A Bengaluru startup has raised ₹21.15 crore to develop an indigenous short-wave infrared sensor. The funding is a small round by venture-capital standards. The technology it is trying to commercialise points to a much larger question for India’s semiconductor ambitions: can the country build the sensing hardware that gives autonomous machines the ability to perceive the physical world?
- A different kind of vision for machines
- What the ₹21.15 crore round will fund
- Why defence is the first proving ground
- The wider opportunity: physical AI needs better perception
- The difficult journey from laboratory to chip
- Why this is a signal for India’s deep-tech ecosystem
- What to watch next
- FutureIsNow Signal
Artificial intelligence is increasingly moving out of chat windows and into machines. Drones inspect infrastructure, robots navigate warehouses, vehicles interpret roads and defence systems process information from increasingly sophisticated sensors. In each case, the quality of the machine’s decisions depends on what its cameras and sensors can detect.
That is the problem Bengaluru-based Nova Spectral AI is working on. Founded by Anish Priyadarshi and Sanjeev Kumar, the company is developing short-wave infrared (SWIR) image sensor chips using colloidal quantum dots integrated with standard silicon CMOS processes.
A different kind of vision for machines
Conventional cameras capture visible light. They are effective in ordinary conditions, but visibility can deteriorate sharply in darkness, smoke, haze, dust and light fog. SWIR imaging operates beyond the visible spectrum and can help cameras capture information in some of these difficult conditions.
SWIR itself is not a new scientific discovery. The commercial challenge has been the cost and complexity of the sensors. Many existing high-performance SWIR systems rely on indium gallium arsenide (InGaAs), a technology that can make broad deployment expensive.
Nova Spectral AI is pursuing a different materials and manufacturing route. Its stated approach uses lead-free indium arsenide colloidal quantum dots, integrated with standard silicon CMOS readout technology. The ambition is to make SWIR sensing more affordable and scalable than conventional solutions.
The distinction matters. The opportunity is not simply to build another camera. It is to develop a key component in the perception stack for machines operating in environments where ordinary imaging is less reliable.
What the ₹21.15 crore round will fund
On October 27, 2026, Nova Spectral AI announced a ₹21.15 crore seed round, approximately $2.25 million, led by Shastra VC, with participation from Equirus InnovateX Fund and angel investors. The company said the capital will support team expansion and development of a field-ready prototype within 18 months.
More Read
The funding is an early product-development milestone, not evidence of a commercially deployed sensor. The company is still working through materials, device and CMOS-integration challenges that sit between a promising laboratory process and a reliable, manufacturable chip.
On its website, Nova Spectral AI describes work underway on indium arsenide quantum-dot ink chemistry, film deposition, charge-transport layers and integration with CMOS wafers. Its published roadmap targets CMOS-wafer test structures in 2027, a sensor-chip prototype in 2028 and pilot fabrication and field pilots in 2029. The company separately says it is targeting a field-ready prototype within 18 months; these are company-stated development goals, not independently verified delivery milestones.
Why defence is the first proving ground
Nova Spectral AI identifies defence and intelligence, surveillance and reconnaissance (ISR) as its first market. Drones, night-vision systems and surveillance platforms need to operate in conditions where visibility can be limited. A more affordable SWIR sensor could, if it meets performance and reliability requirements, improve the range of platforms that can use this form of imaging.
There is also a strategic supply-chain dimension. The company argues that important infrared detector components used in Indian night sights, drone payloads and ISR systems are largely imported. Building more of the sensing stack domestically could strengthen local capability in a part of the electronics chain that is critical to advanced defence systems.
But domestic design alone does not establish supply-chain independence. Materials quality, fabrication access, packaging, testing, yields and qualification by customers all matter. Defence applications also require demanding validation before a new sensor can be trusted in operational environments.
The wider opportunity: physical AI needs better perception
The longer-term opportunity extends beyond defence. Nova Spectral AI lists aerospace, autonomous vehicles, industrial inspection, surveillance, wearables and consumer electronics among potential applications. In cars, SWIR could help detect certain road conditions; in industrial settings, infrared imaging can reveal material or moisture differences that are difficult to see with standard cameras.
These are potential markets, not announced customer deployments. Each has different requirements for sensor size, power consumption, cost, image quality, operating temperature and certification. A technology that works in a specialised defence payload will not automatically be ready for a smartphone or mass-market vehicle.
Still, the direction is significant. As AI systems interact more directly with the physical world, the bottleneck is not always the model doing the reasoning. It can be the quality of the data entering that model. Better perception hardware can expand what autonomous systems are able to detect, interpret and respond to.
The difficult journey from laboratory to chip
Nova Spectral AI’s approach has to clear several engineering hurdles. Quantum-dot materials must remain stable and consistent. Films need to be deposited uniformly. The photodiode stack must convert incoming light efficiently, and the device must integrate with CMOS readout circuitry without compromising performance.
Then comes manufacturing. A working device in a laboratory is only the beginning. The company will need repeatable wafer-scale processes, reliable packaging, robust testing and competitive production economics. Its ambition to make SWIR sensors dramatically cheaper than existing InGaAs solutions is a company target that will need to be demonstrated through measured performance and production costs.
There is also a commercial test. The startup will need to secure design partners and customers willing to evaluate the technology, integrate it into their systems and eventually commit to procurement. No commercial customer deployment has been established in the funding announcement.
Why this is a signal for India’s deep-tech ecosystem
India’s semiconductor conversation is often dominated by fabrication capacity, chip design services and large manufacturing investments. Yet the semiconductor value chain is broader. It includes materials, sensors, specialised components, packaging, testing and the intellectual property that differentiates one device from another.
Nova Spectral AI is working in one of those specialised layers. Its progress will depend on scientific execution as much as on capital, and the path to a product is longer than the funding headline might suggest. That is precisely why the company is worth following: it is attempting to turn a materials-science capability into a component that could matter to defence and the emerging physical-AI market.
What to watch next
The next meaningful milestones are technical and commercial: progress on CMOS-integrated test structures, independent performance data, a packaged sensor prototype, field evaluation with prospective users and evidence that the cost advantage can survive manufacturing scale-up.
For now, the ₹21.15 crore seed round funds a serious attempt to solve a difficult hardware problem. It does not yet prove that the company has solved it. The distinction between those two stages is important in deep tech, where years of engineering separate a promising concept from a dependable product.
FutureIsNow Signal
The next phase of physical AI will depend on what machines can see, not only on how well they can reason.
Evidence: Nova Spectral AI has raised ₹21.15 crore to develop SWIR image sensor chips based on indium arsenide colloidal quantum dots and silicon CMOS, with a field-ready prototype targeted within 18 months.
What remains unproven: Commercial-grade sensor performance, repeatable manufacturing, the claimed cost advantage and customer adoption.
Implication: India’s deep-tech opportunity includes the specialised sensing components that allow autonomous systems to perceive difficult environments.
Watch next: CMOS integration results, prototype validation, defence evaluation partners and evidence of a manufacturable cost structure.
