Despite late start, can India make gains in global chip race?

Industry executives are exultant that global semiconductor companies will see the India advantage, but a large local demand holds the key

Harichandan Arakali
Published: Mar 20, 2024 01:41:33 PM IST
Updated: Mar 20, 2024 03:51:14 PM IST

India’s semiconductor demand was estimated at about  billion in 2022 and projected to reach 2 billion by 2032. The country’s electronics ministry envisions an overall products manufacturing output of 0 billion by 2026, versus  billion in 2022.
Image: ShutterstockIndia’s semiconductor demand was estimated at about $26 billion in 2022 and projected to reach $272 billion by 2032. The country’s electronics ministry envisions an overall products manufacturing output of $300 billion by 2026, versus $75 billion in 2022. Image: Shutterstock
 
“The ecosystem from across the globe will mobilise to have India as their preferred semiconductor destination,” N Chandrasekaran, chairman of Tata Sons, said in a statement on March 13, to mark the beginning of construction of two chip factories by the Indian conglomerate.
 
One will be a wafer fabrication plant and the other an OSAT (outsourced semiconductor assembly and testing) facility. The former in Dholera, Gujarat, and the latter at Jogiroad, Assam, “2,500 km apart, will have a lasting impact on the nation,” the Tata Group boss said.
 

The previous month Prime Minister Narendra Modi’s cabinet had approved these projects and another one led by the Chennai-headquartered Murugappa Group—also in Gujarat. Together, they will see investments of $15.2 billion, at least half of which will come from the government’s $10 billion fund launched in 2021 to attract global semiconductor manufacturers to India.
 
In the case of the Tatas, group company Tata Electronics will set up the semiconductor fab in partnership with Taiwan’s Powerchip Semiconductor Manufacturing Corp (PSMC).
 
Previously, the government had approved a $2.75 billion ATMP (assembly, testing, marking and packaging) project, in Gujarat, led by America’s Micron Technology, a memory chip specialist—the contribution from the government fund is about two-thirds, with Micron investing $800 million. The hope is that together, these projects and those that may follow, will finally drag India into the hi-tech age as a more self-reliant maker of products and not just a consumer.
 

Local demand

“The biggest factor in my mind is, where is the demand,” Karthee Madasamy, founding managing partner at the US based deep-tech focussed VC firm MFV Partners, told Forbes India in a recent interview, speaking from Silicon Valley. “That, I think, needs to be still figured out. How many box makers, how many PCB makers are here to consume the chips that will come out of this fab.”
 
The VC investor himself is a semiconductor chip designer by training, who has contributed to developing 16 chips in his career. Previously, he was a managing director and vice president at Qualcomm Ventures, the VC arm of smartphone chip giant Qualcomm.
 
A fab is viable only when it can perform at high volumes and low costs or at least comparable costs to alternatives. What is different today is that India is consuming a lot more semiconductors than in, say, 2006 or 2012—the last time there was serious talk of fabs here. There is a significant volume of local assembly, with the presence of large international manufacturers.
 
India’s semiconductor demand was estimated at about $26 billion in 2022 and projected to reach $272 billion by 2032. The country’s electronics ministry envisions an overall products manufacturing output of $300 billion by 2026, versus $75 billion in 2022.
 
What is missing is local value addition, which today is widely accepted to be only in the single-digit proportion.
 
The hope is that the Tata-PSMC partnership will attract other ecosystem players and boost the local high-value-add component. They are also being careful in that they're going after some specific, relatively low-hanging, opportunities, starting with wafer-etching processes that result in “mature nodes”, as such tech has been around for 10-15 years or even more.
 
The most advanced chips today are nodes of 3nm to 10nm (1nm is one billionth of a metre) –such as the processors for top-end smartphones and laptops. The Tata-PSMC fab will produce chips made from 28nm, 40nm, 55nm, 90nm and 110nm processes, Randhir Thakur, CEO and MD, Tata Electronics, said in a press release on February 29.
 
Historically, the “process node” or simply “node” derived its name from the physical dimensions of an individual transistor, with billions of them packed in the chip. Today, it’s more a terminology of convenience, including for marketing, because the process itself has completely changed, Markus Vomfelde, head of business development for the EMEA region at Renesas Electronics Corp., noted in a July 2023 blog on the company’s website.
 
Renesas is the tech partner for Murugappa Group’s CG Power and Industrial Solutions, which is building its ATMP plant in Gujarat.
 
Starting with the older tech also means that the money needed for such plants is relatively more modest compared with the “frontier nodes”, Madasamy points out.
 
The new Tata-PSMC semiconductor fab will manufacture chips for applications such as power management integrated circuits, display drivers, microcontrollers, and high-performance computing logic, addressing the growing demand in markets such as automotive, computing and data storage, wireless communication and artificial intelligence, Tata Sons said in a press release on March 13. “With a manufacturing capacity of up to 50,000 wafers per month, the fab establishes India as a key supply chain partner in the global semiconductor industry,” the Tatas say.

Also read: India's semiconductor dream gets a push as PM lays foundation for three manufacturing facilities

 

China’s leap

In comparison, China produces millions of wafers per month. And with 44 existing fabs, Chinese wafer makers, with already about a third of the global mature-node market share led by companies such as SMIC, have plans to add several more in the coming years, market researcher TrendForce notes.
 
The Tata-PSMC plant is expected to be ready for commercial production in about three years.
 
China is estimated to spend $150 billion in chip-making through 2030, the Wall Street Journal noted in a July 2022 report after the US announced a $77 billion plan to boost domestic fabrication. Global chip giant South Korea had a $260 billion five-year plan, the EU was pushing a $40 billion public-private investment and Japan would spend $6 billion, the Journal noted at the time.
 
For Tata-PSMC, and India, this all means a large local demand is crucial, notwithstanding the so-called China-plus-one strategy of global companies to de-risk their supply chains.
 
“Wherever there is a gap in the industry, that is where the lowest hanging opportunity is, that is where Tata and PSMC will work on together,” says Neil Shah, research vice president at Counterpoint Technology Market Research.
 
The order pipeline will initially be local, Shah says, with applications in space, defense, industrial and, for example, 300 million or so houses will need smart meters in next five to six years.
 
Automotive is another mature industry in India, he points out. If one can create the smaller components, which can be locally sourced, it could add up to a lot of local demand.

Also read: Vibrant Gujarat 2024: India's first semiconductor chip will be produced in Gujarat this year

 

National imperative

And for the same reason that China is seeking to dominate the market, and for India’s own long-term geopolitical security, “this whole fab is an asset which we absolutely need”, says Vishesh Rajaram, founding managing partner at Speciale Invest, a deep-tech focussed VC firm in Chennai.
 
Speciale has invested Mindgrove Technologies, a chip design startup in Chennai, and another as yet undisclosed venture. “We have great design capability, and ultimately, the ability to realise the chip tape-out has to be localised from purely faster product, and I think geopolitical reasons, and building India competencies,” Rajaram says. “Having this infrastructure positions you to be a partner for the globe, not just India.”
 
Tata Electronics’s plan to turn out 50,000 wafers per month is “not small, that's large”, from an India point of view. “It gives faster turnaround time, prototyping, go-to-market,” he says. And sectors such as power management chips for electric vehicles, automotive in general, telecom, and defence are “really fast-growing emerging sectors”. “Therefore, having captive capacity to solve for internal demand and then to also go global is meaningful.”
 

Missed opportunities

India’s efforts at producing hi-tech electronics date back at least to the 1960s. The debacle against China’s border aggression in late-1962 exposed how the lack of hi-tech defence electronics handicapped India’s armed forces.
 
The following year, India’s top technocrats, led by Homi Bhabha, came together to form an Electronics Committee. The committee’s report was submitted in 1966 by Vikram Sarabhai after the untimely death of Bhabha.
 
The report went into great detail in every relevant area and even envisioned a 10-year roadmap. Since then, a series of government decisions, as well as denial of tech from the richer countries, contributed to India falling behind. And after the first wave of liberalisation in the 90s, our pivot to IT services put paid to any further enthusiasm on the hardware front.
 
A precious handful of companies such as Tejas Networks in telecom, and chip design specialist Saankhya Labs, which in today’s lingo would be seen as a deep-tech startup, persevered—the former, now a part of Tata Group, and has also acquired the latter.
 
Around 2005-06, there were efforts again to build a fab in Hyderabad, Bangalore, or Chennai. Then again in 2012-13 there was much fanfare around how various consortia, including one led by IBM, would build fabs in India. None of those materialised.
 
Today, “the underlying rationale and constraints are still the same, and we need to answer all those things correctly”, says Madasamy. “But I like some of the aspects that are being addressed to make it happen. The jury's still out, but there're more favorable conditions now,” he adds.
 
Perhaps the obvious challenge is that semiconductor fabs are incredibly expensive to set up. Then there is the technology learning curve and the need for a good technology partner can’t be overstated. With PSMC, the Tatas have significantly shortened the learning curve.

Also listen: How the Tata wafer plant can transform India's semiconductor landscape
 

40nm sweet spot

“PSMC is a very respected old name. That is very encouraging,” says Shashwath T Ramkumar, co-founder and CEO of Mindgrove. And when the plant is ready, “we could be one of their customers”.
 
“That they are doing 40nm is very encouraging because it is one of those real nice workhorse nodes that we use for a lot of things,” Ramkumar says. And “the capacity they have chosen is actually slightly above what I would consider minimum capacity to actually breakeven or to make economic impact.”
 
Therefore, what has been done by the government, the Tatas, their tech partner, and other stakeholders, “they are not rushing into this, they have chosen the sweet spot for a lot of things, and it's just hats off to them for getting this far”, Ramkumar says.
 
The 40nm node is about 15 years old, 28nm is about 10 years old now—think cellphone from ten years ago as one application. Over the years, both nodes have been repurposed to a variety of applications including IoT, automotive, connected devices, smart appliances and so on. They can still be found in cellphones, but not just as the main processors.
 
While the Tata-PSMC fab won’t be targeting advanced chips in the foreseeable future, the other industrial and vertical segments that their mature nodes can sell into are all high-growth areas in India and around the world, Counterpoint’s Shah says. Each of those segments could represent a single-digit billion-dollar opportunity, he says.
 
For example, “once we have display driver microprocessors and integrated circuits, then we can also attract display fabs”, Shah says. Samsung already has one in India and TCL is planning to set up its own, he says. “Slowly the entire ecosystem could develop around different components which go into different applications.”
 
Therefore, while today’s projections for local demand are in the single-digit billions of dollars as total addressable market, “but if you look at global, it's multi-billion, and multi double-digit billion-dollar opportunities”, he says.

Madasamy echoes this thought process: “Ultimately, it has to make economic sense. And this time, the efforts are more thoughtful in that direction.”