and military domains.
By designating AI‑intensive firms such as Baidu, Alibaba and robotics or sensor manufacturers as “Chinese military companies,” Washington is effectively acknowledging that AI capabilities and high‑end compute are dual‑use by default. Systems that optimise logistics, enhance surveillance, assist intelligence analysis and operate autonomous platforms cannot be neatly separated into “civilian” and “military” buckets, especially under a deliberate military‑civil fusion strategy.
Navigating the New Strategic Landscape
For both policy professionals and technically literate folks, several implications follow.
- Export controls and investment screening will increasingly focus on the full AI stack: compute, interconnects, operating systems, model development, data access and specialised talent – not just traditional hardware categories. LineShine’s ascent with domestic CPUs and interconnects is likely to be read in Washington and allied capitals as a warning that compute sovereignty is now a moving target, not a static constraint.
- Corporate governance must assume that geopolitics is now a core dimension of technology risk. For firms operating across US–China fault lines, particularly in AI‑adjacent sectors, board‑level discussions will need to treat defence‑linked designations, national lists and compute export controls as recurring features of the operating environment rather than edge cases.
- AI governance debates that focus solely on ethics and safety, without engaging with the strategic deployment of AI and compute in great‑power competition, will feel increasingly detached from reality. The same architectures that power recommendation systems, fraud detection, logistics optimisation and generative models are being treated as levers of national power - backed by infrastructure like LineShine - and policy frameworks must be honest about that.
If war is politics continued by other means, then AI and compute are now among those means: embedded in how states categorise companies, control capital, shape standards and redraw the boundary between civilian innovation and military capability. The challenge for both policy and technology communities is to accept that entanglement as a starting point - and then decide what responsible practice looks like inside a world where a change in the TOP500 rankings or an update to a “military company” list is also a political signal.
