L&T’s semiconductor plan: Fabless, fabrication and chiplet packaging

A few years ago, the concept of semiconductors might have seemed complex to the average person. But the COVID-19 pandemic changed all that. When $10,000 cars were stuck in production due to a $2 chip shortage, the vital role of semiconductors—also known as chips—suddenly became clear to governments, industries, and consumers alike. The realisation was stark: chips power everything from automobiles to space missions, and any disruption in their supply chain can have far-reaching consequences.

The Indian government's decision to take the semiconductor industry seriously, after several failed attempts over the past few decades, comes with the launch of the India Semicon Programme. With a financial outlay of ₹76,000 crore, this ambitious initiative aims to establish semiconductor and display fabs within the country.

While Indian conglomerates like the Tata Group and Murugappa Group have sought government subsidies to set up massive semiconductor fabs or testing plants, L&T made a bold decision: it would focus on semiconductor design, building its own intellectual property (IP) and creating a new footprint in the tech ecosystem.

Sandeep Kumar, chief executive, L&T Semiconductor Technologies (LTSCT) told Fortune India: “It is great to have manufacturing, but that still doesn't make India self-reliant. It brings in manufacturing capability, but self-reliance comes from being able to build products. Even with the PLI scheme on EMS, when we bring in (manufacture) Apple iPhone into India, it's still not an Indian phone. It is manufacturing, it creates jobs and starts creating the ecosystem. The real value or breakthrough is when you start building products over here and build a company like Cisco, Texas Instruments, or Microsoft in India, which delivers products. So that's the difference, and that’s why we are setting up L&T Semiconductor.” 

Going fabless

Shortly after the pandemic, L&T faced the strategic challenge created by its complete dependence on imported semiconductors for its defence vertical, as all of L&T’s semiconductors were sourced from abroad. “About 80% of the semiconductors we consume in India come from China — though not necessarily Chinese-made, but manufactured there — and if we include Taiwan, nearly all of our semiconductors originate from that region. In defence, there has always been a drive to achieve some level of independence from the global supply chain. That’s how L&T’s journey with semiconductors began,” explains Kumar.

But semiconductors is a business that requires a large investment, and, therefore, it requires big markets to make it financially viable. Defence procurement, although strategic, does not have sufficient volumes to justify a standalone plant. So instead of just getting into semiconductor manufacturing, L&T’s vision evolved to “let’s bring semiconductor capability and capacity into India, and once that is there on a product level, and we have that knowledge to build it here, and the talent, experience and the business wherewithal, then you can also build solutions for defence,” adds Kumar.

Roughly 20% of the world’s integrated circuit (IC) design resources come from India, yet the country lacks semiconductor giants like Intel. However, whether it’s Intel, Micron, or any other major semiconductor player, nearly every global chip company has its second-largest, if not its largest, design team based in India.

Kumar explains, “To set up a fabless chip company, you need the other two parts – the front-end definition and the back-end support or customer service again. And then you need a solid engineering team to implement it.” He adds as the front-end talent and the back-end talent doesn't exist in India, the company has set operations in Munich, Tokyo, Austin and Detroit, mainly in Austin, where a lot of the architecture work is happening. In parallel, he is setting up teams in India to learn from those so that the capacity and capability can also be built in India slowly.

Another challenge faced by fabless companies is the significant investment required, with developing a single product potentially costing anywhere from $10 million to $50 million, in addition to the expenses involved in marketing and selling the chip. In the case of LTSCT, the company is taking a strategic approach by first targeting customers, securing their business, and gaining a deep understanding of their needs before moving forward with product development.

“And because we have L&T with deeper pockets than a startup running around trying to raise capital from the VCs, we are able to create a pipeline of 15-20 products in parallel. Almost all startups, the conventional ones, can only do one product because doing a single product is a $10-$50 million commitment. Here we're talking about doing 15 to 20 products in parallel. So multiply that number by 20,” adds Kumar.

In the fabless space, LTSCT is working on power semiconductors, high-performance analogue, RFICs, sensors, and processes for targeted markets where the company will go down into the seven-nanometer room. With these solutions, LTSCT caters to the automotive, industrial and energy sectors. “Everything we are doing is relevant for defence,” adds Kumar.

To start with, LTSCT has already acquired Indian startup SiliConch Systems for Rs 183 crore. “We are doing power, high-performance analogue, sensors and RF. To jumpstart each one of these functional elements, we are looking at acquiring to bring in talent, IP, some products and customers, in that order. For us, talent is the most important. Assembling a brand new team of 60 people and having them work together in a cohesive way can take a year. But if I can pick a team of 60 people who have already been working together for many years, I can jump-start. Then having good technology then I can jumpstart it and having a product that can jumpstart in that order.”

Chip fabrication

LTSCT has adopted a fabless model to start with, focusing exclusively on chip design and outsourcing the manufacturing process. In contrast to Tata Electronics, which is establishing a semiconductor fab in Gujarat, LTSCT has no plans to enter the foundry business—where a chip factory is set up to produce chips for other companies. This approach is similar to the semiconductor landscape in Taiwan, where most operations are centred around the foundry business, with major players like TSMC handling the production of chips designed by others.

“We would be setting up a factory down the road, at whatever appropriate time, to build our products, not other people's products. We will be an Integrated Device Manufacturer,” says Kumar. LTSCT’s plan goes like this. The company will first develop its IPs and products. It will initially use a foundry to build (print the chips) them. Once the company ramps up volume, it will then consider building a factory. “It’s a trade-off between gross margins and profitability and having your factory can help in achieving a more aggressive pricing or higher gross margins,” says Kumar.

Chipset packaging

The next step after chip fabrication is packing, and LTSCT has no plans to venture into conventional testing and packing of semiconductors. About 90-95% of the chips today are packaged in this way, which is advanced technology but no breakthrough. For its chips, Kumar is open to leveraging whosoever sets it up (in India) – be it CG Power or Tata.

Then there is 5% advanced packaging -- around power modules, where instead of microamps or amps of current one is delivering gigawatts or megawatts of power in a package. That requires special techniques. Then there are chiplets, which help build and stack 12 chips into a single die, one on top of another. “These are advanced packaging, and we are looking into it. That would be like 5% today of the total semi-volume that goes through it, maybe it will go to 10% in five years. So we would be having the same ratio and that 90% of our products will go through an external OSAT house and maybe 10% of our revenue and products comes from our advanced packaging facility,” says Kumar.

Investments and going public

Kumar reveals that for the fabless segment, the company has committed $100 million, with plans to increase it to $300 million. Depending on how the venture progresses and evolves, the investment could surpass this amount. When the time is right for L&T to enter semiconductor manufacturing, the company is prepared to allocate the necessary funds to support the expansion.

Responding to Fortune India’s question on whether or how soon the company plans to go public, Kumar stated, “Of course, L&T is cautious, conservative, and rightfully so, in my mind. So, once we have good revenue, we have clear growth beyond the revenue for the foreseeable future. Once we hit those two, then it will get listed.”