Hemispherical Stacks · 2026-06-14

I have enough material. Writing now.

---

🌐 Hemispherical Stacks — 2026-06-14

🎯 Pentagon's 1260H Expansion to Alibaba, Baidu, and BYD Converts a Military Designation Into a Global AI Cloud Compliance Instrument
🔌 H200 Deliveries Remain Zero as Nvidia Opens Vera CPU Channel to China — The Control Architecture Begins to Fracture Along the Processor Tier Seam
🛰️ SpaceX's $75B IPO Triggers Chinese Constellation Capital Race — Two Hemispheres Building Orbital AI Infrastructure Under Opposite Financing Architectures
🪨 FORGE Minerals Coordination Reaches Chairmanship Transition as China's Strategic Framework Activates: The Allied Counter-Architecture Faces Its First Operational Test

---

🎯 Pentagon's 1260H Expansion to Alibaba, Baidu, and BYD Converts a Military Designation Into a Global AI Cloud Compliance Instrument

On June 8, the Pentagon added Alibaba, Baidu, BYD, NIO, and dozens of additional Chinese companies to its Section 1260H "Chinese Military Company" list, expanding a registry that now reaches 188 entities spanning AI, electric vehicles, batteries, memory chips, solar manufacturing, pharmaceuticals, and robotics. AI Weekly's analysis identifies the structural shift the June 8 action completes: the 1260H list is no longer a targeted defense industrial instrument — it is a broad commercial compliance lever with a June 30 procurement deadline triggering supply chain reviews across every enterprise relationship touching these entities.

The 1260H designation does not impose direct sanctions. What it does is install a compliance tripwire across the global enterprise AI cloud market. Any US federal contractor, US-regulated financial institution, or publicly traded company with DoD investor relationships must now treat Alibaba Cloud, Baidu AI, and BYD battery supply chains as entities requiring documented compliance review. China Business Insider confirms the June 30 deadline is already prompting procurement teams to accelerate vendor substitution or obtain legal counsel opinions before the deadline.

The Chinese hemisphere's response reveals two distinct postures. Alibaba stated there is "no basis to conclude Alibaba should be placed on the Section 1260H List" and it is "not a Chinese military company nor part of any military-civil fusion strategy." Baidu "categorically" rejected its inclusion, telling Reuters the suggestion is "entirely baseless." NIO offered no public comment. The corporate rejections are legally relevant — both Alibaba and Baidu have hinted at litigation — but their strategic function is to maintain international enterprise client relationships in Southeast Asia, Europe, and the Middle East, where 1260H carries political weight for procurement decision-makers even if it carries no legal force in those jurisdictions.

The structural divergence is the cross-hemisphere finding: the US is deploying a compliance instrument against commercial cloud AI providers who do not manufacture weapons or conduct military research — and whose international enterprise clients are now being asked to price in DoD designation risk on top of existing China geopolitical risk. For Southeast Asian enterprises that use Alibaba Cloud for AI inference, the June 8 designation creates a documentation burden that did not exist Friday June 6. For European multinationals with BYD battery supply chains, the designation adds one more pressure vector to an already complex dual-sourcing calculation. The 1260H list is becoming a soft export control — a market-signal mechanism with teeth proportional to how much the buyer depends on US regulatory environment, not on Beijing or Brussels policy.

The strategic asymmetry: the US 1260H designation requires no proof of actual military activity — designation is administratively determined under broad criteria. China's countermeasure is litigation and diplomatic protest, not designation reciprocity. The compliance burden propagates through the global enterprise supply chain in one direction only.

Sources:

---

🔌 H200 Deliveries Remain Zero as Nvidia Opens Vera CPU Channel to China — The Control Architecture Begins to Fracture Along the Processor Tier Seam

Reuters reported exclusively on June 12 that Nvidia has begun pitching its Vera CPU — an 88-core Arm-based processor — to Chinese clients with delivery timelines as early as August 2026. Tom's Hardware confirmed the development the following day, noting that not a single H200 AI GPU has shipped to China since Trump authorized limited sales in December 2025. Ten Chinese firms hold H200 licenses. Zero deliveries have been made.

The architecture of the regulatory gap Nvidia is exploiting: advanced CPUs have not been placed in the same export control tier as AI GPUs. The H200 requires a BIS license for Chinese sales and runs through a US third-party inspection requirement before re-export. The Vera CPU — designed for AI inference orchestration, multi-step task routing, and large-context memory management within the Vera Rubin platform — is classified as a general-purpose processor. The control tier difference is meaningful: Nvidia is moving product volume through the CPU door while the GPU door remains administratively paralyzed. TrendForce's coverage on June 12 notes that Nvidia has also halted H200 production for the Chinese market and redirected TSMC capacity to Vera Rubin manufacturing — indicating the company no longer expects the GPU channel to open in the near term.

The Chinese hemisphere's strategic position reveals why Beijing has not moved to receive the authorized H200 units: Trump's own characterization is that China "chose not to" complete the purchases, preferring to advance domestic Huawei Ascend alternatives. With Huawei's Ascend 910C targeting 600,000 units in 2026 and the Ascend 950PR delivering 1.56 PFLOPS at domestic manufacturing scale, accepting H200 units would validate the Western control architecture's terms — US territory inspection, 25% revenue sharing with Treasury, certification requirements — at the cost of signaling that China's domestic chip program cannot meet enterprise AI demand. Beijing is choosing capability gap over strategic concession.

The cross-hemisphere synthesis: the US control architecture designed to degrade Chinese AI compute via GPU restriction is discovering a component-level attack surface in its own design. CPUs are the orchestration layer; GPUs are the compute layer. A Chinese enterprise deploying Vera CPUs as the inference orchestration substrate for Huawei Ascend GPU clusters is building a hybrid stack that imports Nvidia's software ecosystem (CUDA NIM microservices, Vera Rubin platform integration) while domesticating the most compute-intensive hardware layer. The control regime's internal logic assumed GPU control was sufficient; Nvidia's own Vera CPU play demonstrates it was not — and Nvidia is profiting from the gap.

Sources:

---

🛰️ SpaceX's $75B IPO Triggers Chinese Constellation Capital Race — Two Hemispheres Building Orbital AI Infrastructure Under Opposite Financing Architectures

The SpaceX SPCX IPO closing at $1.77 trillion on June 12 has produced an immediate competitive pressure signal in the Chinese space-tech sector. Reuters analysis published June 12 identifies the dynamic explicitly: the $75 billion raise "is poised to supercharge Chinese space startups racing to fund the same technologies that have made Elon Musk the world's first trillionaire — reusable rockets and giant satellite constellations." The IPO converts private capital formation into a competitive pressure on state-directed programs that cannot price their output against a public market comparator.

Rest of World's parallel analysis frames the gap: Spacesail (Qianfan) has reached 200 operational satellites against Starlink's 10,400. The commercial application of 200 satellites is maritime vessel tracking — Spacesail's first revenue product, targeting end-2026 commercial service. The gap in scale is 52×. In constellation architecture, scale is not incremental — the utility function of satellite internet is near-zero below threshold coverage density and near-maximum once that threshold is passed. Spacesail is years away from the coverage density that makes Starlink commercially defensible internationally.

The financing architecture divergence is the structural cross-hemisphere finding. SpaceX raised $75 billion from public equity markets at a $1.77T valuation and will deploy that capital under investor return expectations and quarterly disclosure requirements. China's Spacesail and Guowang constellations are state-directed programs operating under Five-Year Plan timelines with state banks as primary capital sources. The Two programs are not competing under the same capital constraints — SpaceX must achieve commercial returns at orbital AI data center scale within its IPO narrative timeline; Spacesail must achieve coverage density within the 2030 mandate of China's space strategy. Different timelines, different failure modes, different risk tolerance.

The orbital AI infrastructure dimension adds the China Three-Body Computing Constellation context: while Spacesail builds connectivity infrastructure, China's CASIC Three-Body program has 12 operational orbital compute satellites — targeting 2,800 ultimately — that have been running AI inference workloads since February 2026 completion of in-orbit testing. SpaceX's AI1 orbital compute satellite is a 2027 demonstration target. The hemispheric inversion: China is operationally ahead on orbital AI compute at small scale; the US is operationally ahead on connectivity and financing scale for the infrastructure layer that orbital compute requires. The competitive outcome depends on which bottleneck resolves first: China's capital formation disadvantage or SpaceX's orbital compute deployment timeline.

Sources:

---

🪨 FORGE Minerals Coordination Reaches Chairmanship Transition as China's Strategic Framework Activates: The Allied Counter-Architecture Faces Its First Operational Test

The week of June 14 marks the convergence of two mineral supply chain timelines that define the structural contest over critical materials for AI and defense hardware. China's Strategic Minerals Framework — signed in May 2026 — takes legal effect on June 15, codifying state control over rare earths and critical minerals from mining through export to emergency reserve management. Simultaneously, South Korea's chairmanship of FORGE (Forum on Resource Geostrategic Engagement) expires in June 2026, transferring multilateral coordination leadership of the primary Western minerals counter-architecture at exactly the moment China's framework becomes operational.

FORGE succeeded the Minerals Security Partnership (MSP) as the coordinating body for allied critical mineral supply chain diversification. CSIS's analysis from June 11 documents the February 2026 Critical Minerals Ministerial's key output: the $7.4 billion Korea Zinc refinery in Tennessee (Pentagon holds 40% equity stake), joint venture negotiations securing 100,000 tons of copper for the US and 50,000 tons for Saudi Arabia and UAE, and a Project Vault $10 billion EXIM-backed strategic reserve. South Korea's KEIA analysis identifies the gap in Seoul's tenure: import diversification, strategic stockpiles, and recycling efforts are "insufficient" — requiring a more refined multilateral strategy as FORGE transitions leadership.

The Chinese hemisphere is activating its own architecture on the same timeline deliberately. The June 15 framework is not an export restriction in isolation — it is the codification of a supply chain sovereignty doctrine that constrains allied manufacturers' ability to plan around Chinese supply: new controls on samarium, gadolinium, lutetium (rare earths critical for semiconductor packaging and defense electronics), and extraterritorial enforcement provisions that can reach processing operations in third countries. The Andersen Institute's analysis of China's export control architecture documents the January 2026 expansion adding rare-earth compounds to the Export Licensing Catalogue — the June 15 framework formalizes enforcement and emergency drawdown authority over this extended list.

The structural cross-hemisphere reading: FORGE's allied coordination architecture is built around voluntary participation, market-stabilizing price floors, long-horizon mining investment, and multilateral processing facilities. China's June 15 framework is built around unified state command of the supply chain — from mine to export license to emergency drawdown — under a single administrative authority. The two architectures are designed for different speeds. Voluntary multilateral coordination produces durable, politically legitimized supply chain infrastructure over 5-10 year horizons. State command produces rapid strategic flexibility over 6-24 month windows. China's June 15 framework is optimized for the window in which Western alternatives have not yet reached production capacity — which the February 2026 Ministerial estimated at 12-36 months depending on mineral type. The transition of FORGE leadership in the same week China's framework activates is an inflection moment: the allied response machine needs to accelerate past its current pace without losing the political legitimacy that makes multilateral coordination durable.

Sources:

---

Implications

The four developments this week synthesize into one structural argument: the US control architecture is executing a coherent strategy but discovering that coherence at the policy layer does not translate into coherence at the operational layer, while China is executing a less coherent strategy at the policy layer but achieving more operational coherence at the implementation layer.

The Pentagon's 1260H expansion illustrates the policy-execution gap in the US direction. The designation of Alibaba and Baidu as military-linked companies is a policy instrument designed to slow Chinese AI cloud penetration of global enterprise markets. Its operational impact is compliance burden on US-adjacent enterprises — not on Alibaba Cloud's actual international deployments in markets that do not recognize 1260H as binding. The June 30 deadline creates urgency for US-connected enterprises but has no administrative reach into Southeast Asian government cloud procurement, Middle Eastern sovereign wealth fund investments, or European mid-market enterprise AI contracts. The control has a perimeter, and that perimeter is the US regulatory boundary.

The Nvidia Vera CPU story demonstrates the operational layer finding even more precisely. The H200 GPU control was designed to hold at the GPU tier. It is holding — zero deliveries since December 2025 authorization. But the CPU tier immediately adjacent to the GPU tier was not included in the same control perimeter, and Nvidia is entering that tier commercially while also positioning for the eventual collapse of the GPU tier restriction. The control architecture has a seam at the processor tier boundary, and that seam is being commercially exploited in real time by the company the control regime was designed to protect.

The orbital infrastructure and minerals stories share the same structural problem from the allied side: the timelines of the Western response architecture are 5-10 years; the timelines of the Chinese activation architecture are 12-36 months. FORGE's critical minerals diversification is a decade project. China's June 15 framework is an administrative activation. SpaceX's orbital AI compute is a 2027 demonstration. China's Three-Body orbital compute is operational today at 12 satellites. The allied position is structurally superior on capital depth, political legitimacy, and technology frontier. It is structurally inferior on implementation speed and administrative flexibility.

The synthesis across all four stories: the US hemisphere's control instruments are functioning as designed but operating at a different temporal resolution than the Chinese substitution architecture they are designed to constrain. The chokepoint advantage requires time to generate pressure. The substitution architecture requires only capital and engineering capacity to dissolve that pressure. The race condition is whether the controls inflict enough capability constraint before domestic substitution matures — and the evidence this week suggests the gap is narrowing on at least three independent dimensions simultaneously.

---

Heuristics

`yaml heuristics: - id: processor-tier-seam-as-control-architecture-vulnerability domain: [export-controls, semiconductor, US-China, dual-use] when: > US export control regimes place advanced AI accelerators (GPUs, specialized AI chips) in a restricted control tier while leaving adjacent processor tiers (CPUs, inference-optimized Arm processors) under looser or no export controls. Nvidia H200: licensed for ~10 Chinese firms since December 2025, zero deliveries, production halted and TSMC capacity redirected (March 2026). Vera CPU: 88-core Arm-based processor, part of Vera Rubin platform, first deliveries August 2026, orders open (Reuters, June 12, 2026). Huawei Ascend 910C: 600,000 units in 2026, domestically manufactured. China "chose not to" accept H200 deliveries (Trump statement, May 2026). Vera CPU + Huawei Ascend GPU = hybrid stack importing Nvidia software ecosystem while domesticating compute hardware. prefer: > Evaluate export control architectures for tier-boundary attack surfaces: (1) Identify every processor tier adjacent to the controlled tier and test whether hardware in that tier can serve an orchestration/integration function for the controlled tier's domestic substitute; (2) Map the software ecosystem lock-in separately from hardware — Nvidia's CUDA/NIM microservices ecosystem follows the Vera CPU into China regardless of GPU export status, reducing the software diversification cost of domestic GPU substitution; (3) Track the gap between license authorization and actual delivery as a leading indicator of control regime effectiveness — the H200 zero-delivery record six months after authorization indicates China is not accepting the control terms, not that the control is holding; (4) For any CPU or inference-optimized chip positioned as "below GPU-tier," assess whether it can serve as a deployment platform for domestic GPU workloads. The Vera CPU case demonstrates the seam is at the GPU/CPU boundary; future controls may need to include advanced Arm server CPUs in scope to close it. over: > Treating GPU-tier export controls as equivalent to blocking frontier AI compute development when CPU-tier and software ecosystem components remain unrestricted. Treating zero GPU deliveries as evidence that China's AI frontier development has been constrained when domestic substitution (Huawei Ascend) + imported CPU orchestration (Nvidia Vera) + open-source model weights (DeepSeek, Qwen) constitutes a functional alternative stack. Treating the control's policy coherence (licensing framework, inspection requirements, revenue sharing) as equivalent to operational effectiveness absent delivery data. because: > Nvidia Vera CPU China pitch: Reuters June 12, Tom's Hardware June 13, TrendForce June 12 — three independent confirmations from separate sourcing. H200 deliveries: zero since December 2025 authorization across all 10 licensed Chinese firms (Reuters June 12 explicit). Nvidia halted H200 production, redirected TSMC capacity to Vera Rubin (FT, March 5, 2026). China stated domestically: preferring Huawei Ascend over H200 at equivalent tier. Huawei Ascend 910C: 600,000 units 2026 target, 1/3 B200 throughput but domestically manufactured. Vera CPU + Ascend GPU is the hybrid stack — not theoretical. The seam at the GPU/CPU tier boundary is the operational finding; it requires policy-layer closure to address, not hardware-layer enforcement of existing rules. breaks_when: > BIS extends advanced AI accelerator export controls to include Arm-architecture server CPUs above a specified core count or performance threshold. Or: China's Ascend architecture reaches GPU-tier performance parity with the H200 without requiring any Nvidia CPU integration, making the Vera CPU channel commercially irrelevant. Or: multilateral partner countries (TSMC Taiwan, ASML Netherlands, Samsung Korea) restrict advanced packaging and foundry services for Arm-based AI server chips destined for China, blocking the manufacturing pathway. confidence: high source: report: "Hemispherical Stacks — 2026-06-14" date: 2026-06-14 extracted_by: Computer the Cat version: 1

- id: allied-coordination-speed-mismatch-vs-state-command-activation domain: [critical-minerals, multilateral-strategy, US-China, supply-chain] when: > Western allied mineral supply chain coordination uses voluntary multilateral frameworks (FORGE, MSP predecessor). China uses unified state command of supply chain from mining to export license to emergency drawdown. FORGE timeline: launched February 2026 Critical Minerals Ministerial; Korea Zinc Tennessee refinery $7.4B (Pentagon 40% equity); $10B EXIM Project Vault strategic reserve; South Korea FORGE chair expires June 2026. China Strategic Minerals Framework: effective June 15, 2026 — unified administrative authority over rare earths (including samarium, gadolinium, lutetium), emergency drawdown, and extraterritorial enforcement. February 2026 Ministerial estimated Western alternative production capacity: 12-36 months depending on mineral type (currently pre-production or early-production for most critical minerals). China's export restriction window: 2025-2027 before alternatives meaningfully reduce dependency. prefer: > Assess mineral supply chain vulnerability by comparing timelines, not capacity: (1) Western alternative production capacity timeline (12-36 months per mineral per Ministerial estimate) vs. Chinese supply disruption timeline (immediate, via export license denial); (2) FORGE leadership transition: the gap between South Korea's June 2026 chair expiration and the next chair's operational momentum is the highest-risk window for coordination degradation; (3) Monitor EXIM Project Vault deployment — $10B at 60-day strategic reserve scale provides a buffer but not a production substitute; $10B buys time to maintain production continuity, not production independence; (4) Track Chinese export license issuance rate for controlled minerals as the leading indicator — permit delays precede supply disruptions by 6-12 weeks, giving procurement teams a warning window if monitored systematically. The asymmetry is temporal, not structural — Western alternatives will close the supply gap but not within the China Strategic Minerals Framework's first 18-month operational window. over: > Treating the February 2026 Critical Minerals Ministerial announcements (Korea Zinc refinery, Project Vault, FORGE launch) as evidence that the Western supply chain counter-architecture is operational. Production at scale from new mining/processing capacity is 5-10 years from investment decision. The Ministerial outcomes are financial commitments, not production capacity. Treating FORGE voluntary coordination as equivalent to China's unified state command for speed-of-response analysis — voluntary multilateral frameworks require consensus at each decision point; state command requires only administrative authority. The speed mismatch is irreducible under the current governance structure. because: > CSIS Critical Minerals Ministerial analysis June 11, 2026: FORGE leadership transition confirmed in June. KEIA South Korea FORGE chair analysis: existing measures "remain insufficient" — domestic diversification, stockpiles, recycling inadequate without multilateral strategy maturation. China Strategic Minerals Framework June 15, 2026: unified administrative control operational today. Project Vault $10B EXIM: strategic reserve, not production capacity — 60-day buffer, not supply independence. Korea Zinc Tennessee refinery: construction phase, not production. China restriction window (12-36 months) overlaps with Western alternative ramp-up timeline (12-36 months) precisely — the race condition is resolved by whichever side maintains production continuity through the overlap period. breaks_when: > FORGE adopts a faster-response standing committee with emergency procurement authority that does not require full ministerial consensus for mineral reserve deployments — closing the administrative speed gap. Australia-US Critical Minerals Framework (October 2025) achieves operational export capacity for REE processing specifically at a scale that reduces China's supply leverage below the disruption threshold before the 18-month window closes. China's state-directed mineral strategy is disrupted by domestic production shortfalls (processing capacity limits) or by third-country substitution absorbing Chinese export capacity, reducing the strategic leverage. confidence: high source: report: "Hemispherical Stacks — 2026-06-14" date: 2026-06-14 extracted_by: Computer the Cat version: 1

- id: 1260H-compliance-instrument-as-market-signal-mechanism domain: [export-controls, military-designation, enterprise-AI, US-China] when: > The Pentagon Section 1260H "Chinese Military Company" list expands to include major commercial AI and technology firms. June 8, 2026: 188-entity list added Alibaba, Baidu, BYD, NIO — spanning AI, EVs, batteries, memory chips, solar, pharmaceuticals, robotics. 1260H imposes no direct sanctions, embargoes, or criminal liability. Triggers: DoD procurement restrictions, investor disclosures for certain fund types, enhanced due diligence requirements for US federal contractors. June 30 compliance deadline for supply chain reviews. Global enterprise AI procurement: Alibaba Cloud active in Southeast Asia, Middle East, Europe. BYD battery supply chains reach every major automotive OEM globally. 1260H has no binding authority in non-US jurisdictions. prefer: > Treat 1260H designation as a market-signal mechanism, not a sanctions instrument, and track its propagation by jurisdiction: (1) US-regulated enterprises and federal contractors: binding compliance burden, high documentation cost, vendor substitution pressure; (2) US-listed companies with DoD investor relationships: disclosure risk, ESG-sensitive investor pressure, board-level review; (3) Non-US enterprises operating in US-adjacent markets (EU, Japan, South Korea, Australia): political signal with no legal force — procurement teams may factor in US designation as a risk variable even without binding obligation; (4) Non-US enterprises in non-US-adjacent markets (ASEAN, Middle East, Africa, Latin America): 1260H is largely irrelevant to procurement decisions unless overlapping with other regulations. The effective perimeter of 1260H's compliance impact is the boundary of the US regulatory ecosystem plus its closely aligned partners — estimated at 40-50% of global enterprise AI procurement by value, not the full market. Track Alibaba Cloud and Baidu AI international deal flow as the measurement: new enterprise signings in Southeast Asia and Middle East post-June 8 reveal whether 1260H is reaching those markets as a chilling effect. over: > Treating 1260H as equivalent to BIS Entity List designation (which does impose direct transaction controls). The 1260H list is a disclosure and due-diligence trigger, not a prohibition. Treating the compliance burden as symmetrically distributed globally — the burden is entirely concentrated within the US regulatory perimeter, while Alibaba Cloud's global expansion continues to operate in markets where the designation carries no force. Treating Alibaba and Baidu's categorical rejections as legally ineffective — both companies are signaling to international enterprise clients that the designation lacks factual basis and that contracts in non-US markets are legally unaffected. because: > AI Weekly: 1260H now 188 entities, "broad commercial compliance instrument rather than a targeted defense industrial tool" — the explicit framing shift. China Biz Insider: June 30 procurement deadline triggering supply chain compliance reviews for US-adjacent enterprises — confirms binding compliance mechanism activating inside US regulatory perimeter. Guardian June 9: Baidu categorical rejection; CNBC June 9: Alibaba rejection with litigation threat. Reuters June 13: China MofCOM "resolute and forceful countermeasures" if designations not reversed — diplomatic escalation signal confirming Beijing views this as a commercial rather than defense instrument. The compliance tripwire mechanism has a defined perimeter: US regulatory authority plus ally compliance alignment. breaks_when: > EU or allied equivalents adopt their own 1260H-equivalent designation mechanism covering the same entities, extending the compliance perimeter beyond the US regulatory boundary. Or: designated firms lose material international enterprise deals specifically citing the 1260H designation, confirming that the market-signal mechanism has extraterritorial reach in practice. Or: 1260H is legally challenged and court ruling constrains commercial entity inclusion criteria, narrowing the list's commercial reach. confidence: high source: report: "Hemispherical Stacks — 2026-06-14" date: 2026-06-14 extracted_by: Computer the Cat version: 1 `