Col. Nestor D’Souza, Research Fellow at the Manohar Parrikar Institute for Defence Studies and Analyses (MP-IDSA) spoke on “The Role of Artificial Intelligence in China’s Military Modernisation: Implications for India” at the Monday Morning Meeting held on 20 July 2026. Dr M.S. Prathibha, Associate Fellow, MP-IDSA, moderated the discussion. The meeting was attended by the Director General, Ambassador Sujan R. Chinoy and scholars from the Institute.

Executive Summary

The Speaker examined the footprint of Artificial Intelligence (AI) across the theory, practice and means of the military modernisation of the People’s Liberation Army (PLA), arguing that its influence is visible at every stage from the framing of Military Strategic Guidelines to the development of operational concepts, force structuring and capability acquisition and that the relationship is iterative rather than linear. He highlighted that intelligentisation, notwithstanding its prominence in Chinese discourse, has not yet displaced informatisation as the Military Strategic Guideline, and that at present it is being used to enhance informatised warfighting rather than to replace it. Tracing the concepts of systems-of-systems warfare, multi-domain integrated joint operations, multi-domain precision warfare and kill webs, the Speaker assessed that full mission autonomy remains a distant prospect, constrained as much by the Party’s reluctance to cede control as by the maturity of technology. The presentation concluded with implications for India, including operating under persistent surveillance, the risk of a widening and potentially generational capability gap, and the case for indigenous and asymmetric capability development.

Detailed Report

Dr. M.S. Prathibha, the moderator, set the context by situating the discussion within the intensifying technological competition between the United States and China, noting that announcements as recent as June 2026 of investments of the order of US$ 260 billion in data-centre strategies underline that the race between the two most consequential players in the international system is firmly underway. While such investment is largely commercial in character, its military significance is considerable. She observed that the United States continues to lead in the development of frontier AI models, and that several debates are underway within the PLA on how AI will reshape warfare. Chinese writings, she noted, underscore that science and technology constitute a link that cannot be bypassed in building a world-class military, with improvements to radar systems among the earliest concrete claims of AI-derived military benefit. Chinese discourse, she added, identifies three defining features of future intelligentised warfare: machines as principal combatants, the interconnection of all things, and algorithmic dominance.

The Speaker, Col. Nestor D’Souza, opened with two caveats. Since military modernisation is a classified subject in any country, and particularly so in China, the arguments advanced were largely inferential, drawn from Chinese publications and from Western and Japanese scholarship that follows the PLA closely; the presentation therefore dealt in the realm of possibilities rather than certainties. Second, it confined itself deliberately to the military dimension of modernisation, setting aside the political drivers of reforms such as theaterisation, on the premise that such decisions are taken at a level above the military leadership. The presentation was structured around the strategic bridge of ends, ways and means. Theory, the Speaker argued, is drawn from a nation’s values and historical experience- a particularly significant consideration in the case of a civilisational state such as China -which shape national interests and, in turn, national objectives or ends. These inform national policy and strategy across the diplomatic, informational, military and economic instruments, and within the military instrument produce the Military Strategic Guidelines, which shape doctrine, operational concepts, the design of operations and force structures. The means follow: weapon systems, support systems, logistics and training.

Locating the ends, the Speaker noted that China’s national strategy is framed by the China Dream of great national rejuvenation and the restoration of China to what it regards as its rightful place in the comity of nations after a century of humiliation. From this flows the grand strategy of building the PLA into a world-class military by 2049, with the tasks of protecting economic growth and territorial sovereignty and, above all, of being prepared to fight and win wars, pursued under the guiding strategy of active defence, in which the qualifier “active” carries a distinctly offensive connotation.

Tracing the evolution of the Military Strategic Guidelines from People’s War in 1949 through People’s War under Modern Conditions, Local War under Modern Conditions and Local War under High-Technology Conditions in the 1980s, to Informatisation in 2014, the Speaker observed that the early emphasis on protracted guerrilla warfare became untenable as industrial and agricultural capacity improved and the requirement shifted to halting an adversary at or before the border, with the 1980s bringing a conscious move away from a peasant-army self-image towards mechanisation. The 1991 Gulf War proved decisive, as the PLA leadership was deeply impressed by the coalition’s use of information to link sensors and shooters into network-centric operations. Each of these stages, he emphasised, was defined by human analysis; intelligentisation, which entered the discourse in the second decade of this century, envisages machine analysis substituting for or augmenting it – less creative, perhaps, but more accurate, more resilient and immune to fatigue. Crucially, however, intelligentisation does not yet enjoy the status of a Military Strategic Guideline. Citing the PLA scholar Li Minghai, the Speaker summarised the logic in a single line – that the dominant means of production shapes the Military Strategic Guideline – and noted that because the field is still evolving, informatisation continues to be the guideline in force.

Explaining that a Military Strategic Guideline in PLA construct answers six questions, the Speaker identified the opponent as the reigning hegemon, the United States; the primary strategic direction as having shifted from a ground-centric to a maritime orientation; the basis of preparation as systems-of-systems warfare; the basic form of operations as Multi-Domain Integrated Joint Operations, as designated in the 2020 edition of The Science of Military Strategy; the style of fighting as Multi-Domain Precision Warfare; and the strategic guiding thought as securing military advantage through information superiority. Informatisation, he noted, rests on three pillars – systems warfare, integrated joint operations, and cross-domain coordination extending beyond land, sea, air and space to the electromagnetic spectrum, cyber and the cognitive domain- each of which intelligentisation seeks to enhance through autonomous hardware, human–machine teaming and machine-speed decision support.

Turning to systems-of-systems warfare, the Speaker noted that discussion of information systems-based systems-of-systems operations began around 2005, during the Hu Jintao era. Integrated joint operations had been spoken of earlier but could not be operationalised, largely because the PLA’s older structure of general departments and military regions, dominated by the ground forces, made it unworkable; the reforms and theaterisation of 2015–16 created joint structures and, with them, the space for such concepts to succeed. The premise of systems warfare is that an integrated aggregation of systems should project greater power than the arithmetic sum of its parts, its aim being to strike the adversary’s vulnerabilities while protecting one’s own, and its objective being paralysis rather than annihilation – leaving the enemy’s faculties intact but preventing him from applying them coherently. The operational system comprises five principal components – firepower, ISR, information networks, command and control, and support – each disaggregating into successive layers of subsystems, so that the military is viewed not as army, navy and air force, nor even by domain, but as a set of systems defined by function. The adversary is correspondingly viewed as a target set of four kinds: information links and nodes, whose destruction produces paralysis; the five elements of his operational system; his operational structures and organisations; and, once these have been struck, his operational momentum, that is, the speed at which he observes and decides.

Executing systems warfare, the Speaker explained, requires subordinate concepts that tell forces how to interact. Multi-Domain Integrated Joint Operations, designated the basic form of operations in 2020 and formalised through a joint operations outline, or gangyao, issued in November 2020, is an upgrade on the 1999 formulation, whose structures could not effectively integrate the fruits of cyber, space and cognitive-domain operations into warfighting. In 2021 the PLA articulated Multi-Domain Precision Warfare as its new core operational concept, drawing on well-established Chinese structures for extracting lessons from contemporary conflicts, in which non-contact precision warfare had become increasingly prominent. The Speaker stressed why such terminological precision matters: once a concept is specified in this manner it enters planning, and force structuring organises itself around it, whereas concepts that remain nebulous are not implemented. AI-enabled capabilities are envisaged to enhance the networks required for precision strike, executed through kill webs – the practical manifestation of the entire chain of concepts. A kill web is modular, flexible and mission-oriented; such structures do not exist in peacetime, and it is AI and machine-based decision support that permit one to be assembled in preparation for, or during, an operation, aggregating capabilities across domains at machine speed, with hybrid intelligence determining the attachment and detachment of constituent systems as the target picture changes.

Turning to means, the Speaker outlined the broad heads under which PLA combat capabilities are being developed – autonomous vehicles, autonomous weapons and intelligent munitions; ISR; command, control and communications; decision support; sensor fusion and targeting; and information and electronic warfare, which together constitute C5ISR&T – to which are added predictive maintenance and logistics, simulation and training, and cognitive domain operations, an upgrade on the earlier Three Warfares that seeks to use algorithms to design cognitive effects. These heads, he cautioned, were separated only for clarity; to imagine them working in silos would defeat the very concept of systems warfare, since an unmanned aerial vehicle is simultaneously a weapon and an ISR platform.

Summarising his inferences, the Speaker observed that the footprint of AI is visible at every stage of military modernisation and that the relationship is iterative, with AI shaping concepts and concepts in turn directing AI research. At present, intelligentisation is powering informatised warfare rather than replacing it: it is proceeding at the level of tactical equipment, where ethical questions do not arise and fully autonomous systems are therefore envisaged, but technology has not matured to a point where the killing of a human being can be delegated to it higher up the hierarchy. Full mission autonomy, he assessed, is not imminent, since the PLA does not consider that it has achieved the requisite levels of informatisation or jointness, its joint structures are still being consolidated, and its operational commanders are not assessed as sufficiently competent to execute joint operations. More fundamentally, the ceding of control runs against the grain of the Party, which is unwilling to cede authority even to its own generals, making the surrender of control to machines a distant prospect. Consequently, lower-level systems are being intelligentised wherever this does not threaten the PLA’s way of functioning, while supervisory systems are simply becoming better informatised.

In his concluding remarks, the Speaker identified the implications for the Indian armed forces. India will have to operate in an environment of persistent technical and cognitive surveillance, with data being gathered in peacetime to feed databases and algorithms that seek to predict Indian responses, the more dangerous prospect being the sharing of such information with inimical neighbours; there is consequently considerable scope for improving efforts at information denial and deception. Any adversary that persists with domain-oriented or siloed approaches may be in for a surprise, since systems warfare permits considerable flexibility in the application of power. The machine-versus-human contest is already a reality, as videos of FPV drones pursuing soldiers demonstrate, and extends beyond hardware such as robotic platforms and humanoids to decision-making itself, where creativity may remain a human advantage but speed and resilience will favour the machine. A growing number of Indian observers assess that the qualitative and quantitative gap is widening and must be mitigated before it becomes generational. Finally, the approach to imbibing technology into warfighting should preferably be unique and indigenous, so that India retains the element of surprise in seeking military advantage.

Q & A Session

Ambassador Sujan R. Chinoy, Director General, MP-IDSA, complimented the Speaker on the choice of a subject of considerable relevance to India’s security. AI, he observed, is a field in which even marking time means falling behind. Roughly 85 per cent of global work in the field is attributable to the United States and about 15 per cent to the People’s Republic of China, with the rest accounting for very little, which raises the central question of where that leaves India.

Pointing to the World Artificial Intelligence Conference held in Shanghai from 17 to 20 July 2026, the Director General drew attention to the establishment of the World Artificial Intelligence Cooperation Organisation (WAICO), the agreement for which was signed by 29 countries on 16 July 2026, on the eve of the Conference. This is not China’s first such initiative: the Asia Pacific Space Cooperation Organisation similarly competes with the Japan-led Asia-Pacific Regional Space Agency Forum, of which India is a member while remaining outside the Chinese platform. The WAICO signatories are drawn from Asia, Africa, Europe and elsewhere, and include Russia; no major Western state is among them, but several countries of consequence to India in its neighbourhood, notably Pakistan, and in Southeast Asia, are closely aligned with China and have joined. India, by contrast, is part of the US-led Pax Silica initiative on semiconductors and allied critical technologies.

Beyond the question of funding India’s own research, the Director General asked what India’s plugging points globally are for keeping abreast of progress elsewhere. India has been active in the AI summit process, having hosted one summit with another hosted by France, and has argued at the level of the Prime Minister that AI should be open and accessible, that data should be protected, that ethical considerations should be observed and that the interests of the Global South should be safeguarded. This position, he noted, is not far removed from what was articulated at the Shanghai Conference – including the striking assertion, which drew headlines, that man should not cede control to the machine, an ethical formulation that bears directly on the future of lethal autonomous weapon systems. Whether this reflects an appeal to the Global South, or a realistic Chinese assessment that others are ahead and an attempt to make a virtue of necessity, remains an open question. What is clear is that AI governance has not yet been put together and that the contest will continue.

Turning to implications, the Director General noted that Indian platforms will increasingly depend on AI and on intelligentised warfare systems, raising questions of system integrity and protection against intrusion, from simple systems being compromised to larger ones. His central question was whether India needs to take a fresh look at whom it cooperates with, and whether it should be plurilateral across all the emerging contending camps rather than committing to a single grouping in a fractured landscape – since no one grouping may meet all of India’s requirements in the future, and capabilities may yet be denied.

Dr. M.S. Prathibha, the moderator, observed that international assessments of AI power increasingly turn on the ability to build AI models, an area in which the United States leads on the strength of its data centres, which explains China’s data-centre strategy and investments reported at around US$ 295 billion. She also noted that Chinese analysts regard new-energy and hypersonic weapons as principal threats to intelligentised systems, and concluded the session by observing that Chinese thinking appears to envisage command shifting between the centre and forward troops depending on the state of battlefield awareness, rather than a straightforward move towards centralisation or decentralisation.

A participant raised the question of India’s own AI and digital foundations, noting that policy focus has rested on subsidies and on encouraging research, yet a majority of the GPU capacity available in the country belongs to foreign providers rather than Indian entities, and argued that sovereign AI must be examined in terms of investment, intellectual property, and whether Indian models are built from the ground up or layered on open-source foundations. He further asked about the role of the PLA and its research institutions in China’s AI development, the vulnerabilities that arise if Indian scholars and innovators build on Chinese models offered to the world, and the risk of embedded components or hidden functions proliferating into government and military systems. Another participant offered three illustrations of how Chinese effort has translated into implementable capability over roughly two decades: the prominence given in the official Chinese press to the death of a PLA Colonel in a road accident in July 2023, a graduate of the National University of Defence Technology who had studied statistics abroad before returning to work on an AI decision-support system, now in its third iteration, which reportedly generates draft plans, conducts risk assessment and provides contingency options on the basis of incomplete data; a 2023 study finding that of 340 articles on military theory in a year’s output of the PLA’s official newspaper, 132 dealt with AI, concentrated in decision-making, unmanned systems, cognitive warfare and situational awareness; and displays at the Zhuhai air show in which a Chinese defence major demonstrated fused sensors tracking a target, passing instructions to shooters and conducting post-strike damage assessment. Further interventions related to the realignment of Indian manpower and training policies for the future battlefield – a ten-minute decision cycle being of little value if reporting takes twenty minutes – the retention of technically skilled personnel in service, a clarification on whether informatisation is primarily about disrupting an adversary’s decision-making or about sharing information within one’s own system, and a question on how far these Chinese concepts were visible in Pakistani operations during Operation Sindoor.

Responding, the Speaker agreed that dependence on foreign-origin technology remains a vulnerability across critical technologies, not AI alone, as experience shows such dependencies are leveraged in times of conflict, and that this can be addressed only through foundational research and capacity. On the PLA’s role in AI development, he pointed to military-civil fusion as the germane framework, a subject deliberately left out of the presentation as it is driven by the national leadership rather than the military; most of the technologies listed are dual-use, sensor fusion in space using the Internet of Things and laser inter-satellite communication being presented as a means of measuring ocean temperature even as the same capability can track a ship — a framing that affords both flexibility and deniability. Recalling the AlphaGo–Lee Sedol match and the celebrated “Move 37”, he acknowledged that machines have a demonstrable role in decision-making, but assessed that neither the technology nor the confidence exists at present to hand decision-making over to them, with the existential dimension weighing heavily since no one knows what artificial general or superintelligence will look like. He noted that Chinese authorities have used facial recognition and cognitive AI in social engineering applications, and affirmed that fusing sensors and coupling them to kinetic elements is the entire principle of systems-of-systems warfare. He agreed that the widening gap must first be mitigated, with overtaking being a much later question, and that the human remains central, noting that the PLA has itself revised its training curriculum to develop officers with better joint operational skills as a precursor to employing technology. Elaborating on paralysis, he offered the analogy of the enemy as a human being whose brain is his command and control and whose spine is his network of nodes and connections: with these removed, his limbs and organs may function perfectly well but he cannot fight back coherently. This may be achieved through hard kill, cyber or electromagnetic means, selective jamming, or selective decapitation – not by killing the commander but by keeping him out of touch – the choice being determined by the operational effect sought. The ambition of systems warfare, he concluded, is that anyone should be able to communicate with anyone, an infantry soldier being able to reach a satellite to call in a hypersonic missile or naval fires from a destroyer at sea, what matters being not the chain but the effect brought to bear on the target. On Operation Sindoor the Speaker declined to comment, noting only that commercial satellite imagery providers had made detailed imagery openly available on the internet, an arrangement that remains deniable and cannot be firmly attributed.

The Report has been prepared by Ms. Khyati Singh, Research Analyst, North America & Strategic Technologies Centre, MP-IDSA.