The Eastern Himalaya and the Terms of Stability

Executive summary

The Eastern Himalaya is one of the few places where two nuclear-armed states with incompatible territorial claims share an undemarcated land frontier along which both maintain large forces. That fact alone would justify attention. What makes the Uttara Dialogue necessary is that the dispute has entered a phase that is neither war nor settlement: conduct is calmer in one sector, claims are harder in another, and the machinery built to manage the line is more elaborate than the agreement that defines it.

The report's central finding is that the dispute is legally unresolved but administratively settled. India administers Arunachal Pradesh as a state of the Union. China claims the entire territory and has never accepted the McMahon Line. No treaty delimits the boundary; the agreements of 1993–2013 govern conduct, not title; and neither side accepts adjudication.¹²³ A report that treats either the claim or the administration as the whole truth will be dismissed by one of the two governments it seeks to inform.

The legal record is strong on process and weak on consent. India can show that China attended the Simla Conference, that its representative initialled the 1914 Convention, and that no formal boundary objection was raised for four decades. China can show that it never signed or ratified the Convention and that the alignment was fixed in an exchange of notes between British and Tibetan plenipotentiaries.¹⁴ Both are true. The evidence does not establish which legal case is correct; it establishes that neither is frivolous.

The management record is real but shallow. The October 2024 patrolling understanding stabilised Depsang and Demchok, but its text remains unpublished and it restored access, not the pre-2020 baseline.⁵ The August 2026 eight-point consensus committed the Expert Group on Boundary Delimitation and the Working Group on Border Management, created at the 24th round, to agree terms of reference, alongside two meeting points and two hotlines, and set the 26th Special Representatives round in India in 2027.⁶ It did not publish terms of reference, name a sector, or define "early and substantial harvest". In 2026 the live military file moved east: face-offs in Arunachal Pradesh were confirmed on the record by the Director-General of the Indo-Tibetan Border Police, with PLA tents reported 6–7 km inside India's perception of the LAC.⁷ That figure is credibly reported, not independently verified.

Pressure since 2020 has been layered rather than singular. Military instruments produced the only deaths. Administrative, cartographic, water and export-control instruments produced cumulative leverage without direct force, on different clocks: villages and dams over a decade, renaming annually, licensing weekly.⁸⁹ Intent is stated for naming and export controls and inferred for military and water instruments. The paper keeps capability, intent and effect separate.

Economic exposure is concentrated and asymmetric. Two-way goods trade was $151.1 billion in FY2025–26 on India's basis, with a deficit of $112.16 billion; China supplied about 17 per cent of India's merchandise imports and 85–93 per cent of rare-earth magnet imports depending on the measure.¹⁰¹¹¹² A lethal clash did not reduce trade after 2020, because imports are inputs. Author-constructed scenarios put a contained clash at 0.1–0.3 percentage points of GDP, a prolonged confrontation at 0.9–1.8 per cent, and a systemic shock at roughly 2.5–3.1 per cent.¹³

Deterrence is stable at the top and fragile in the middle. Both sides declare no first use, and both are modernising in ways that could shorten decision time.¹⁴¹⁵ The dangerous thresholds are the first use of firearms at the line and attacks on airbases, command posts or nuclear-adjacent assets, not a deliberate strategic first strike.

Multilateral bodies can legitimise a norm but cannot adjudicate the dispute. The United States recognises Arunachal Pradesh as Indian territory; the Quad, Japan and Australia oppose change by force; the 2023 G20 formula is the highest-consensus sentence available.¹⁶¹⁷¹⁸ None of these texts names the line, and none binds China.

The report sets out four scenarios for 2027–2032 — managed rivalry, limited clash, prolonged confrontation and a conditional settlement track — and 14 sequenced recommendations. None requires either side to abandon its claim. Three tests will show within two years whether the August 2026 text is operative: terms of reference with a non-prejudice clause; activation and testing of the meeting points and hotlines; and an incident-review function that can establish what happened without attributing blame in public.

1. Introduction: a border dispute as a test of order

The Uttara Dialogue 2027 meets in New Delhi on 1 and 2 February 2027. China describes the border as generally stable; India says incidents recur; neither has withdrawn its claim.⁵²²⁰ That combination frames this report and explains why it begins with law and history: the incidents of 2020–2026 are intelligible only against a legal disagreement a century old.

This report uses a fixed formulation, applied without variation across the paper, programme and correspondence:

China claims the entire territory of Arunachal Pradesh and has never accepted the McMahon Line. India administers the state as an integral part of its Union. Along the Line of Actual Control, both sides have built infrastructure and improved their positions since 2020. India records repeated Chinese incursions; China attributes tension to Indian forward policy. India's objection is not to negotiation over the boundary's alignment, but to the premise that the state's status is negotiable at all.

The first sentence states a claim; the second states an administrative fact.¹²¹ Neither asserts a legal conclusion the record cannot carry. Together they avoid two errors that disfigure public commentary: treating a claim as an accomplished transfer of territory, and treating a militarised frontier as a manageable irritant.

The formulation is asymmetric because the two positions are: one side denies the instrument on which the other relies, the other denies that the territory is in question. The asymmetry is in the positions, not in the paper's treatment of them.

The dispute matters beyond the bilateral relationship for four reasons. First, it tests whether a territorial claim can be advanced through administrative and military pressure without crossing the threshold of war; the instruments are familiar — names, maps, roads, forward posts, periodic incursions — but the combination of scale, legal coherence and nuclear risk is not. Second, it is among the most developed experiments in bilateral border management between two major powers, and the architecture of 1993–2013 and 2024–2026 would be studied closely if it failed.²²³ Third, it is coupled to the economic relationship: India's External Affairs Minister has linked border calm to normal ties and, in the same conversation, raised market access, trade balance and supply-chain predictability.²⁴ Fourth, it carries systemic risk, because two states with unresolved land claims, large armies in proximity and deep integration into global supply chains cannot confine a miscalculation to a Himalayan valley.

The report separates three registers that public argument conflates: the legal claim, conduct, and consequence. Sections 2 to 4 address the claim; Sections 5 to 7 address conduct and consequence; Sections 8 to 10 address futures and recommendations. Evidence is labelled. "India holds" and "Beijing argues" mark positions, not findings. Where the record is incomplete, the paper says so rather than filling the gap with inference.

Three propositions organise the analysis. The legal dispute is genuine: each case is internally coherent and rests on a proposition the other rejects. The management regime is real but shallow: it has restored patrolling and created channels while leaving the line undefined and the state's status untouched. China's policy operates at two levels — an absolute public claim and a managerial operational practice — a gap that is stable rather than hypocritical.

2. The line and the law

2.1 The instrument of 1914

The eastern boundary was not created by a single signed treaty. It emerged from a conference China attended, a convention China initialled but did not sign, and an exchange of notes between the British and Tibetan plenipotentiaries to which China was not party.¹

The Simla Conference ran from October 1913 to July 1914. The Convention Between Great Britain, China, and Tibet is dated 3 July 1914 and runs to eleven articles; its preamble names McMahon, Ivan Chen and Lonchen Shatra. Article 9 provides that "the borders of Tibet, and the boundary between Outer and Inner Tibet, shall be as shown in red and blue respectively on the map attached hereto".¹ Article 2 recognises that "Tibet is under the suzerainty of China" and "the autonomy of Outer Tibet", and binds Britain not to "annex Tibet or any portion of it".¹ The convention therefore presupposed Chinese standing in Tibet even as China declined to sign.

The India–Tibet sector was settled separately. Letters between McMahon and Lonchen Shatra on 24 and 25 March 1914, confirmed at the seventh full conference meeting on 22 April and shown on maps attached to the convention, carry the alignment China calls the McMahon Line.²⁵ Chen initialled on 27 April; on 3 July only the British and Tibetan representatives signed. Aitchison records that the convention "after being initialed by the Chinese Plenipotentiary was not signed or ratified by the Chinese Government", and was "accepted as binding by the two other parties as between themselves".¹ Britain and Tibet signed a separate declaration the same day.²⁶

The line runs from the Bhutan–Tibet–India trijunction northeastward along the Himalayan crest to the Tibet–Burma–India trijunction, leaving Tawang and the main valleys south of it. It was drawn by reference to ridges, passes and watersheds, not surveyed coordinates.²⁵ India's maps show it as the state's northern boundary; Chinese maps place the boundary along the Himalayan foothills, so the two positions do not overlap in the eastern sector.²⁷ Beijing's 1962 statement put the Indian claim at "more than 90,000 square kilometres of Chinese territory in the eastern sector", a claim area rather than an agreed measurement.²⁷

2.2 China's legal case

Beijing's objection is not simply that China was absent. Its case, stated in Zhou Enlai's letter of 8 September 1959 and repeated in the government statement of 24 October 1962, rests on four propositions: the boundary was never formally delimited; the eastern sector was fixed in a secret exchange in Delhi on 24 March 1914 behind China's back; Tibet had no right to conclude treaties separately; and the line was a product of British colonial policy imposed during China's weakness.⁴²⁷

The 1962 statement calls it a line the British "attempted to force upon China", "illegal" and never recognised. Zhou wrote that it "was never discussed at the Simla Conference, but was determined by the British representative and the representative of the Tibet local authorities behind the back of the representative of the Chinese Central Government through an exchange of secret notes at Delhi on March 24, 1914"; Tibet had "no right to conclude treaties separately"; and the line "has never been recognized by any Chinese Central Government".⁴

The argument is one of original voidness, not voidability. Because the act was unlawful from the outset, no later Indian act — statehood in 1987, the Ladakh Union Territory in 2019, the names of 2023–2026 — can cure the defect.²⁸ That explains why Chinese spokespersons rarely debate alignment: on this account there is nothing to align.²⁸ One nuance complicates the picture. Zhou's letter of 7 November 1959 proposed that the two armies "each withdraw 20 kilometers at once from the so-called McMahon line in the east, and from the line up to which each side exercises actual control in the west" — treating the line as a reference for disengagement while denying its legality.²⁹ India cites this as engagement in practice; China treats it as a provisional military arrangement with no legal significance.²⁹

2.3 India's legal case

New Delhi's case rests on the 1914 instruments and on Tibet's treaty capacity at the time. The 1961 Report of the Officials states that "the traditional boundary in the Eastern Sector had been formalized in 1914 by an exchange of letters between India and Tibet", that "Tibet had enjoyed the power to sign treaties and to deal directly with neighbouring States on matters regarding the boundary", and that "the Chinese Government had recognised these rights enjoyed by Tibet and had been aware of this formalization".³⁰

India does not argue that Tibet was independent. The same report records that "India did not regard Tibet as independent" and that independent India "had always dealt with the Central Government of China on matters pertaining to Tibet".³⁰ The argument concerns Tibet's capacity to conclude a boundary instrument in 1914, not Tibet's statehood — a distinction frequently misstated on both sides.

India also argues that China's objection is late and inconsistent: Chen knew of and initialled the convention, China raised no formal boundary objection between 1914 and 1959, and Zhou said in January 1959 that the Chinese government "finds it necessary to take a realistic attitude towards the McMahon Line".²⁵ These are Indian characterisations of an incomplete record, and they are contested.²⁵

The Indian case is strong on process and weak on consent; the Chinese case is strong on consent and weak on administration. Neither side has submitted the question to adjudication, and no court or arbitral body has decided it.²⁵ The evidence establishes that both cases are arguable; it does not establish which is correct.

2.4 1962 and its aftermath

Large-scale fighting began on 20 October 1962. China's statement says Indian forces "started a massive general offensive in both the eastern and western sectors" and that Chinese frontier guards "strike back in self-defence"; India's account is that China attacked.²⁷ The date is common ground; the initiation is not.

China announced a unilateral ceasefire on 21 November 1962, effective 22 November, and a withdrawal from 1 December to 20 km behind the LAC as it stood on 7 November 1959 — in the eastern sector, 20 km north of the McMahon Line. All Chinese troops were north of the line by 1 March 1963. China retained Aksai Chin.³¹ Indian analysts infer acquiescence from the withdrawal; Beijing replies that the operations were a border counter-attack and the withdrawal a unilateral peace gesture conferring no recognition.³¹²⁷ The legally significant outcome is the stable de facto line in the east that China has not since crossed in force, alongside its retention of Aksai Chin — the physical foundation of the "package settlement" idea codified in 2005.³

2.5 The management architecture

Between 1993 and 2013 the two governments built a conflict-management regime that deliberately left the legal question untouched. The 1993 agreement requires both sides to "strictly respect and observe the line of actual control" and to "jointly check and determine" disputed segments; references to the LAC "do not prejudice their respective positions on the boundary question".² The 1996 agreement bars the use of military capability, commits each side not to seek "unilateral military superiority", and reaffirms that "no activities of either side shall overstep the line of actual control".²³

The 2005 agreement is the only instrument that sets out the shape of a settlement: "mutually acceptable adjustments to their respective positions", a "package settlement" that "must be final, covering all sectors", taking into account "historical evidence, national sentiments, practical difficulties and reasonable concerns and sensitivities of both sides".³ The 2012 agreement creating the Working Mechanism for Consultation and Coordination states that the body "will not discuss resolution of the Boundary Question"; the 2013 Border Defence Cooperation Agreement adds information exchange, flag meetings and hotlines without delimiting the LAC.³²³³ Read together, these are a conflict-management regime, not a settlement regime.

2.6 From Galwan to the twenty-fifth round

A standoff began in April–May 2020. On the night of 15–16 June troops clashed in the Galwan Valley without firearms; the Indian Army raised its death toll from three to 20, and China confirmed four of its soldiers killed in February 2021.³⁴³⁵ India described the clash as "a result of an attempt by the Chinese side to unilaterally change the status quo"; China accused Indian troops of crossing the border and carrying out "provocative attacks".³⁴

On 21 October 2024 India announced an agreement on "patrolling arrangements along the Line of Actual Control" leading to disengagement; Depsang and Demchok were completed by the end of October and patrolling resumed. The text has not been published.⁵ Beijing described the same event as a "resolution" of "relevant border issues", not a boundary agreement.³⁶

The procedural track moved quickly. The 23rd round was held in Beijing on 18 December 2024;³⁷ the 24th, in New Delhi on 19 August 2025, produced ten points, including the two technical groups, an agreement to reopen three border markets and to share emergency hydrological information;³⁸ the 25th, in Beijing on 25 August 2026, produced eight points published on 26 August: a framework under the 2005 agreement, terms of reference for the groups, two meeting points, two hotlines and a 26th round in India in 2027.⁶

Normalisation proceeded in parallel: Modi and Xi met at Kazan in October 2024 and Tianjin in August 2025; direct flights, the Kailash Mansarovar Yatra and the BRICS summit in New Delhi followed.³⁷ The dispute did not soften. China's Ministry of Civil Affairs published a sixth batch of standardised names for places in Zangnan on 10 April 2026, bringing its total to 112 since 2017.²² India rejected the move, and on 7 August 2026 the Government of India published 27 standard names for Arunachal locations on Survey of India maps.³⁹ The pattern is management without recognition: deconfliction advances while both sides continue unilateral acts that reaffirm their claims.⁶²² The 2026 "early harvest" language is procedural; no sector, line or acreage has been agreed.⁶

2.7 What the record does not settle

Six questions remain open: Tibet's treaty capacity in 1914; the effect of Chen's initialling; whether China acquiesced between 1914 and 1959; the alignment on the ground; the definition of the LAC; and what the 2024 understanding obliges. On the first, China says Tibet had "no right to conclude treaties separately" while India says it "enjoyed the right to sign treaties" and that China knew and acquiesced; no court or arbitral body has decided the point.⁴³⁰ On the second, India treats initialling as evidence of awareness and China as expressly conditional; the Aitchison note does not resolve the legal effect.¹ On the third, the record is incomplete and both readings are defensible.²⁵ On the fourth, the line was drawn on maps, not fixed by joint survey, and the 1993 promise to "jointly check and determine" disputed segments has not been fulfilled.² On the fifth, no jointly agreed map or definition exists, which is the central operational gap.² On the sixth, the text is unpublished, its binding character unclear and its scope unresolved: the arrangement is sector-specific, covering Depsang and Demchok, and it created no inspection, data-exchange or incident-attribution provision.⁵

Both governments can therefore state their positions with internal consistency, and neither can demonstrate them to the standard a court would require.

3. India's position

3.1 Three tiers, and a fourth

India's public position has three tiers. The first is sovereignty over Arunachal Pradesh. The second is the boundary, where India is a negotiating party rather than a claimant demanding recognition. The third is conduct along the LAC. Since 2024 a fourth, quieter tier has been added: economic reciprocity — market access, trade balance and supply-chain predictability — pressed as a test of whether normalisation is worth having. It is stated to China directly but has never been presented publicly as a formal condition, unlike border peace, which India calls a "pre-requisite".³⁷²⁴

The separation is deliberate: sovereignty is asserted as self-evident and not argued, while boundary and conduct language is procedural and conditional. It allows India to negotiate about where the line runs without conceding that the state's status is in question.⁴⁰

3.2 Sovereignty: the self-evident claim

The core formula has been stable and was used repeatedly in 2025–2026. On 25 November 2025, responding to China's detention of an Indian citizen from Arunachal Pradesh transiting Shanghai, the Official Spokesperson said: "Arunachal Pradesh is an integral and inalienable part of India, and this is a self-evident fact. No amount of denial by the Chinese side is going to change this indisputable reality."²¹

On 12 April 2026, responding to a further batch of Chinese names, the same spokesperson said: "India categorically rejects any mischievous attempts by the Chinese side to assign fictitious names to places which form part of the territory of India. Such attempts by China at introducing false claims and manufacturing baseless narratives cannot alter the undeniable reality that these places and territories, including Arunachal Pradesh, were, are, and will always remain an integral and inalienable part of India."⁴¹ On 11 August 2026 the formula was repeated after India itself standardised 27 names and China objected: "Arunachal Pradesh is an inalienable and integral part of India, and this is a fact which is self-evident."⁴²

Three features matter. The claim is phrased as self-evidence, not as an argument about the McMahon Line. It is attached to people as well as territory, since the November 2025 statement arose from consular protection. And India now mirrors China's naming practice, making the contest reciprocal.⁴¹ The Prime Minister's register is different: his Statehood Day message of 20 February 2026 is developmental and cultural, praising "majestic landscapes" and "extraordinary cultural diversity".⁴³ Sovereignty language is delegated to the spokesperson, keeping the Head of Government's statements low in legal content and preserving space for summit diplomacy.²¹

3.3 The boundary: negotiation without prejudice

India accepts that the boundary as a whole has not been delimited and that the LAC "does not have a common understanding in some areas". It does not accept that the eastern sector is undelimited: its case is that the 1914 instruments delimited it, and that the dispute is China's non-acceptance rather than an absence of instruments. The distinction matters, because it is the difference between negotiating an alignment and negotiating a title. The External Affairs Minister told the Lok Sabha on 3 December 2024: "While there is a Line of Actual Control (LAC), it does not have a common understanding in some areas. We remain committed to engaging with China through bilateral discussions to arrive at a fair, reasonable and mutually acceptable framework for a boundary settlement."⁴⁰

The framework is the 2005 agreement and the mechanism is the Special Representatives track. The eight-point consensus of 26 August 2026 anchors negotiations in the 2005 principles. Its second point records both sides' commitment to "advancing negotiations on a framework for the settlement of the boundary question" under that agreement; its third commits the Expert Group on Boundary Delimitation and the Working Group on Border Management, created at the 24th round and placed under the WMCC, to pursue an "Early and Substantial Harvest of boundary delimitation and Border Management", with the first task for both groups being to agree their respective Terms of Reference.⁶ Point 4 routes incidents through existing channels and commits the sides to improving mutual "understanding of LAC in appropriate areas"; point 5 adds two meeting points and two military hotlines in the Eastern and Middle Sectors; point 8 sets the 26th round in India in 2027.⁶

The hedge is the most important sentence in the 2026 documents. Terms of reference are not published, and no sector has been named. "Early and substantial harvest" implies staged delineation before a final settlement — a position India does not advertise at home, because it sits awkwardly with the claim that the state's status is self-evident. Nothing is prejudged until the terms of reference are settled, and the 26th round is due in India in 2027, after the conference.⁶

India's red line is therefore the premise, not the process. It will discuss where the line runs and how the border is managed; it will not accept that whether Arunachal Pradesh is India's is negotiable.⁴⁰ That is the conference's fixed formulation restated in diplomatic practice.

3.4 Conduct: the three principles

India's settled language on conduct is expressed in three principles that "must be observed in all circumstances": both sides should strictly respect and observe the LAC; neither should attempt to unilaterally alter the status quo; and past agreements must be fully abided by in their entirety.⁴⁰ The linkage to the wider relationship is a necessary condition, not a preference: "India-China relations cannot be normal in the absence of peace and tranquility in the border areas."⁴⁰ In 2026 the same condition is repeated as a "pre-requisite for normal ties".²⁴

On disengagement, India's account is that the 2024 understanding has been implemented but that de-escalation has not followed. The Minister of State told the Lok Sabha on 8 August 2025 that the agreement "has since been effected and implemented as per modalities and timelines agreed" and that earlier terms "continue to hold in relevant areas in Eastern Ladakh".⁴⁴ The External Affairs Minister has said disengagement is complete in Eastern Ladakh, while "the next priority will be to consider de-escalation, that would address the massing of troops along the LAC".⁴⁰ After the 36th WMCC meeting on 6 August 2026 the Ministry said the sides would use existing diplomatic and military channels, including the WMCC and local commander meetings, to avoid "misunderstanding and miscalculation".⁴²

3.5 Infrastructure and the civilian face

India presents border infrastructure as developmental, not as a build-up narrative. The Border Roads Organisation recorded its highest-ever expenditure of ₹16,690 crore in 2024–25, with a target of ₹17,900 crore for 2025–26, and dedicated 250 projects in 2024–25.⁴⁵ The External Affairs Minister cites the Atal, Sela and Nechiphu tunnels, the Umlingla Pass road and the extended Zojila opening as evidence that infrastructure "has made such effective counter-deployment possible".⁴⁰

The Vibrant Villages Programme is the civilian face of the same policy. A written reply of 28 July 2026 records 662 priority villages, of which 455 are in Arunachal Pradesh; ₹953.47 crore released; 111 road projects worth ₹2,481.93 crore for 134 unconnected villages; and 528 villages with 4G, 625 with piped water and 621 with electricity.⁴⁶ A later official feature puts the combined outlay of the programme's two phases at ₹11,639 crore and calls border villages "strategic partners in national security and development".⁴⁷ India publishes infrastructure and de-escalation status, but not Chinese force levels opposite Arunachal.⁴⁴ That silence limits what outside analysts can assert about the local balance.

3.6 Red lines, silences and an assessment

India's red lines are compact. No acceptance of Chinese renaming or mapping in Arunachal; such acts are "fictitious names" that "cannot alter the undeniable reality".⁴¹ No unilateral change of the status quo, and past agreements "fully abided by in their entirety".⁴⁰ No normalisation without peace and tranquillity.²⁴ No dilution of "integral and inalienable" and no official description of the state as disputed.²¹ And, by consistent practice rather than explicit statement, no third-party adjudication or mediation: India's documents describe only bilateral mechanisms, and the 2026 text routes every contingency through them.⁶

The silences matter as much. The 2024 arrangement's text, maps and patrol limits are unpublished; Parliament was given outcomes only.⁴⁴ India has not said what "early and substantial harvest" would cover, nor published the terms of reference.⁶ It has set no deadline for de-escalation and made it no formal precondition for the normalisation already under way — flights, visas, pilgrimage, border trade.³⁷ And it does not use "integral and inalienable" for Aksai Chin as it does for Arunachal, a lexical difference that implies a different legal treatment no official explains.⁴⁰

India's strengths are consistency and defensibility: the formula has not changed under provocation. Its weaknesses are opacity and tension. A status described as self-evident sits awkwardly with a track that contemplates delimitation. Its own 2024 text, maps and patrol limits remain unpublished. And a strategy that rests on the LAC cannot fully succeed while the LAC remains undefined.²

4. China's position and strategy

4.1 The claim in Beijing's words

Beijing's claim is total: not to the McMahon Line, but to the territory south of it. Spokesperson Lin Jian set out the current formulation on 25 March 2024: "The China–India boundary has never been delimited. It consists of the eastern, middle and western sectors and the Sikkim sector. The eastern sector, the Zangnan region, has always been Chinese territory. Before India's illegal occupation, China continuously exercised effective administrative jurisdiction over Zangnan. This basic fact cannot be denied. In 1987 India established the so-called 'Arunachal Pradesh' on illegally occupied Chinese territory; China issued a statement at the time."²⁸

The standard formula, repeated after each Indian move, is: "Zangnan is China's territory. The Chinese government has never recognized and firmly opposes the so-called 'Arunachal Pradesh' illegally established by India."⁴⁸ In its 2026 English statement the Ministry uses the pinyin "Zangnan", not "South Tibet": "Zangnan region is China's territory. China does not recognize the so-called 'Arunachal Pradesh' illegally set up by India. India's move of attempting to replace names long been used by China with its so-called 'standard names' is illegal, null and void."⁴⁹

The terminology is substantive. "Zangnan" asserts that the area is southern Xizang, not a separate disputed territory; "South Tibet" makes the same assertion legible in English but invites the Indian reply that it is a strategic fiction.¹⁹ This paper uses "Zangnan" for Chinese usage and "Arunachal Pradesh" for India's. Beijing describes the territory as the Cona, Lhunze, Medog and Zayu counties of the Xizang Autonomous Region, and the Indian state as an entity created in 1987.²⁸ The same template is applied in the west: after India published 28 names in Ladakh in October 2026, the Foreign Ministry called the move "illegal, null and void".⁵⁰

4.2 The strongest Chinese arguments

The legal skeleton is set out in Section 2. Before any rebuttal, three arguments should be stated at their strongest.

The legal defect is original, not incidental. The McMahon Line was not an agreement between two sovereign states. It was drawn in a secret exchange of notes in Delhi on 24 March 1914 between McMahon and the Tibetan representative, behind the back of the Chinese central government and before the Simla Convention was signed. The Chinese representative initialled the convention on 27 April 1914 but did not sign it on 3 July. Tibet was a local administration without independent treaty-making capacity. A secret act by a colonial power and a local authority cannot dispose of Chinese territory, and no successive Chinese government has recognised the line. The defect is original and incurable: the line is void ab initio, not merely voidable, and Indian statehood, Union Territory status and place names cannot cure it.⁴

History and administration support China, not colonial cartography. China administered the area through the Tibetan local government before the British drew the line; Chinese records treat Cona, Lhunze, Medog and Zayu as the relevant counties. The Indian state was created in 1987 on territory China regards as illegally occupied, and the 2019 Ladakh Union Territory is a second unilateral act. India's claim rests on maps inherited from the colonial power, not on a treaty China signed. China's place-name standardisation is domestic administration, legally different from India's unilateral redrawing of administrative boundaries.⁴⁸

China has been the restrained party, and its restraint is conditional. China says it has never sought to hold territory south of the McMahon Line: its forces withdrew after the 1962 counter-attack, and Zhou stated in 1959 that Chinese troops had not crossed the line. China seeks a negotiated package under the 2005 principles, accepts "mutual understanding and accommodation", and has resumed the SR process, border trade and pilgrimage; a river-data mechanism met in September 2026, but data sharing has not resumed.³ In Beijing's account, the changes that destabilised the border were Indian: forward infrastructure, the 2019 Union Territory, the 1987 statehood and the renaming campaign. The 2024 understanding restored patrolling without a territorial concession; the 2026 proposal is to build a "positive narrative" and place the boundary in an "appropriate position" in the relationship.⁵¹ If India wants peace, the argument runs, it should stop changing the status quo and negotiate on the 2005 basis.

A Chinese analyst would add two contested points often missing from Indian accounts: that the 2020 clash was triggered by Indian forward infrastructure on the Chinese side of the LAC, and that the 2024 deal was possible only because India accepted new border rules.⁵²

4.3 The strategic literature

Chinese strategic writing is not monolithic, but the spectrum is narrower than Western commentary assumes. No mainstream scholar or official in the sources reviewed publicly accepts Indian sovereignty over the territory China calls Zangnan; the debate is about how hard to press, what to trade, and whether settlement is desirable now. The axis is legalist rigidity versus managerial flexibility.⁵³

Five positions can be distinguished. The legalist-maximalist current, represented by the Foreign Ministry and Global Times, treats the claim as indivisible and the line as void; its function is to define the outer limit of acceptable debate.⁵⁴ The assertive-realist current, represented by Lin Minwang, argues that the pre-2020 rules were biased against China and gave India room to "nibble", and that 2020 and 2024 "reshaped the rules"; he supports negotiation and suggests an early harvest in Sikkim.⁵³⁵⁵ The pragmatic-managerial current, represented by Hu Shisheng, treats the boundary as subordinate to development, describing the relationship as "stabilising after the fall, slowly recovering, with competition and cooperation coexisting" and the opening as "floor-building, not a reset".⁵⁶ The accommodationist strand, represented by Shen Dingli, argues that an agreement on Zangnan would produce permanent peace — a position implying mutual concession and not official policy.⁵⁷ The nationalist-public strand, represented by Gao Zhikai, performs rhetorical maximalism; his proposed counter-line was, by his own account, a device "to use magic to defeat magic".⁵⁸

The most important feature of the spectrum is what it does not contain: in the sources reviewed, a serious Chinese argument that the area China calls Zangnan is legitimately Indian. The debate begins after the claim, not before it. A conference that stages a "hardline versus moderate" debate without recognising this will misrepresent the field.⁵³

4.4 Strategy: two levels

China's policy operates at two levels: an absolute public claim and a managerial operational practice. At the level of claim, the position is absolute and admits no public compromise. At the level of conduct, Beijing presents itself as managerial: the 2024 understanding is a package solution on patrolling that does not affect either side's sovereignty position; the SR process produced six points in 2024, ten in 2025 and eight in 2026; and the 2026 meeting added hotlines, meeting points and border trade.⁵⁹ Beijing describes the two levels as consistent; whether that is strategy, institutional habit or both cannot be settled from open sources, but it is stable and it survived 2020, and any negotiation has to work with it rather than around it. Lin Minwang, the Chinese analyst quoted most often on the 2024 deal, describes it as a remaking of the rules rather than a restoration, writing that "the old rules were unfavourable to China and left institutional loopholes for India's continuous 'nibbling'"; Hu Shisheng describes a shift "from crisis management to a new stage of border management", with the border entering a de facto "armed coexistence".⁵³⁵⁶

The two levels are connected by the 2005 agreement, the only text both sides accept that contemplates adjustment of claims. Article 3 commits them to "meaningful and mutually acceptable adjustments to their respective positions" and a package settlement; Beijing treats the text as the framework, but its public rhetoric rarely mentions "adjustment".³ That is the seam between legalist claim and managerial practice: the legalist says the claim is indivisible, the 2005 text says claims are adjustable in a package.⁵¹

The strategy that follows is patient and procedural. The 2025 Central Conference on Neighbouring Work defines the approach as good-neighbourliness and a shared future, with "properly managing contradictions and differences" alongside "consolidating strategic mutual trust".⁶⁰ The 2019 defence white paper states that China "never seeks hegemony, expansion or spheres of influence" and settles territorial disputes through negotiation, while recording that land border disputes have not been completely resolved.⁶¹ The vocabulary legitimates a development-first approach without implying concession on sovereignty. That time and asymmetry favour China is an inference, not a stated intention.⁵⁶

4.5 The Indian reply

An Indian analyst would press several weaknesses. None refutes the legal argument; most shift the burden to China to specify and verify.

China's own conduct engages the line it calls void: Zhou Enlai proposed in November 1959 that both armies withdraw "from the so-called McMahon line in the east", and the 1962 withdrawal was measured from it. A line with no legal existence cannot be a reference for disengagement without conceding practical relevance. Beijing's answer — that a military arrangement confers no recognition — is a real answer that any Indian argument must meet.

The capacity argument proves too much: if Tibet could not conclude treaties, China's claim to have governed through the Tibetan local government is weakened, since the same authority that allegedly could not bind China is the one through which China says it administered the area. "Never delimited" is asserted, not mapped: China has published no precise eastern-sector line, coordinates or map of its traditional customary line, so claims that India crossed the LAC and China did not are unfalsifiable.²⁸ The historical record south of the line is thin: pre-1950 Chinese central administration is largely indirect, while India has administered the territory continuously since the 1950s, with elections, statehood and local government; Indian analysts invoke acquiescence and estoppel.

The renaming campaign contradicts the language of peace. Six batches in nine years, timed to diplomatic moments and Indian moves, are not neutral administration but low-cost, reversible pressure that changes the symbolic status quo.²² The 2024 deal is described by Chinese analysts as "rewriting the rules", which is not the language of restoration: if the old rules were unfavourable and left loopholes, the new rules are a Chinese gain.⁵³ The 2005 text commits both sides to mutual adjustment, in tension with the public insistence that all of Zangnan is non-negotiable.³ "Normal patrol" is a conclusion, not a description: after the October 2026 ITBP allegations Beijing said it conducts "normal patrol and management activities" without denying that patrols had changed.⁶² And "Zangnan" is a claim, not a fact: the pinyin has not changed the underlying assertion.⁴⁹

4.6 Assessment

The two positions are structurally similar. Both say sovereignty is non-negotiable; both accept some negotiation over conduct; both accuse the other of changing the status quo. The difference is baseline and sequencing. India wants the pre-April 2020 position restored; China's working preference, inferred rather than stated, is that the post-2024 arrangements stand. India says border peace must precede normal relations; China says the boundary should be placed in an "appropriate position" within the relationship.⁵¹⁶ The fixed formulation records both readings for that reason.

Those differences, not the exchange of legal formulas, are the real agenda. They explain why the management track can advance while the claim hardens, and why a procedural agreement can be signed in the same season as a new batch of names. They also explain the limits of a Track-II conference: it cannot settle the legal question, but it can clarify the two baselines, test what "early harvest" might mean, and identify where the management regime is thinnest.

The dispute is not a misunderstanding that better communication would resolve. It is a genuine disagreement about which instrument creates obligation, which facts count, and whose administration is lawful.

5. The pressure record, 2020–2026

China's pressure on India since 2020 has been layered rather than singular. The instruments fall into four families — military presence, administrative claim-hardening, infrastructure and economic leverage — and run on different clocks: villages and dams over decades, renaming annually, incursions seasonally, licensing weekly. A single pressure index would obscure that asynchrony.

Table 1. Instruments, examples and evidentiary status

Only the military instrument has produced deaths, and those deaths were inflicted without firearms.³⁴ Beijing states an intent for the legal and economic instruments and is silent on the military ones; pressure is therefore an inference from timing, targeting and effect. Capability is better documented than effect: renaming has changed no ground but has produced repeated crises.³⁹

5.1 The western sector, 2020–2024

The first phase began in April–May 2020, when forces built up at Pangong Tso, Galwan, Depsang and Demchok.⁵ On 15–16 June the Galwan clash killed 20 Indian soldiers; Beijing acknowledged four dead only in February 2021.³⁴³⁵ More than 100,000 troops remained on each side in eastern Ladakh as of December 2024.⁵ Pressure did not wait for Ladakh to settle: about 200 Chinese troops crossed near Yangtse in Tawang in late 2021, an episode resting on a single outlet;⁶⁹ a clash there on 9 December 2022 caused minor injuries on both sides.⁷⁰ The October 2024 understanding covered Depsang and Demchok, with disengagement complete by December.⁵ The arrangement is narrow, reciprocal and observable in the only sense available: both governments reported disengagement on the same timetable. It is not verifiable in any agreed sense — the text is unpublished, it is sector-specific, and there is no inspection, data-exchange or incident-attribution provision. It remains the one durable gain of 2024, and it is durable only in the two locations it names; the 2026 eastern-sector face-offs are the evidence that it does not travel.

5.2 The eastern turn, 2025–2026

After October 2024 the live military file moved east. The fifth renaming batch, of 27 names, was published in May 2025.⁷¹ Construction of the Medog project began on 19 July 2025.⁶⁵ Hydrological data has been suspended since June 2023, following the expiry of the Brahmaputra memorandum; an RTI reply reported in May 2025 found that no data had been shared since 2022.⁶⁸ Rare-earth licensing began on 4 April 2025.⁹

The 2026 record is denser. China published a sixth batch on 10 April 2026, bringing the total to 112 since 2017.²² On 26 June the Nah Welfare Society named five sites in a memorandum to the Deputy Commissioner of Upper Subansiri, including Oying (Asaphila), and alleged roads, bridges and camps.⁷² A confrontation at Taksing on 25–26 July took about a week to calm.⁷² India published 27 standard names on 7 August, including Longju, Maja and Bisa.³⁹ The 36th WMCC meeting was held on 6 August, the 25th SR round on 25 August, and the first Eastern Sector corps-commander meeting at Vacha-Damai on 6 September.⁷ On 12 August the MEA called border incidents "most serious" and said they would reflect on the wider relationship.²⁰ On 22 October the ITBP Director-General confirmed face-offs "at several locations in Arunachal", with PLA tents about 6–7 km inside India's perception of the LAC near Maja.⁷

5.3 Grading the evidence

The 2026 episode mixed a real grievance file with a rumour ecosystem. This report therefore grades every claim.

Documented — official on record, or two independent credible sources: the Galwan fatalities and China's February 2021 acknowledgement; the October 2024 understanding; the renaming batches and totals; the dam's start; the data halt; rare-earth licensing; India's 27-name counter-list; and the ITBP statement of 22 October 2026.^{343552265689397}

Credibly reported — a reputable outlet, officials neither confirming nor denying: tents 6–7 km inside India's perception of the line; the Nah Welfare Society site list; the reported gaps in Indian patrolling at Taksing; the November 2025 Pudong detention; the October 2021 Tawang transgression; and the September 2020 detention of five Indian nationals near the Tibet border.^7726669

Rumoured, contested or debunked: claims of a 50–60 km PLA advance, rejected by the government's fact-checking unit; claims that China had taken control of Asaphila, Longju and Taksing, covered by the Army's denial and unverified by any source reviewed; the figure of 38–40 Chinese dead at Galwan, inconsistent with Beijing's acknowledgement of four; claims of organised militia settlement on the Indian side, for which no verifiable instance was found; and assertions that water is already being diverted, which no source reviewed supports.⁷²³⁵⁶⁸

5.4 What October 2024 changed

The thaw is differential. De-escalation occurred where both sides had something to trade, at Depsang and Demchok; claim-hardening continued where only China is setting facts, through villages, names, the dam and licences.⁵²²⁶⁵⁹ October 2024 therefore bought quiet in one sector and time in the other; 2026 is what the second looks like when the time is used. The Indian record is not uniformly comfortable: restrictions on grazing and hunting access are themselves part of the local grievance, and a framing that treats local communities purely as victims will not survive contact with the evidence.⁷²

6. Economic exposure and global spillover

6.1 The bilateral baseline

Two-way goods trade reached a record $155.62 billion in calendar 2025 on Chinese customs data: Chinese exports to India $135.87 billion, Indian exports $19.75 billion, deficit $116.12 billion.⁷³ On India's April–March basis the figures for FY2025–26 are $151.1 billion, $131.63 billion, $19.47 billion and $112.16 billion.¹⁰ The $4 billion difference between the series is a reporting artefact of valuation and classification, not a disagreement about direction.⁷⁴ The deficit rose from $73.31 billion in FY2021–22 to $112.16 billion in FY2025–26, and reached $105.93 billion in January–September 2026, up 20.7 per cent year on year.¹⁰⁷⁵

The relationship is asymmetric both ways: China supplied about 17.0 per cent of India's merchandise imports in FY2025–26, while India sent 4.4 per cent of its exports to China; China's exports to India were about 3.6 per cent of its global exports in 2025.¹⁰¹²⁷³

6.2 The chokepoint map

Dependence is concentrated. India imported 53,700 tonnes of rare-earth permanent magnets in FY2024–25, 93 per cent from China by one account and 84.8–90.4 per cent by quantity on official data; inventories fell to two to three weeks after the April 2025 controls.¹¹¹² It imported $4.35 billion of APIs and bulk drugs in FY2024–25, of which China supplied 73.7 per cent, with 41 critical molecules identified as single-source.⁷⁶ China supplied 83.6 per cent of lithium-ion battery imports worth $3.9 billion in FY2025–26, 59.6 per cent of solar imports in Q4 2025, and 48.9 per cent of semiconductor imports, down from about 64 per cent a year earlier.¹²⁷⁷ Across 71 eight-digit electronics tariff lines, China supplied at least 80 per cent of imports in FY2025–26, against 44 such lines in FY2018–19.¹²

The supplier's position is structural: China refines 19 of 20 critical minerals at an average share of about 70 per cent, and about 92.1 per cent of rare-earth separation, on 2025 estimates.⁷⁸ A supplier with that share is not a price-taker, and licensing decisions become strategic even when they are legally generic.

6.3 What past shocks teach

The Galwan precedent is the most important elasticity finding. Indian imports from China were $58.71 billion in calendar 2020, making China the largest source in the year of the clash; they rose to $97.5 billion in 2021 and two-way trade reached $125 billion, despite app bans and investment restrictions.⁷⁹⁸⁰ A lethal clash did not reduce trade: the imports are inputs, not finished goods.

The Red Sea and Hormuz episodes supply the price template. Suez trade fell about 50 per cent year on year in January–February 2024; UNCTAD simulated a 0.6 per cent rise in global consumer prices by end-2025, while CEPR put the trade cost at 1.3 percentage points in 2024.⁸¹⁸²⁸³ In 2026 Hormuz was effectively closed from late February; Brent peaked at $138 on 7 April against $71 in February, and India's crude bill for April–July rose 56.5 per cent year on year to $63.4 billion.⁸⁴⁸⁵ War-risk cover moved from 0.15–0.25 per cent to 7.5–10 per cent of hull value, and a single voyage from about $250,000 to $7.5–10 million.⁸⁶⁸⁵ Reinsurers cancelled Indian Ocean cover effective 16 August 2026; the Bharat Maritime Insurance Pool, with a ₹12,980 crore guarantee and $1.5 billion of capacity, restored availability but not pre-crisis prices.⁸⁶⁸⁷

6.4 Six transmission channels

A border war becomes a global price shock through six channels, each with a different latency. Insurance and shipping move first, through war-risk premia and Cape of Good Hope rerouting. Energy moves next, through the risk premium on crude and LNG. Critical inputs follow as inventories of magnets, cells and APIs exhaust within weeks. Semiconductors transmit through Taiwan, where about 90 per cent of advanced chip production is located and where TSMC held 73 per cent of the pure-play foundry market in Q2 2026.⁸⁸⁸⁹ Food and fertiliser transmit through LNG and ammonia feedstock, and finance through equity and exchange-rate adjustment. Each channel moved in 2024 or 2026; insurance moved first and hardest.⁸⁶

6.5 Three scenarios and their arithmetic

Common baseline, FY2025–26: nominal GDP $3,920 billion; imports from China $131.6 billion; oil import bill $123 billion at about $71 a barrel; current-account deficit $25.2 billion, or 0.6 per cent of GDP; reserves $734.6 billion as of 2 October 2026.⁹⁰¹⁰⁸⁴ Every output below is author-constructed, not an institutional forecast.

Scenario A — Contained. A clash confined to one sector for under 30 days; trade and diplomacy open; no export controls; a $5–10 oil premium for one quarter; insurance repricing for four to eight weeks.

• Oil: $5–10 a barrel × 1.8–2.0 billion barrels = $9–20 billion annualised. Over one quarter: $9–20 billion ÷ 4 = $2.3–5.0 billion.
• As a share of GDP: $2.3 billion ÷ $3,920 billion = 0.06 per cent; $5.0 billion ÷ $3,920 billion = 0.13 per cent.
• Net: GDP −0.1 to −0.3 percentage points; CPI +0.1 to +0.2 points; current-account deficit +0.1 to +0.2 points.

Scenario A's first-round cost is 0.06–0.13 per cent of GDP. The growth effect is larger because a one-quarter energy shock propagates through demand and the current account; we apply a 1.5–2× multiplier over two quarters. That is an assumption, not an estimate, and the range should be read accordingly. A is not the current base case: it is the pre-2025 template — a short clash in a market without export controls and without an energy shock — included because it isolates the border variable. The current baseline is A's political event on B's economic conditions, which is why the report does not offer a single "contained clash" cost for 2027: the honest range runs from A's figures to roughly half of B's, depending on whether the energy shock persists.

Scenario B — Prolonged. Twelve to twenty-four months of tit-for-tat export controls; restrictive licensing of rare earths, magnets, graphite and battery technology; Indian procurement preferences; oil at $95–110 a barrel; Cape rerouting; Hormuz open.

• Input prices: 10–25 per cent on $131.6 billion = $13.2–32.9 billion; ÷ $3,920 billion = 0.34–0.84 per cent of GDP.
• Energy: +$24–39 a barrel against $71, on 1.8–2.0 billion barrels = $43–78 billion; ÷ $3,920 billion = 1.1–2.0 per cent of GDP.
• First-round total: 0.34 + 1.1 = 1.4 per cent; 0.84 + 2.0 = 2.8 per cent.
• Net of a 20 per cent overlap and a 20 per cent offset from services, remittances, diversion and fiscal support: 1.4 × 0.8 × 0.8 = 0.9 per cent; 2.8 × 0.8 × 0.8 = 1.8 per cent.
• Net: GDP growth −0.5 to −1.0 percentage points; CPI +0.5 to +1.0 points; current-account deficit 1.5–2.0 per cent of GDP.

• Global: the WTO's high-energy-price scenario shaves 0.3 points off 2026 GDP and 0.5 points off merchandise trade.⁹¹

Scenario B's 0.9–1.8 per cent is a level loss. Spread over 12–24 months it is 0.5–1.0 percentage points of growth per year — the figure used above.

Scenario C — Systemic. Hormuz and Bab el-Mandeb constrained for six to twelve months; Brent at $130–140; China suspending 50 per cent of exports to India by value; Indian Ocean reinsurance withdrawn.

• Energy: $130–140 − $71 = +$59–69 a barrel × 1.8 billion barrels = $106–124 billion; ÷ $3,920 billion = 2.7–3.2 per cent of GDP.
• China cut: 50 per cent × $131.6 billion = $65.8 billion. With a gross-output multiplier of 1.5–2.0 and a value-added share of 35 per cent: 65.8 × 1.5 × 0.35 = $34.5 billion; 65.8 × 2.0 × 0.35 = $46.1 billion; ÷ $3,920 billion = 0.9–1.2 per cent of GDP.
• Combined gross, with partial overlap: 2.7 + 0.9 = 3.6 per cent; 3.2 + 1.2 = 4.4 per cent.
• Net of a 30 per cent offset: 3.6 × 0.70 = roughly 2.5 per cent; 4.4 × 0.70 = 3.1 per cent of GDP.
• External account: current-account deficit 3.5–4.5 per cent of GDP, or $137–176 billion; rupee depreciation of 10–15 per cent; reserve drawdown of $60–100 billion.⁹⁰
• Global: growth 0.5–1.5 points below the IMF's 3.1 per cent baseline for 2026, and inflation 1–2 points above its 4.7 per cent baseline.⁹² The IMF's severe fragmentation scenario implies a permanent loss of about 2.3 per cent of global GDP, rising to 7 per cent with slow adjustment — a ceiling, not a forecast.⁹²

The offsets (20 per cent in B, 30 per cent in C) are author assumptions standing in for services exports, remittances, trade diversion and fiscal support. A sensitivity run at 10 and 40 per cent moves the net figure by roughly ±0.4 percentage points in B and ±0.6 in C. The assumption is disclosed because it is the weakest link, not because it is precise.

B is the current trajectory, with export controls and an energy shock already in place; C requires a maritime closure and an embargo simultaneously, neither of which has occurred in the India–China relationship since 1962, though both occurred separately in 2025–26.⁸⁶⁹

6.6 Who pays: the global spillover

Losses fall unevenly. The Gulf and wider Middle East are already in contraction: the IMF projects −2.0 per cent GDP for the region in 2026 and Middle East merchandise exports down 17.2 per cent.⁹²⁹³ ASEAN, with $1.05 trillion of trade with China in 2025, is exposed through electronics and batteries, though diversion gains show in 24.7 per cent growth in January–July 2026.⁹⁴ The European Union ($828.1 billion) and the United States ($559.7 billion) face slower input-cost and asset-price losses, the latter through Taiwan semiconductors.⁷³⁸⁸ Japan, South Korea and Africa are exposed through energy, foundries and commodity demand.⁷³⁹⁵

6.7 Water

The Brahmaputra is the one channel that is bilateral and permanent rather than market-mediated. China began the Medog project on 19 July 2025: five cascading stations, about 60 GW of design capacity, roughly 300 TWh a year at design output, an estimated cost of $167.8 billion, with initial grid connection targeted for 2030 and full operation by 2033.⁹⁶⁹⁷ A Reuters report citing an Indian government analysis says the project could allow diversion of up to 40 billion cubic metres, just over a third of annual flow at a key border point; an Indian analysis puts the likely dry-season reduction at about 10 per cent.⁹⁸⁹⁷ These are potential-impact estimates, not observed flows.

7. Deterrence, escalation and the nuclear shadow

7.1 The conventional balance

The Line of Actual Control is not a surveyed boundary; the two sides hold different perceptions of its alignment across roughly 3,380 km — estimates range to about 3,490 km — in three disconnected sectors.⁹⁹ India's Ministry of Defence recorded that PLA deployment along the northern borders fell during 2025 but that ten combined-arms brigade-sized formations remained.⁶³ Indian reporting has estimated 50,000–60,000 Chinese troops in Ladakh alone as of 2024, an estimate rather than an official figure.⁹⁹ India's ground formations facing Arunachal Pradesh are 3 Corps at Rangapahar and 4 Corps at Tezpur.¹⁰⁰

The aggregate gap is wide. SIPRI estimates China's military expenditure at $336 billion in calendar 2025, up 7.4 per cent in real terms, against India's $92.1 billion, up 8.9 per cent and equal to 2.3 per cent of GDP.¹⁰¹ China's 2026 core defence budget is estimated at RMB 1.935 trillion, or about $273 billion.¹⁰² India allocated ₹7.85 lakh crore in FY2026–27, including ₹2.19 lakh crore of capital expenditure.¹⁰³ The totals are not directly comparable: India's figure is a budget commitment, while SIPRI measures calendar-year spending at market rates.¹⁰¹¹⁰³

7.2 The escalation ladder

The ladder below is an analytical construct built from documented incidents, not a prediction. Each rung names the threshold that separates it from the one below.

The 1996 agreement is the formal threshold at rung 4: neither side shall open fire or conduct blast operations within two kilometres of the line, and combat aircraft are barred within ten kilometres.²³ That is why Galwan was fought with clubs and fists. The more dangerous transition is rung 5 to 6, where attacks on airbases, command posts or SSBN bases become indistinguishable from a counterforce threat.

7.3 Why limited war is hard to keep limited

Five mechanisms make containment difficult, and none depends on leaders being reckless. Because the line is not surveyed, a local move can be described by each side as defensive; the Taksing episode of 2026 is a case in point, with no public agreed account of the ground.¹⁰⁰ The sectors are disconnected, so a crisis in one does not communicate to the others. India plans against a two-front contingency, and the May 2025 conflict exposed how Pakistan's Chinese-origin systems interact with an India–China crisis.⁹⁹¹⁰⁵ Both sides have publicly committed to positions from which retreat is costly, and nuclear opacity forces each side to guess.

The May 2025 India–Pakistan crisis is the best-documented regional test. It involved Indian cruise-missile strikes, Pakistani short-range ballistic missiles and drone attacks, yet no manned aircraft crossed into the other's airspace, and both sides calibrated escalation.¹⁰⁵ Control therefore depends on shared tacit rules, functioning communication and a mutual interest in stopping. None can be assumed in a China–India crisis.

7.4 The nuclear dimension

The nuclear balance is asymmetric and widening. SIPRI estimates China at about 620 warheads and India at about 190 as of January 2026, with China expanding faster than any other state.¹⁴¹⁰⁶ Both sides declare no first use. China reaffirmed the principle in its white paper of 27 November 2025.¹⁵ India's Cabinet Committee on Security statement of 4 January 2003 committed India to no first use and to "massive retaliation" against a nuclear attack on Indian territory or Indian forces anywhere.¹⁰⁷ The consensus is under stress: SIPRI assesses that China may be moving toward an early-warning counterstrike posture, while Indian statements since 2019 have described no first use as conditional in practice.¹⁴¹⁰⁸

India's triad became more credible in 2025–26. INS Arighaat was commissioned in August 2024 and INS Aridhaman on 3 April 2026, giving India three hulls in principle.¹⁰⁹¹¹⁰ The submarines carry the K-15 at about 750 km and are transitioning to the K-4 at about 3,500 km; the Agni-5, above 5,000 km, was flight-tested with multiple warheads in 2024.¹¹⁰ SIPRI assesses that India may have begun deploying a small number of warheads on a single SSBN, and that canisterisation and sea patrols suggest a possible shift toward peacetime mating.¹⁴ That is an assessment, not a confirmed change, but it erodes the separation that has underwritten crisis stability.

Two functions follow. Nuclear posture deters escalation to strategic strikes, and it sets a ceiling on conventional punishment. The second is more fragile, because conventional disadvantage creates an incentive to signal resolve earlier than doctrine implies.

7.5 The maritime and partner dimension

China's Indian Ocean presence rests on access rather than bases: Djibouti and port access at Gwadar, Hambantota, Bangladesh and Myanmar.¹¹¹ By late 2025 Chinese survey vessels were mapping undersea terrain; the Observer Research Foundation estimates that about 80 per cent of the 64 Chinese research and survey vessels active in the region have presumed PLA links.¹¹²

Partners change the cost calculus but not the core dyad. India and the United States have LEMOA and COMCASA in place, with defence trade above $20 billion by 2020, but no mutual-defence treaty.¹¹³ France contracted 26 Rafale-Marine aircraft in April 2025; Russia supplies the $5.43 billion S-400 system; India approved about $25 billion of procurement in March 2026.^114115116 None of these constitutes a guarantee in a China contingency.

7.6 Defining stability

A realistic threshold for 2027 is not a boundary settlement. It is a year in which the Working Mechanism meets at least twice with a joint readout; the Special Representatives process continues; the Eastern Sector corps-commander channel meets again without a prior incident; border trade at all three passes operates continuously, following the reopening of Nathu La on 1 August 2026; no firearms are discharged within two kilometres of the line; both sides reaffirm no first use publicly; and hydrological data sharing resumes.^{63100104231568}

8. Scenarios, 2027–2032

8.1 Method

Scenarios are not forecasts but disciplined descriptions of pathways, defined by observable assumptions, indicators and implications. No probabilities are assigned, and no point forecasts are made. The restraint is evidential: Chinese intentions are not established in open sources, and Indian and Chinese readings of the October 2026 patrol reports diverge.⁶²

Three drivers organise the set. The first is implementation: whether the August 2026 eight-point outcome becomes terms of reference, meeting-point locations, hotline activation and an incident procedure.⁶ The second is political direction: whether leaders sustain the process through the 26th Special Representatives round, due in India in 2027.¹¹⁷

The third driver is domestic politics: India's general election due by 2029, and the Chinese party congress of 2027. A border process that depends on two leaders' willingness to absorb domestic criticism is exposed to any change in either capital, and the 2024–26 record shows that the process advances fastest in the months around summits.

8.2 Baseline: managed rivalry

Assumptions. The 26th round is held on schedule; both technical groups agree terms of reference; the meeting points and hotlines are activated; no fatal clash occurs; border trade and pilgrimage continue, and the river-data mechanism meets; neither side invites a third party. Each element is plausible on the 2026 record: the eight points make terms of reference the first task of both groups, and the two governments released identical texts.⁶⁵¹

Indicators. Publication of the terms of reference; announcement of the meeting-point locations, still undeclared in late August 2026;¹¹⁸ an activation date and test for each hotline; the number of meetings held on schedule; the duration of any face-off; and whether either side publishes another naming batch.

Implications. The border remains calm without becoming settled. De-escalation is deferred, as it has been since October 2024.⁴⁰ Construction continues on both sides. The economic relationship widens rather than deepens: the goods deficit reached $112.16 billion in 2025–26.¹⁰ The strategic risk is not war but ossification — a quiet, militarised frontier that is permanently contested, with the domestic cost of concession rising on both sides.

8.3 Limited clash: reversal

Assumptions. A violent incident in the eastern sector causes fatalities or serious injuries; domestic pressure prevents de-escalation; both sides reinforce locally but neither seeks a general war; trade continues; third parties issue statements of concern without offering mediation. The October 2026 allegations describe how such an incident could begin: a patrol encounter where the two sides' perceptions of the line differ by several kilometres.⁶²

Indicators. Casualty reports and the official characterisation; suspension of scheduled meetings; non-use of the hotlines; reinforcement and airlift activity; war-risk premia for Indian Ocean cargo; the language of the two foreign ministries; and the suspension of Track-II contacts.

Implications. The episode resembles 2020 with a thinner cushion. The 2020 standoff exposed the architecture's central defect: no inspection, no data exchange and no consequence for violation.¹¹⁹ A clash in 2027 would follow two years of declared normalisation, making the political damage greater than the military one.

8.4 Prolonged confrontation: attrition

Assumptions. No settlement and no de-escalation; repeated standoffs without fatalities; tit-for-tat export controls; cyclical place-naming; both sides build infrastructure and revise domestic law; neither accepts a third-party role. The assumption is not hostility but inertia: the 2005 agreement requires a package covering all sectors and sets no timetable, so inaction is the default.³

Indicators. The frequency of corps-commander meetings; defence budget growth; the time taken to license magnet exports; the number of naming batches; the bilateral deficit; and Chinese investment approvals under the eased Press Note 3 regime.¹²⁰

Implications. This is the costliest peaceful scenario. Dependence is concentrated in inputs India cannot replace quickly: China supplies 85–93 per cent of its magnets depending on the measure.¹²¹ Attrition does not stop trade; it raises its price and turns industrial policy into a security instrument. The risk is that managed rivalry becomes a managed decline, with the boundary no closer to settlement in 2032 than in 2026.

8.5 Settlement track: conditional progress

Assumptions. Leaders authorise a bounded negotiation; the terms of reference state that no sectoral outcome prejudices the final package; an early harvest is attempted in a sector where the two perceptions of the LAC already coincide; a Chinese analyst has suggested Sikkim, a suggestion the report records but does not adopt;⁵³ a verification pilot accompanies it; de-escalation begins with a published timetable.

Indicators. The text of the terms of reference, particularly the non-prejudice clause; the naming of a sector; a joint survey or agreed coordinate list; published incident statistics; border trade beyond the three passes reopened in August 2026; and any official Indian statement that describes Aksai Chin as disputed rather than occupied.¹²²

Implications. A settlement track would not settle Arunachal Pradesh, and this report does not suggest otherwise. Its value is procedural: it would create a technical record, test whether mutual adjustment can operate, and establish that a sectoral outcome does not prejudice the package. It requires the most political capital and yields the least immediate domestic reward.

A package that leaves the eastern sector untouched can only be built in the west. The report does not recommend a western-sector trade; that is a decision for governments and would be premature in a Track-II text. But the 2005 text's "mutual adjustment" has no operational meaning unless the western sector is eventually on the table, and India's lexical restraint on Aksai Chin — its phrase is "illegal occupation of 38,000 sq. km", never "integral and inalienable" — is the space in which such a trade would be discussed. A settlement track that cannot name that space is a settlement track in name only.

8.6 Reading the set

None of the four scenarios produces a boundary settlement by 2032. Two are continuations of the current trajectory (baseline and attrition), one is a rupture within it (limited clash), and one is a departure from it (settlement track). The agenda in Section 9 is designed to be useful in all four, which is why it is procedural rather than political. The most informative warning indicator is procedural, not military: non-publication of the terms of reference, or non-activation of the hotlines, would show that the August 2026 text was declaratory rather than operative.⁶¹¹⁸

9. An eight-year agenda to 2034

9.1 Selection rule

Fourteen recommendations follow. Each is procedural, reversible and silent on sovereignty, assigned to an actor, phased, costed in administrative or fiscal terms and labelled with a precedent or "no precedent". Phases run from consolidation in 2027–2028 to verification in 2029–2031 and institutionalisation in 2032–2034.

The governing constraint is the 2005 agreement: settlement must be a package covering all sectors, and pending settlement both sides respect the line.³ The opportunity is the August 2026 outcome, which committed the two technical groups created at the 24th round to agree terms of reference as their first task.⁶

Three of the fourteen — 5, 6 and 11 — cannot be implemented without a change in China's position on verification, on the joint ascertainability of the line, or on the sovereignty-neutrality of export licensing. They are included because a Track-II agenda should state the ask precisely rather than settle for the lowest common denominator. Each carries a fallback that does not require that change.

9.2 Bilateral

Recommendation 1. Publish the terms of reference of the Expert Group on Boundary Delimitation and the Working Group on Border Management, with a non-prejudice clause. Actor: MEA East Asia Division, with China's MFA. Sequence: before the 26th SR round, 2027. Cost: negligible. Precedent: Article III of the 2005 agreement requires a package settlement; the 2026 outcome makes terms of reference the first task of both groups.³⁶

Recommendation 2. Publish activation terms for the two additional meeting points and two hotlines: location, staffing, counterpart and test schedule. Actor: Ministry of Defence and Army Headquarters. Sequence: 2027–2028. Cost: low; staffing within existing formations. Precedent: the sixth hotline was announced in August 2021 with its location and counterpart.¹²³

Recommendation 3. Restore an annual calendar of working mechanism, SR and corps-commander meetings, with dates published in advance. Actor: MEA. Sequence: from 2027. Cost: low. Precedent: the working mechanism had met 35 times by May 2026 and the SRs 25 times by August 2026; the calendar lapsed after 2020.³⁷¹¹⁷

Recommendation 4. Establish a director-general-level military hotline between army headquarters, without the DGMO title if that assists agreement. Actor: Ministry of Defence. Sequence: 2028–2029. Cost: low. Precedent: the India–Pakistan DGMO hotline restored the February 2021 ceasefire; a top-level hotline with China was proposed in 2013, agreed in 2015 and never established.¹²⁴¹²³

Recommendation 5. Adopt a written no-firearms and no-tailing protocol with a two-kilometre separation rule at named face-off sites, annexed to maps. Actor: Ministry of Defence and Army Headquarters. Sequence: 2029–2030. Cost: medium; survey and map annexes. Precedent: the 1996 agreement barred firing within two kilometres and combat aircraft within ten kilometres; Korea's armistice used a two-kilometre withdrawal.²³¹²⁵ Fallback: if maps and toponyms cannot be agreed, the protocol can be recorded in descriptive terms at three named locations, with a toponym annex marked "not prejudicial to either side's position" — the device the 1993 agreement already uses for the LAC. The rule also risks freezing Indian patrol access, which is a reason to test it in one sector first.

Recommendation 6. Create a joint incident-review cell with an agreed evidence standard, private for two years and publishing quarterly statistics thereafter. Actor: MEA and MFA, with military liaison officers. Sequence: pilot in one sector, 2029; standing cell by 2032. Cost: medium; eight to ten staff. Precedent: joint review exists in the Korean Armistice Commission and the Sudan–South Sudan mechanism; published statistics have no precedent.¹²⁵¹²⁶ Fallback: if Beijing will not accept joint attribution, parallel national logging to a published common template, with an annual exchange of aggregate incident counts and no findings, is weaker than a joint cell and stronger than nothing, and it is the most China has accepted in any border forum.

Recommendation 7. Begin joint clarification of the line in one stretch where the two perceptions of the LAC already coincide, and record the result in coordinates without boundary language. The Arunachal sector is excluded, in line with India's position that its boundary is settled. Actor: MEA Boundary and Ocean Affairs, with the Survey of India. Sequence: 2029–2031. Cost: high; cartographic and domestic political. Precedent: the 1993 agreement provided for joint checking; the 2026 outcome commits both sides to improve understanding in appropriate areas.²⁶ The exclusion has a cost: the sector where the 2026 incidents occurred will remain undefined, and a management regime that cannot say where the line runs cannot verify who crossed it. Excluding the sector is a political choice, not a technical one.

Recommendation 8. Commission an Indian technical assessment of imagery-verification methods — satellite revisit rates, coordinate conventions and chain-of-custody for incident evidence — published as an Indian methodology paper and offered to the 26th round as a national contribution. No joint text, no Quad branding, no partner role in the border question. Actor: MEA with the Survey of India and a research partner. Sequence: 2028. Cost: low. Precedent: the 1961 Report of the Officials was a national technical product offered into a bilateral process.³⁰

9.3 Minilateral

Recommendation 9. Open an India–European Union supply-chain resilience dialogue alongside the free trade agreement expected to operate from early 2027. Actor: Ministry of Commerce and Industry. Sequence: 2027. Cost: low. Precedent: the EU's shift from decoupling to de-risking, and the agreement itself.¹²⁷¹²⁸

9.4 Economic

Recommendation 10. Fund a second tranche of rare-earth magnet capacity, adding processing and recycling to the 6,000-tonne scheme approved in November 2025. Actor: Ministry of Heavy Industries. Sequence: 2028–2030. Cost: high; the approved scheme is ₹7,280 crore.¹²⁹ Precedent: the November 2025 scheme.

Recommendation 11. Propose a bilateral consultation channel on export controls, with a thirty-day notification window before licensing changes affecting critical inputs. Actor: DGFT with China's Ministry of Commerce. Sequence: 2028. Cost: low. Precedent: none between India and China. Fallback: present the thirty-day window as a reciprocal commitment covering both sides' export-control measures, so that India's own procurement preferences and licensing restrictions sit inside the same transparency envelope. As drafted, the recommendation asks China to accept a constraint India does not accept.

Recommendation 12. Establish a published annual border-trade calendar and fund customs, storage and road improvements at the three passes reopened in August 2026. Actor: Ministry of Commerce with the governments of Sikkim, Uttarakhand and Himachal Pradesh. Sequence: 2027–2028. Cost: low; trade peaked at $9.5 million in 2016, so the case rests on livelihoods, not revenue.¹²² Precedent: the 1991–92 arrangements and the 2026 reopening.

9.5 Domestic

Recommendation 13. Publish an annual border-transparency report on Indian infrastructure, village programme delivery, patrol access and incident counts, without disclosing unit locations. Actor: MEA with the Ministry of Defence. Sequence: annually from 2027. Cost: low. Precedent: none; the Indian record appears in statements to Parliament, not a periodic report.⁴⁴

9.6 Track-II

Recommendation 14. Fund a standing Track-II secretariat, jointly convened by the Institute of Chinese Studies and a Chinese partner, to maintain a technical annex and glossary under the Chatham House Rule. Actor: MEA public diplomacy, with Indian foundations. Sequence: from 2027; review in 2030. Cost: low; three to five staff. Precedent: the Manesar dialogue of March 2019; a published glossary has no precedent.¹³⁰

9.7 What the agenda does not propose

The agenda does not propose third-party mediation, arbitration or adjudication; a United Nations force on the line; a demilitarised buffer; a territorial swap; a referendum; economic sanctions; a dated roadmap; or a joint communiqué. China treats the boundary as bilateral, a buffer would freeze post-2020 positions, and a Track-II meeting cannot bind governments.⁵¹³⁷

10. Conclusion

The boundary dispute is not a problem to be solved within a decade. It is a condition to be managed while the legal positions, the balance of power and the domestic politics that sustain them change. The choice before 2032 is between a managed rivalry that preserves the possibility of settlement and a drift that consumes the machinery built in 2024–2026.

Three tests will show within two years whether the process is real: terms of reference with a non-prejudice clause; activation of the meeting points and hotlines; and an incident-review function that can establish what happened without public blame. If all three fail, the August 2026 text will be remembered as a declaration, not an agreement.

India's advantage lies less in material balance than in institutions: a settled legal position, a professional army and a research community capable of technical work governments cannot do in public. None of the recommendations requires China's consent to be studied or published; several require it to be implemented. The Uttara Dialogue cannot settle the boundary. It can make the next negotiation better prepared, and argue that verification, not declaration, is the currency of stability.

Annex A. Chronology, 1914–2027

Annex B. Agreements and mechanisms

Annex C. Economic data

Figures are sourced in Sections 5 to 7; agreement texts are cited in Section 2.