The Gist: Geopolitics & Sovereign Positioning

Wednesday, March 04, 2026

Top Line

• Pentagon-Anthropic clash escalates into defining test of AI sovereignty: The DoD has terminated its $200M Anthropic contract and designated the company a supply chain risk after disputes over autonomous weapons and mass surveillance restrictions — while OpenAI secures expanded Pentagon access, creating a bifurcation in the US AI defence industrial base. New York Times, BBC, Lawfare

• China accelerates military AI integration as US debates ethical guardrails: Chinese forces are rapidly deploying AI across military operations while the US military's commercial AI partnerships face internal restrictions, potentially widening the capability gap in autonomous systems and battlefield intelligence. Foreign Affairs, ChinaTalk

• Iran conflict becomes testing ground for AI-enabled warfare: US forces are using costly precision missiles against cheap Iranian drones in combat that's generating AI-manipulated battlefield imagery at scale, while prediction markets hit record volumes on war-related bets — demonstrating both AI's military applications and its information warfare risks. Bloomberg Tech, Financial Times

• Trump administration debates Tencent gaming carve-out in broader China tech restrictions: Internal administration discussions on whether to exempt Tencent's investments in Epic Games and other US studios from security reviews signal potential fractures in unified approach to Chinese tech dependencies. Financial Times

• UK commits £40M to sovereign AI research lab amid push for strategic autonomy: Britain launches state-backed frontier AI research facility targeting breakthroughs in science, healthcare, and transport — a modest but symbolic effort to reduce dependence on US AI capabilities. Financial Times

Key Developments

Pentagon Fractures US AI Industrial Base With Anthropic Ban

The US Department of Defense has terminated its $200 million contract with Anthropic and designated the AI company as a supply chain risk, barring all military contractors from using its Claude model. The rupture stems from Anthropic's refusal to remove restrictions on its technology being used for mass surveillance of Americans or fully autonomous weapons systems — restrictions the company maintained even after signing the Pentagon contract in 2025. The DoD gave Anthropic a deadline to comply before pulling the contract, according to BBC and EFF.

Meanwhile, OpenAI has secured expanded Pentagon access, with CEO Sam Altman telling staff the company "doesn't get to make the call about what the Defense Department does" with its technology, per Bloomberg. OpenAI has amended its contract to explicitly prohibit use for domestic mass surveillance or intelligence services — restrictions Altman admitted made the original deal look "opportunistic and sloppy" — but maintains broader Pentagon cooperation. The Guardian reports Anthropic is nearing a $20 billion annual revenue run rate despite the Pentagon dispute, more than doubling its late 2025 performance, according to Bloomberg.

Why it matters: The US now has competing AI defence contractors with fundamentally different restrictions on autonomous weapons and surveillance — creating strategic uncertainty about which capabilities the military can reliably access while signalling to allies and adversaries that America's AI defence posture lacks coherence.

What to watch: Whether other US AI companies align with OpenAI's permissive approach or Anthropic's restrictive stance, and whether Congress intervenes to standardise acceptable use policies across military AI contractors.

China's Military AI Push Proceeds Unencumbered by Ethical Debates

Chinese forces are rapidly integrating AI into military modernisation with no equivalent internal debate over autonomous weapons or surveillance restrictions, according to Foreign Affairs analysis by Georgetown's Center for Security and Emerging Technology. While US military contractors navigate commercial AI companies' ethical constraints, China faces no such friction between its defence industrial base and AI developers. The head of Alibaba's Qwen AI model development has stepped down following a major model launch, per TechCrunch and Bloomberg, though the strategic implications of the departure remain unclear.

CSET experts warned in multiple interviews that the US-Anthropic standoff could widen capability gaps in autonomous systems at precisely the moment when AI-enabled warfare is being tested in real combat conditions. The New York Times characterised the dispute as "a decisive moment for how A.I. will be used in war."

Why it matters: China can move faster on military AI integration without navigating commercial sector ethics constraints, potentially gaining first-mover advantages in autonomous systems — while the US debates whether its own defence contractors should face restrictions its adversaries don't observe.

What to watch: Whether China demonstrates AI-enabled military capabilities that exceed US systems constrained by commercial ethics policies, forcing a policy reckoning in Washington.

Iran Conflict Becomes AI Warfare Laboratory and Disinformation Testbed

The US-Israeli strikes on Iran are generating dual AI challenges: militarily, the US is using expensive precision missiles against cheap Iranian drones in asymmetric warfare that Bloomberg analysts say raises sustainability questions; informationally, AI-manipulated satellite imagery and battlefield videos are flooding social media, with the Financial Times documenting how modified strike images circulate as the conflict intensifies. X announced it will suspend creators from revenue-sharing for 90 days if they post unlabelled AI-generated armed conflict videos, per TechCrunch and The Guardian.

Prediction markets hit record volumes on war-related bets, with Bloomberg reporting Polymarket opened more than a dozen new Iran conflict contracts including bets on regime change and Houthi strikes. Tech companies are closing Middle East offices and moving staff to remote work as the region — which has positioned itself as an AI hub following billions in investment — faces operational disruption, according to BBC.

Why it matters: Real-world conflict is simultaneously testing AI's military applications and its capacity to generate persuasive disinformation at scale — demonstrating both the technology's strategic value and its information warfare risks.

What to watch: Whether AI-generated battlefield content decisively shapes public perception of the conflict, and whether the US develops effective countermeasures to AI disinformation in contested information environments.

US Strategic Dependencies: Tencent Exemption Debate and Energy Pressures

Internal Trump administration discussions are exploring whether to exempt Tencent's investments in Epic Games (Fortnite creator) and other US game studios from ongoing security reviews, per Financial Times. The debate signals potential fractures in unified approaches to Chinese technology dependencies, with economic and cultural considerations competing against security concerns. Separately, soaring US electricity prices driven partly by AI datacenter demand have politicians "scrambling" for solutions ahead of congressional elections, with Bloomberg reporting Trump administration efforts to ease voter concerns may be complicated by Iran war-driven energy costs.

The Financial Times notes Trump is pushing datacenter operators to build their own power supplies as AI demand adds to costs, though experts expressed doubt that tech companies' promises can check fast-rising electricity prices even without war impacts.

Why it matters: The US faces simultaneous pressures from Chinese technology dependencies in consumer sectors and domestic energy constraints from AI infrastructure — both creating political vulnerabilities that adversaries can exploit or that constrain US AI scaling.

What to watch: Whether the Tencent exemption debate establishes precedent for carving out economically significant Chinese tech from blanket restrictions, potentially weakening unified technology competition frameworks.

UK and European Efforts at AI Strategic Autonomy

Britain is investing £40 million in a new state-backed frontier AI research laboratory seeking breakthroughs in science, healthcare, and transport, per Financial Times. While modest compared to US and Chinese AI investments, the initiative represents explicit pursuit of strategic autonomy in AI capabilities. The European Commission is hosting a stakeholder forum on "AI for 3D Digital Twins in Cultural Heritage" on March 23, according to EU documentation — a sector-specific AI application that demonstrates Europe's focus on differentiated capabilities rather than direct competition in frontier models.

Why it matters: European efforts at AI sovereignty remain vastly under-resourced compared to US-China competition, but targeted investments in specific application domains could create pockets of European advantage and reduce dependence on American AI infrastructure.

What to watch: Whether UK and European AI research investments translate into commercially viable capabilities or remain primarily academic exercises without market impact.

Signals & Trends

The "ethical AI" competitive disadvantage is now explicit doctrine: The Pentagon-Anthropic split codifies what was previously implicit — that US military AI adoption will be constrained by commercial sector ethics policies that adversaries don't observe. This creates measurable capability gaps: while the US debates autonomous weapons restrictions, China integrates AI into military operations without friction between defence and commercial sectors. The question is no longer whether ethics constrain capability, but whether those constraints are strategically sustainable. Watch for Congressional attempts to mandate "ethics-free" military AI development pathways, or alternatively, for US development of AI capabilities specifically designed to function within ethical constraints that adversaries can't match.

AI is replacing export controls as the primary technology competition mechanism: Traditional export controls on chips and manufacturing equipment remain in place, but the real competition is shifting to who can deploy AI fastest in military, economic, and social domains. The Anthropic-Pentagon dispute isn't about restricting technology transfer to China — it's about whether the US can match China's pace of domestic military AI integration. Similarly, the UK's £40M research lab isn't trying to block Chinese AI development but to create sovereign British capabilities. This represents a fundamental shift from containment to competitive deployment as the primary technology strategy.

Global South positioning on prediction markets and AI governance: The surge in war-related betting on prediction markets, reaching record volumes per Bloomberg, demonstrates how AI-enabled platforms operate beyond traditional regulatory reach — and disproportionately in jurisdictions outside US and European control. Polymarket opening contracts "beyond Washington's reach" signals emerging economies becoming venues for AI applications that developed nations find politically unacceptable. This pattern extends beyond gambling to content moderation, surveillance technology, and autonomous systems — with Global South nations positioned as regulatory havens for AI capabilities restricted in the West, creating leverage they didn't possess in previous technology cycles.