Sandeep Kumar, CEO of L&T Semiconductor Technologies

BBack in 1986, five freshly minted college students who had joined semiconductor giant Texas Instruments (TI) were on the verge of their big break. As Apple, Microsoft and other companies were making microprocessor units (MPUs) for servers, desktop computers and mobile computers, Dallas-based TI put together a new team and tasked the young guns with making microcontroller units (MCUs) that go into refrigerators, washing machines and cars, among other things. TI focused more on microcontrollers because, unlike microprocessors, they do not require complex operating systems to function. “We managed to make the first cell phone chip, the microcontroller with adaptive voltage scaling (AVS) and the motor controller. In ten years, we grew our business from zero to around $700 million,” recalls Sandeep Kumar, CEO of L&T Semiconductor Technologies. After completing his BTech from IIT Delhi, Kumar, a native of Chandigarh, moved to the US and did his MS in electrical engineering from the University of Cincinnati before joining TI. Things came full circle in 2021 when Kumar’s classmate from IIT, who was running L&T’s defense business, suggested he start a semiconductor company in India. Kumar put together a plan and contacted several companies, including L&T. Two years later, Larsen and Toubro showed interest, but instead of investing in the company, they asked him to build the entire business for them. In 2023, L&T Semiconductor Technologies (LTSCT), a wholly owned subsidiary of L&T, was formed and Kumar took over the role of CEO in November. LTSCT is a fab-less semiconductor company that will focus on developing chips. They will outsource manufacturing to US-based Global Foundry, Taiwanese company TSMC and Vanguard. Their product portfolio includes smart devices for MEMS sensors, power, analog, mixed-signal and radio frequency (RF) products to support the automotive, industrial and energy sectors. LTSCT has onboarded six automotive companies and the total value of the business is $150 million per year. “We have a pipeline of about 20 automotive customers in Europe, Japan and India and out of these we are signing six. The rest will also be signed soon,” says 61-year-old Kumar, who lives in Austin but is currently in India. The engineering and construction group is going the “chip design” route rather than “chip manufacturing” because of the lack of indigenous chip design companies in India that can develop designs and intellectual property (IP). There are a number of small startups working in the semiconductor design space but none of them have made it big. LTSCT’s focus on developing custom chip solutions based on application-specific integrated circuits (ASIC) would help them occupy a niche, especially given the growing boom in computing in the age of artificial intelligence, says Neil Shah, Vice President of Counterpoint Research. In a conversation with Forbes IndiaKumar spoke about the areas India needs to focus on in building the upcoming semiconductor ecosystem, L&T’s entry into the future sector and why the government’s Design-Linked Incentive (DLI) scheme needs to change. Edited excerpts:
Q. The Indian government is encouraging chip manufacturing in India. Why did L&T decide to enter into chip design?
Three years ago, the Indian government was putting a lot of emphasis on manufacturing and the incentives were geared towards bringing in 7 nanometer technology, which is a very advanced technology. In my view, nobody is going to bring that and we need to set up a manufacturing company. I had prepared a plan for that. After some time, the government also realised that that was not going to happen. They changed the incentives to 180 nanometer to bring the technology into the country anyway. But the fundamental gap, in my opinion, is still the same: never build a factory if you don’t know what product you are going to make in it.
Q. Can you tell us more about why the focus is specifically on the automotive, energy and industrial sectors?
Most semiconductor manufacturers don’t like to get involved in these markets because it takes three years to develop and qualify the product and the actual sales start in five years. Also, the quality and reliability requirements are very high. The positive side of this is that they will buy from you for 10 years. Also, there is a lot of room for new innovation because there are elements of sensing, edge AI, control systems and it’s not just about computation. You have a human interface, a mechanical interface that goes into an electronic interface. We need to understand the electromechanical system in the industrial or automotive space. In the energy space, we need to understand power distribution, power generation, turbines. So it’s not just about data manipulation like it is in mobile phones or PCs. Here you need to have cross-functional expertise. We are building a capacity to make 10 to 15 products in parallel. To execute each of these products, on average, it takes 30 engineers in a two-year period. Development takes 12 months, testing takes six months, and qualification takes three months, so it will take about one and a half to two years before you are ready for production with a new product.
Q: What size will the chips be that LTSCT will develop?
In the automotive sector, smart analog and smart power devices are generally manufactured at 130 to 45 nanometers. The smart RF and compute devices are manufactured anywhere from 28 to seven nanometers. And the power devices for electric vehicle electrification are built on silicon carbide or gallium nitride, where geometry is not as important. They are large devices measured in micrometers, not nanometers.Also read: Focus on ‘Make in India’ and emerging manufacturing sectors like semiconductors are crucial for new coalition government
Q: Does LTSCT plan to gradually enter chip manufacturing?
This is more of a math game. A silicon factory needs $10 billion in investment. That’s the low end. It can be more. To make money when you build a factory with that much capital, you need to be able to sell at least $1 billion worth of product a year out of that factory. So I need to know how to sell $1 billion worth of product before I build a factory. I can do that in seven to eight years; then I need a factory. Now if I know how to do that with a high degree of certainty in the next three to four years, I can start building the factory in three to four years so it’s ready in seven to eight years, because it takes four years to build.
Q. LTSCT has invested $100 million so far. Have you received any support from the government’s Design-Linked Incentive (DLI) program?
The answer is no, and the answer is partly no because the DLI scheme, at least as of now, does not allow companies like L&T to avail this funding. It is restricted to a few startups, and there is a very strict definition of it. Maybe it will be revised, and then we will see if it makes sense to use it. L&T has deep pockets and enough capital, so I don’t think capital is the issue for us. We easily have three years of startup time, of which we have used nine months. We are looking for acquisitions. We recently acquired a semiconductor design startup, SiliConch Systems. We are looking at several more, and all of them are aimed at accelerating our product portfolio, revenue acceleration and talent acquisition on all three fronts.