In remarks that at once seemed an encouragement and a challenge to India’s defence industry, Chief of Defence Staff General N S Raja Subramani told a gathering of defence manufacturers in New Delhi in October that rather than pointing to a lack of orders, companies should “come up with an innovative product that we are compelled to buy, or which we need to buy. Create the demand first.”

The chief of defence staff (CDS) made the point while answering a question from Baba Kalyani, chairman and managing director of Bharat Forge, who was in conversation with him at a fireside chat and had framed it around this very idea. Following the CDS’ opening remarks, Kalyani said that the United States (US) was moving towards innovation and Europe towards building capacity, with India somewhere in between. To illustrate the American approach, he pointed to Anduril, a US defence-technology company, whose co-founder Palmer Luckey uses what Kalyani called a very simple concept: “Build first, then offer.” Luckey, Kalyani added, says the company was built “as a product company and not as a contract company”. Kalyani linked this to what the CDS had said at the outset: that industry needed volume, scale and speed, and that the armed forces would increasingly tell it the capability they wanted, rather than the product, and work with it. He then put his question to the CDS: the war in West Asia had shown that large quantities of missiles, rockets and ammunition were needed, so what did India need to do in that respect?

The CDS, too, turned to Anduril to illustrate his point. The company, he said, had once been an outsider to the US defence-industrial complex, where the big prime contractors (Boeing, Lockheed Martin, RTX, General Dynamics and Northrop Grumman) held a monopoly on procurement. Anduril broke in, the CDS said, by first building a software, Lattice, which was better than what existed and could work with existing systems, starting with sensors that mapped borders.

But the CDS also made clear that the burden would not fall on industry alone. “I think it’s time for us, that is, the defence forces and the industry, to talk to each other more and work together rather than looking at a supplier-vendor relationship. We have to look at each other as industry partners,” he said. The armed forces’ part of that bargain, he said, would be to give a clear demand signal for the next five to seven years, along with a technology capability roadmap, which has already been laid out in the Technology Perspective Capability Roadmap.

Founded in 2017 and based in Costa Mesa, California, Anduril supplies drones, artificial-intelligence (AI) software and augmented-reality headsets to the US and its allies. A $5 billion funding round in May 2026 doubled its valuation to $61 billion. India’s startups, and sections of its wider private defence industry, are drawn to it by its rapid growth, its multibillion-dollar contracts and its reach into government. The country’s defence establishment, however, may find a more useful lesson in how it does business. Behind Luckey’s “build first, then offer” is a business model: Anduril develops products with its own money, raised through venture capital, and sells to the military afterwards. That reverses the order on which militaries around the world, India’s included, have long relied: the armed forces fix detailed requirements through a formal procurement process, and only then does industry build them. Companies such as Anduril have begun to test that order, and it is this model that both Kalyani and the CDS held up to the Indian industry. The battlefield adds weight to their call. In the war between Russia and Ukraine, aerial drones have passed through eight phases of development in just four years, according to the French Institute of International Relations (Ifri), turning from basic reconnaissance aids into advanced weapon systems partly coordinated by AI. That raises a pressing question: when warfare changes this fast, how should a military tell industry what it wants? 

The challenge

“The big companies, who have the money and the muscle, need to do more. They should look at research and development (R&D) as an investment and not an expense. Maybe, the environment is not enabling enough and the risk-reward is too adverse,” said Air Marshal Narmdeshwar Tiwari (retired), former vice-chief of the Air Staff of the Indian Air Force. In India, he said, defence R&D is done either by the government, where it is mandated, or by startups, for which it is the only way to survive. A culture of private sector R&D  is well established in the US, but still nascent in India.

The defence procurement process, he said, is designed to give everyone a level playing field. Requests for information and proposals, and selection of the lowest bidder, ensure that no one can complain of favouritism. “There were small companies that were very impressive, but we could not move the case, because we cannot buy on a nomination basis, which we can do with DPSUs (defence public sector undertakings).” 

In the US, Anduril and Palantir (a Miami-headquartered maker of data-integration and analytics software), two of a new breed of defence-technology companies, blame both the incumbents and the way the US military buys for holding back innovation in American defence. In January 2025, Palantir’s Chief Technology Officer, Shyam Sankar, told a US Senate committee that the defence market was “functionally a monopsony, where the sole buyer shapes the market with overly prescriptive requirements”. The cost shows, Anduril argues: “There is more AI in a Tesla than in any US military vehicle; better computer vision in your Snapchat app than in any system the Department of Defense owns.” Its answer is to “build to the mission, not to spec”, with companies risking their own money. They have won some ground. In November 2025, US Secretary of Defense Pete Hegseth directed that commercial products be the Department of Defense’s preferred option, with new development undertaken  only “if nothing exists”. In October 2026, Luckey and SpaceX chief Elon Musk were among those picked to lead Project Meridian, a Department of Defense study of future battlefields and the weapons needed to dominate them.

Yet even in the US, the old model is far from finished. In March 2025, Boeing won the engineering and manufacturing development contract for the US Air Force’s sixth-generation fighter, the F-47, a deal reported to be worth more than $20 billion. Hegseth’s own

memo calls for a “high-low mix” that balances “exquisite, high-end solutions” with cheaper, more readily available ones. If the build-first model has not displaced the old order where it was born, India must also ask whether it has what made the model possible there.

The gap runs deeper than famous names. Building first with one’s own money assumes a deep base of research and the intellectual property it yields. India spent 0.84 per cent of its gross domestic product (GDP) on R&D in 2023–24, an all-time high and up from 0.64 per cent in 2020–21, according to the Department of Science and Technology. The US spent 3.43 per cent in 2022, and China 2.58 per cent in 2023. Indian private industry has also stepped up, with its share of national R&D spending rising from 32.2 per cent in 2020-21 to 45.2 per cent in 2023-24. Defence, however, is the exception: of the industrial sectors that spent the most on R&D in 2023–24, it was the only one in which the public sector, rather than the private sector, accounted for most of the spending, according to the department’s data. Despite the overall gains, AI patents also show the distance India still has to travel. It accounted for 0.4 per cent of AI patents granted worldwide in 2024, against 12.1 per cent for the US and 74.2 per cent for China, according to the AI Index Report 2026 by Stanford University’s Institute for Human-Centered Artificial Intelligence. Indian firms may not yet be positioned to work the way Anduril does. That does not leave India stuck with the old system, however.

The opportunity

The CDS’ call at the October gathering in New Delhi for the armed forces and industry to work together as partners points to a third path between the conventional way of buying defence materiel and the model Anduril wants. Anduril’s ambition goes beyond building first. It wants defence companies to have their own “vision of the future”, as the most successful technology firms did, firms whose success, it says, came partly from working on things their customers “did not believe were possible or did not yet understand”. In effect, Anduril wants industry, not just the military, to shape what future warfare looks like. In the third path, perhaps better suited to the present state of India’s defence industrial base, the armed forces set out a capability and a cost ceiling, and industry develops a product with the prospect of orders.

Answering Kalyani’s question on missiles and ammunition, the CDS showed what the clear demand signal he had said the forces would give might look like in practice. Long-range precision missiles, he said, are “the need of the hour”. He has asked industry for a range of 500 to 600 kilometres (km) to begin with, which can be increased later. He also set a cost ceiling for each missile, and said industry should aim to come in below it, “so that we can have more and more”. Companies, he added, would need to tie up their own propulsion, propellants and explosives. “And if you come to us with this, we will ensure that we give you adequate orders, because this is what the demand is at the moment.”

He set out the same deal for other needs. On counter-drone systems, lasers and directed-energy weapons, he asked companies to “come to us with a workable product where you have a clear IP (intellectual property) on it”, after which the forces could work towards partnering and buying.

Every item on that list belongs to a field where, as the war in Ukraine has shown, technology moves fast. Crewed platforms such as submarines and fighter aircraft are a different matter. They take years to design, cost far more and stay in service for decades.

“The industry-first model is best suited to the layer that connects platforms, and not to the platforms — especially crewed next-generation combat aircraft, warships and submarines — themselves,” said Bodhisattwa Sanghapriya, founder and chief executive of IG Defence, a Noida-headquartered maker of drones and loitering munitions whose systems were used during Operation Sindoor in May 2025. Like Anduril’s Lattice, IG Defence’s own battlefield-management and autonomy platform, ‘GRID’, lets one operator command multiple unmanned systems as a coordinated force. Such capability must come from industry first, he said, because it improves with field data, integrates systems across arms and vendors, and rests on technologies moving faster than procurement cycles. The same holds, he said, for systems that detect and defeat drones, for drones operating together in large coordinated groups, for navigation when satellite signals are jammed, and for military mobile communication networks built on 5G and 6G technology. “In all of these areas, the battlefield sets the pace.”

IG Defence has also developed KAL, a one-way attack drone with a range of up to 1,000 km, of the class Iran’s Shahed-136 made prominent, and aims for a ₹10,000-crore defence-business pipeline by 2029–30. “We aim to be India’s Anduril,” Sanghapriya said.

For the faster-moving kind of purchase, several countries have set up routes outside their conventional procurement systems. The US buys commercial technology through the Defense Innovation Unit, using contracting mechanisms outside the normal rules, said Cchavi Vasisht, associate fellow at the New Delhi-based Chintan Research Foundation. Ukraine uses “Brave1”, a government-backed defence-technology cluster, accelerator and online procurement platform, and “DOT-Chain Defence”, a military e-commerce marketplace run by its Defence Procurement Agency, where frontline units choose codified drones and the state pays. “India can adapt the best of each,” she said.

For that to work, each side must change. The military, Vasisht said, needs to move “from rigid specifications to sharing operational problems with industry early”, and the forces could open test ranges, field data and user units for trials. It is fine for the forces to set out requirements in qualitative terms, Tiwari said, but fixed quantitative ones, such as weight or horsepower, can be a challenge, since R&D starts with an idea and needs flexibility. “If all the staff qualitative requirements are to be met, nothing will get done. They should be a guiding rail, not binding restrictions.” Companies can carry development costs, Vasisht said, if the government offers assured orders for proven products. Value-based pricing, timely payments and clear ownership of IP matter too.

If India is to build its own model, it must “repose faith in the private sector”, said Kartik Bommakanti, a senior fellow in the Observer Research Foundation's Strategic Studies Programme. The forces, he said, need to be clear about what they expect from industry on a specific capability, so that a company can commit its own people, technology and money to it. A company may also develop a capability on its own initiative, without a requirement, which the forces then find they can use. Once a product passes tests against the qualitative requirements, he said, the service concerned and the government must compensate the company fully. There is reason for caution. Vasisht cited Amit Cowshish, a former financial adviser (acquisition) at the Ministry of Defence (MoD), who notes that no major project under the Make-I route (a government-funded route for developing prototypes with Indian industry) has come to fruition since the procedure was adopted in 2006. No project under the Strategic Partnership model (under which Indian private firms tie up with foreign manufacturers to build major platforms in India) has matured since 2017. “Industry will remember that record when deciding whether to self-fund,” she said. The draft Defence Acquisition Procedure (DAP) 2026 moves in the right direction, she said, pointing to spiral development, provisions for five years of follow-on orders after Make and Innovations for Defence Excellence (iDEX) projects, a fast-track procedure for buying emerging and fast-changing technologies, and expert input into the services’ qualitative requirements. But the overall process “still largely starts with a services requirement and a tender”. Operationalising the Low Cost Capital Acquisition route, a new acquisition category proposed in the draft DAP 2026 that allows proposals submitted by companies on their own initiative or by a single vendor, would be a welcome move, she said, adding that “a dedicated route to evaluate and buy ready, unsolicited products, including single-vendor and software-licence cases, would strengthen such a push”.

Bommakanti said it was unclear whether the draft DAP 2026 accommodates comprehensive compensation for companies that develop capabilities with their own money. It may need provisions stating specifically that such companies will get full reimbursement and a guarantee of purchase, he said. Most Indian experts, Vasisht said, agree that “the policy intent is now ahead of institutional behaviour.”

How, then, should a military tell industry what it wants? For drones, missiles and counter-drone systems, the answer taking shape in India is a capability, a price and a promise of orders, rather than a detailed specification. For submarines and fighters, the old discipline will hold. And the CDS’ challenge to “create the demand first” runs both ways. Industry must stake its own money on products the forces cannot ignore. The forces, and the government’s rule book for defence procurement, must make that bet worth taking.