Inside India’s push to build deeptech for real-world deployment
A panel at TechSparks 2025 explored India's endeavour to build advanced homegrown hardware and intelligence systems for deployment in real-world systems.
India’s deeptech landscape is entering a new phase, where ambitions extend beyond software into aircraft, factories, and sensing systems that operate in the physical world. As robotics, advanced manufacturing, and AI-driven perception converge, the central question is whether India can build end-to-end systems with the precision, reliability, and integration required for real-world deployment.
At TechSparks 2025, the panel 'Deepening India’s deeptech vision with robotics in real-world systems' explored this shift through three lenses: electric aviation, precision manufacturing, and large-scale 3D perception.
India may have missed the traditional industrial revolution, but electric aviation is creating an opportunity to re-industrialise on modern terms.
Adrian Schmidt of Sarla Aviation highlighted how electric propulsion provides a rare reset for aviation. Electric motors, he explained, open the possibility of indigenising the entire value chain, from motors and powertrains to systems integration and certification.
“Electric motors as a powertrain enable us, for the first time, to bring aviation’s entire value chain end-to-end into India,” he said.
Sarla’s first air taxi, Shunya, is positioned as a formative model rather than a final product, "as the beginning of something, not the final answer," said Schmidt.
He emphasised that government ministries and regulators are increasingly acting as enablers. “Certification and regulation are being treated as accelerators rather than obstacles.”
Kaushik Mudda, Co-founder and CEO, Ethereal Machines, described the hidden complexity of high-precision manufacturing. Ethereal operates a nearly fully automated 'dark factory' with more than 70 machines running 24×7, producing micron-level components for aerospace, defence, medical, and semiconductor sectors.
“Most of these critical components aren’t manufactured in India today because they require micron-level precision, tolerances measured in tens of microns. Maintaining an automated floor continuously requires a combination of software, mathematics, and electronics. It’s a marriage between software, mathematics, and electronics that gives the shop floor the strength to keep running 24×7,” he explained.
Mudda also highlighted a deeptech founder's challenging journey, “It’s extremely brutal to be a deeptech founder. It took many years to raise follow-on capital, and hardware requires many costly iterations.”
The discussion then turned to perception, with Tushar Chhabra, Co-founder and CEO of CronAI, describing how AI for the real world must be built from first principles. “Most people implement existing algorithms; true AI means going down to the basic mathematics and writing the model from scratch,” he said.
Off-the-shelf models fail under the variability of Indian conditions. So CronAI developed its own learning architecture and spent years building datasets across cities such as Bengaluru, Delhi, Mumbai, and Chennai, and Tier II and III regions.
Cron provides a 3D sensor data perception processing platform. This platform uses deep learning AI to process raw data from 3D sensors like LiDAR (Light Detection and Ranging) to deliver real-time, actionable intelligence for various applications.
LiDAR, Chhabra noted, is "an active camera" critical for machines to accurately perceive three-dimensional space.
Real-world deployments demonstrate the stakes involved. “We deployed a mile of sensors along the border, we detected nearly 100 intrusion attempts in the first week, and casualties dropped to zero because no one had to step out of the bunker,” said Chhabra.
Edited by Swetha Kannan