Born in the chaos of COVID-19’s second wave, AGNIT Semiconductors raced from lab to working GaN chip prototypes in just two years. Today, it builds vertically integrated chips entirely at IISc’s pilot facility. Narrating their story, Hareesh Chandrasekar from AGNIT tells EFY’s Janarthana Krishna Venkatesan everything, from wafer growth to final testing.
Q. Can you give a brief intro to AGNIT Semiconductor and what it is working on?
A. AGNIT is a vertically integrated gallium nitride (GaN) semiconductor company. We make our own GaN wafers, then take those wafers to a foundry, process them, put our designs on dies, dice them, assemble, package, and test the devices. We started operations in January 2021 as a spin-off from the Indian Institute of Science (IISc), and our primary focus was to commercialise the GaN technology IISc had developed over the previous 15 years. The outcome is a product that goes into telecom, defence and other systems.
Q. Can you please elaborate on GaN and why it is your choice?
A. Normally, when I say semiconductors, most people are going to think of silicon. GaN is a compound semiconductor with material properties that differ from silicon. Practically, that means it has much higher electron mobility and a higher critical breakdown field: roughly five times the mobility and about three times the breakdown field compared with a silicon channel. In plain terms, it conducts current very efficiently when on and blocks high voltage when off. Those physical advantages let you build switches and amplifiers that are more compact, handle more power, switch faster and dissipate heat more effectively, which is why GaN is attractive for both RF (radio frequency) power amplifiers and power conversion.
Q.What solutions do you make with GaN material?
A. We manufacture GaN transistors for RF power amplifiers. These go into 5G base stations, radars, and secure defence communication systems. Our GaN amplifiers achieve around 60% efficiency compared to 40-45% for silicon or gallium arsenide equivalents, and they are about half the size. GaN makes those amplifiers more compact and efficient, enabling higher power density and wider bandwidth, which is important for 5G, private networks, radars, and other strategic uses.
Q. What is the difference between silicon-based chips and GaN-based chips?
A. The differences come down to material physics and system outcomes. GaN offers higher ‘on-state and off-state’ mobility, which I mean by this: the transistors are basically switches. When the switch is off, it should completely block a certain amount of voltage, and when it is on, it should conduct the maximum possible current. Ideally, GaN is more efficient than silicon at this. At the same time, it offers better thermal conductivity than silicon, making GaN devices smaller and compact for the same power, run cooler, and be more efficient.
Q. Is GaN actually cheaper at the system level?
A. At the device level, a GaN transistor may cost more than a comparable silicon part. But at the system level, the story can be different: GaN’s higher switching speed and power density reduce the size and number of passives (capacitors, inductors), lower heatsink and cooling requirements, and simplify thermal and mechanical design. When you factor in assembly, performance, and operational costs, in many applications GaN provides a compelling price-performance edge and, in some cases, a breakeven or better outcome compared with silicon-based designs.
Q. What wafer material should be used for producing GaN chips?
A. Native GaN wafers are extremely difficult to make, and the defects per centimetre square are exponentially large compared to a silicon substrate. So you typically start with a support substrate such as silicon, silicon carbide (SiC) or sapphire. The choice depends on the end application: RF components often use SiC for its thermal conductivity; many fast-charger components use GaN on silicon because silicon is cost-effective and available in large wafers.
Q. How do the manufacturing processes and investment requirements of GaN fabs compare with those of traditional silicon fabs?
A. The manufacturing processes have overlaps with silicon but also significant differences: GaN requires different chemistries, defect management and thermal-design approaches. In terms of capital, the scale differs dramatically. Cutting-edge silicon logic fabs that chase nanometre nodes are extremely costly; people often cite figures in the tens of billions of dollars. By contrast, you can set up a reasonably sized GaN fabrication line at a much lower scale; we discussed figures in the hundreds of millions rather than tens of billions. That makes GaN a more accessible entry point for countries and companies that cannot commit to the capital intensity of leading-edge silicon logic fabs.
Q. Which R&D hurdles have you encountered while developing your product?
A. You see, semiconductor development is inherently long and capital-intensive. Talent is another challenge: specialised process engineers and RF designers are scarce, and historically, many engineers moved abroad or into multinational design centres. Retaining and recruiting the right people is a continuous effort.
Q. How do you move your innovation from lab to market?
A. Having technology is one thing; commercialising it to solve specific customer pain points is quite another. You can publish papers and file patents, but product development requires aligning the technology to a system-level need, building prototypes, iterating with customers, and securing patient capital. For semiconductors, that cycle is long; prototyping takes time, tape-outs cost money, and you need a mix of grant funding and later venture capital to scale. We focused on building a prototype that customers can sample and test; from there, you refine and move towards production.
Q. Where do you run your production?
A. We run our pilot production and prototyping inside the IISc campus, in a small-volume GaN foundry that MeitY (the Ministry of Electronics and Information Technology) supported. That pilot line gives startups like us access to infrastructure without requiring the full capital outlay of a large commercial fabrication facility. We incubate there and run our processes and designs on the equipment available at that facility.
Q. How do you source materials, and where do you sell?
A. Our primary markets today are telecom, defence and space sectors that have specific technical requirements and procurement channels. Regarding materials and supply chain, fabrication requires many specialised chemicals and gases; hundreds of items across the process flow. The foundry maintains multiple sourcing options across geographies to mitigate geopolitical risks. We work closely with system integrators and customers to align component roadmaps with system needs, ensuring our chips fit into broader supply chains.
Q. How do you treat IP, and why is it important?
A. IP (intellectual property) for us is more than patents. It includes process know-how, trade secrets, and the operational knowledge embedded in how we grow wafers, process devices, and package them. We licensed background IP from IISc when we started, and we have filed our own applications, several in 2025 alone. Capturing IP matters because if the value sits only in services or foreign providers, the country and the company miss out on the long-term returns from product ownership. For semiconductor product companies, owning IP across design and manufacturing differentiates you and enables global competitiveness.
Q. Where do you see AGNIT and GaN going next?
A. Our immediate focus is defence, space and telecom, where there is apparent demand today. Parallel to that, we are exploring consumer and mobility applications; prototypes of compact electric vehicle (EV) chargers and household appliance converters are underway. More broadly, any place that needs efficient wireless transmission or power conversion is a candidate for GaN as the technology and manufacturing mature.
