India’s armed forces turn to startups to build battlefield tech at speed | External Affairs & Defence Security News
At the Bengaluru Tech Summit, officials from the Army, Navy, Air Force, and iDEX (Innovations for Defence Excellence), the Department of Defence Production’s innovation initiative, said closer industry collaboration, co-development, and funding support could help address gaps in manufacturing, artificial intelligence (AI), and technology sovereignty.
A key challenge for startups is moving from working prototypes to large-scale production. Officials also stressed that human oversight must remain central to defence AI.
India spent $92.1 billion on its military in 2025, up 8.9 per cent from 2024, making it the world’s fifth-largest military spender, according to the Stockholm International Peace Research Institute.
The institute credited the rise partly to the May 2025 conflict with Pakistan, which involved combat aircraft, drones, and missiles.
“Our role has changed from interacting with or purchasing equipment to co-developing with startups and industries,” said Wing Commander Gaddam Sandeep, commanding officer of the Regional Aerospace Innovation Division in Bengaluru.
Vipin Yadav, deputy programme director at iDEX, said India’s defence startup ecosystem faces a gap between technology readiness and the ability to manufacture products at scale. “When we are ready for a product, it doesn’t mean we will be able to deliver it in huge numbers,” he said.
iDEX provides grants to early-stage startups and offers minimum pilot order quantities to give them visibility on potential orders after product validation. It has also introduced measures to help startups build business cases and secure repeat orders.
Grants alone may not be enough to establish production lines, said Commodore S K Singh, officer in charge of the Indian Navy’s Technology Development Acceleration Cell, which identifies capability gaps and works with startups to address them.
Under the Defence Acquisition Procedure, startups working with iDEX and the Technology Development Fund, with cumulative contract values of up to ₹250 crore, can access incentives including using intellectual property as collateral, advances of up to 15 per cent of the base contract value, and payments covering 100 per cent of raw material costs against invoices or committed purchases. They may also receive exemptions from liquidated damages.
The armed forces do not prevent startups from selling dual-use products commercially, Singh said. He cited a firefighting robot developed with a ₹1.5 crore grant that has since secured orders from the Army, Navy, Air Force, and oil and gas companies.
Co-development can also expand a product’s capabilities. The Navy initially approached a startup to build an AI-powered underwater vision camera for visibility in turbid waters. The project later expanded to include an echo sounder, sonar, and metal detector. The Navy inducted the integrated system, which Singh described as world-class. The startup subsequently adapted the technology for an autonomous underwater vehicle, enabling persistent underwater operations.
The Navy is working with around 150 startups, more than 30 of them in Bengaluru, Singh said. Sandeep said Karnataka accounts for more than 65 per cent of India’s aerospace and AI industries. He called for centralised testing facilities and a hub-and-spoke model connecting startups with established aerospace companies.
AI built for the front lines
“The focus should not be on building larger AI models. It should be more on application-based small models which can help services in decision-making and surveillance,” Yadav said.
Sandeep said defence AI must evolve beyond prototypes because battlefield conditions and enemy tactics keep changing. Models need rapid iterations and testing in extreme temperatures and actual operational environments. “We don’t want huge computational models. We want small models, but it needs to have iterations faster,” he said.
The Navy needs task-specific models that can operate with limited data, power, and computing capacity, Singh said. Ships have restricted bandwidth when communicating with shore stations and can remain on patrol for two to three months, making frequent software updates difficult. The Navy therefore wants systems that can process data at the edge and detect rare events after long periods of inactivity.
Sandeep urged startups entering aviation to begin by modernising the Air Force’s legacy systems, some of which are 30 to 40 years old. They must also be prepared to maintain and support products throughout their 30-40-year service life.
Who owns the battlefield brain?
Access to classified data remains another challenge. Singh said the Navy wants developers to work alongside its personnel at naval facilities. Sandeep said the Air Force could share open-source and legacy data during development, but classified data would remain with the service once a model enters the defence domain. Further development would take place in a sandbox environment.
Lieutenant Colonel Sumanto Lahiri, officer in charge of the Indian Army Cell at the Army Design Bureau, said indigenous computing capabilities were essential to technology sovereignty. “Nobody is stopping us from developing our own sovereign software stacks,” he said.
The Navy has also launched a sovereign AI challenge through iDEX, requiring participating companies to transfer intellectual property rights and software to the service. The challenge attracted 68 industry proposals, Singh said.
Humans hold the trigger
Officials said AI would not replace human accountability in military operations. Lahiri said the Army had no scope for fully autonomous, AI-driven decision-making. “It is always human in the loop. The final call is taken by the human,” he said.
Singh said AI could independently support observation and surveillance, but human oversight would remain necessary for decisions and actions until the Navy developed sufficient confidence in the technology.
Building India’s next-gen arsenal
- Firefighting robots, underwater vision systems, autonomous vehicles move into military’s tech pipeline
- Scaling prototypes into mass production remains a key battle for defence startups
- Military seeks lean, task-specific AI models for battlefield conditions
- Indigenous computing, sovereign software emerge as critical lines of defence
- AI may scan the battlefield, but humans control military decisions
Source: www.business-standard.com
