Vinod Dham was was 21 when he joined Intel, then a USD1 million startup. Dham helped scale up Intel into a multi-billion-dollar enterprise and earned himself the title ‘father of Pentium’.
He led the group of Intel engineers that developed the flagship microprocessor or logic chip. Processors are ubiquitous, found in an array of products, from household appliances, such as microwaves and dishwashers, to cars and missiles, etc. “If the industrial era was defined by motors, the modern, digital era has a semiconductor or chip in everything,” says Dham.
However, for decades now, India has been trying to find partners to build a facility for large-scale chipmaking. Dham, who is among the advisors to the government’s India Semiconductor Mission, aimed at kick-starting chip manufacturing, doesn’t mince words. “We have to succeed, otherwise we become a secondary nation in the world,” he says.
Based in Silicon Valley, Dham was in India to attend Semicon India 2022, a conclave in Bengaluru aimed at catalysing India’s chipmaking ecosystem. He isn’t worried about Covid-19 — “I am four-times vaccinated with Moderna shots” — but it’s the unusually hot summer here that bothers him. At Semicon India, chipmaking consortium ISMC inked an MoU with the Karnataka government to set up a USD3 billion fab in the state. The Centre is evaluating at least four other proposals for chipmaking in India.
Dham’s strategy includes upgrading a semiconductor facility in Mohali; working with partners with 15 to 20 years’ experience in building fabs; and go for products that have a sizeable market. Even if India gets its act together now, it will take five to 10 years to get the entire ecosystem up and running, Dham tells ET Prime.
In a wide-ranging interview in Gurugram, Dham, 71, discusses India’s attempts at building a fab ecosystem; the strategy and the risks; why Intel slowed down; and more. Edited excerpts:
Much before the current noise around fabs, there was much excitement around fabless. It never took off. What happened?
You can’t exist in a vacuum. The early efforts were driven by individuals who came mostly from abroad. They had connections with some ministers or local administrations. Some of their efforts were genuine attempts. But they didn’t have all the support that was needed. That’s why you haven’t heard of successful startups in semiconductors as you have heard in, say, software.
In the context of the shift caused by the pandemic, the Russia-Ukraine war, and the changing geopolitics, how critical is a fab for India?
If there’s anything that’s most critical, it is a chip. It has become ubiquitous-- just like motors became ubiquitous in the industrial era. Even the hairdryer has a motor. Similarly, chips are now in everything. Cars and smart homes have plenty of chips. They are all over – in water meters, electric meters, dishwashers, microwaves.
Do you think a chip factory will come up [in India]? It’s a massive capital commitment.
Fab is a serious commitment and needs government support. This time there’s a serious effort by the government not only on fab but on the entire ecosystem [so as to] to create local capability.
It doesn’t have to be the fastest node or processor. We’re not competing with Intel, AMD, Qualcomm, or Nvidia. We are building an atmanirbhar, self-sustaining ecosystem, particularly after Covid-19, the Russia-Ukraine war, and the changing geopolitics.
Even private equity, hedge funds, and other sources of money can come into play, but they first want to see: Is it real? Are they really serious? Are they really going to do it? Once we cross that hurdle, money won’t be a major issue.
Do you think it’s the right strategy for India to go for a full-scale 28 nanometre fab?
It will take five to 10 years to create the ecosystem. To start with, we have to build capability for, say, chips like IoT sensors and not get into building a leading-edge microprocessor with the highest speed in the world. We have to make chips for more practical uses, for the local domestic market, for connected cars, smart speakers, LEDs, etc.
The fab at the Semiconductor Laboratory (SCL) in Mohali has the basic infrastructure in place, but it needs to be beefed up. Mohali is 180nm (nanometre), it can go 130nm or 90nm. That’s quite a sweet spot.
We can also start setting up some ‘assembly, test, mark, and package (ATMP)’ type of facilities, which don't require leading-edge capabilities or massive capital. Within a few years, we should have at least the process for building the first fab 28nm or 20nm fab.
So you recommend upgrading the Mohali fab plant and focus on the IoT, sensors and those kinds of micro controllers? And then gradually build the system?
Yes. Nothing will happen overnight. In three years from now, my hope would be that a lot of these [upgrading the Mohali fab, sensors etc.] will be in place — like sensors which go into smart bulbs, smart speakers will be locally sourced.
I see no reason why India won't succeed. In fact, we have no choice. We have to succeed, otherwise we become a secondary nation in the world.
The one good thing about coming in late is, if we had started 50 years ago, every three years, we would have had to spend multi-billion dollars to set up a new node or process. Now, India is not far from where everybody else is going to end up on the treadmill.
What percentage of the market does the 28nm chip account for?
I would say 25%-30%. The leading edge is a smaller end of the market. But it’s richer, the margins are higher, and it’s expensive [10nm needs around USD8 billion investment, while 22nm-28nm needs USD5 billion]. We need to stay in the older generation, fulfil our strategic needs, meet local domestic needs, learn, build a foundation, and then develop on that.
How many years will this journey take?
I'll give you an example. In 2001, I visited SMIC — Semiconductor Manufacturing International Corporation — a fab in Shanghai, China. They were just starting up then. It’s only in the last few years that they have scaled up, and they actually do a lot of older-generation products — 65nm, 90nm.
It does take a decade or so to get to know how to build a cutting-edge fab, which process to adopt, and the tech to build on.
Is there an opportunity in the semiconductor tools space as well?
There are two types of tools — one in which our boys and girls are very good — and that’s design tools, called computer-aided design (CAD). There are two big companies in the world Cadence and Synopsys — both are in India. Nvidia, AMD, Intel, Qualcomm, Texas Instruments — all have a few thousand engineers, mostly in Bengaluru, designing chips. They collaborate with the CAD players, using their platforms to create designs.
The other type of tools, which we don’t have and only few companies in the US and the Netherlands have, like ASML, are lithography tools [used to help pack transistors on computer chips]. In India, we want to get things started first and then see if we can get to a stage, maybe a decade from now, and do some of those very complex tools.
India launched a USD10 billion incentive package in December to attract chipmakers. Is this good enough?
I think it’s attractive and adequate to get the ball rolling. USD10 billion dollars is on the chip side and there’s another USD10 billion for incentivising startups and setting up displays, power electronics, and so on. Sensors are going to be a huge market across the world. We will be talking about a dust of sensors all around — in the streets, walkways, everywhere.
A dust of sensors! That’ll be quite a scary thing going all around.
If we want to get to a point where we want to save humanity, you have to stop wasting 80%-90% of everything around us. Like a fan need not run at a fixed speed. If it senses your sweat and your body temperature and has the AI built in, it will continue to monitor itself and adjust speed based on the surroundings. That’s why having sensors around will make life so much easier.
You will build sensors into the walls and won’t see them. Like in the wall paint working wirelessly, capable of sensing and connecting to devices.
Fab is fairly complex, needs big investment, massive power, and water resources. Given all that, will global companies be interested in setting up a fab in India?
The semiconductor industry’s TAM [total addressable market] will be about USD1 trillion in 10 years. Now, it’s about USD0.5 trillion. It’s a huge market. So, will we just let a few companies in the US and Taiwan do that or will India also play a role at whatever level it can? Besides, the Indian market has sufficient scale now. I think the right thing is to go participate.
What are the risks?
You need to be careful about what to fill it with. You need a high-volume product, otherwise the fab can become a white elephant. If you don’t run it 24/7 365 days a year, the cost of what you make will be 10 times more than the guy next door. Nobody will buy it. That’s a big risk.
So in the Indian context, what could that product category be?
Memory chips, microcontrollers, sensors…. From a device point of view, why does India have to buy an iPhone from America or other smartphones from China? Why does Xiaomi, a Chinese company, dominate India’s cellphone market? Look at the US. The who’s who of American tech companies are run by Indians, and you can’t make a cell phone here. Seriously? I think there’s something fundamentally wrong. Earlier it was argued that there isn’t enough market or scale here. But now the scale is there.
Having missed out on the last 50 years, can India leapfrog now?
India needs to partner with someone. Someone who has done it 15-20 times. So they know how to do it right and help create the local ecosystem. There are several Indian-origin engineers in this field in the US who have done it in the last few decades, and they can also help out.
The world is very different now, and India is very different. I hope it doesn’t deviate from its vision. They have the template of what China has done in the last few decades. I don’t see a scenario where something will blow up and fall into pieces.
In the last 18-24 months, over USD100 billion fresh investment has been announced by companies like TSMC, Texas Instruments, Nvidia, Intel and others, but none in India.
India announced its policy in December. Global companies are not waiting for India to announce the policy and jump on a ship and come here. It doesn't work that way. India doesn’t have the infrastructure. So India won’t be the first obvious choice. We have to convince them [global chipmakers] as to why they should come here. That’s why we’re throwing so much incentive at them.
When you joined Intel, it was a USD1 million startup and later with Pentium and the x86 architectures, it became a multi-billion chipmaker that set industry benchmarks. Today, TSMC and Nvidia are the big players. Has Intel lost out?
The previous management took its eyes off the ball. The new management is trying to regain lost ground. But in this industry, if you miss out, not once but twice, it becomes much harder to climb back to the position that you were in.
Do you think Intel got a bit too complacent?
I think the leadership was defocused. They were getting into many other areas. For a spell, there was also a finance person running Intel! You don't expect a very high-tech company to be run by a finance person [Bob Swan].
Prior to him, the leadership [Paul Otellini, who had a non-tech background, and Brian M Krzanich] started moving into different areas of electronics — virtual reality, augmented reality, and such things. That’s okay if your house is in order. There’s been a series of such missteps. Now they are trying to set it right.
Read More at https://economictimes.indiatimes.com/prime/technology-and-startups/india-must-succeed-in-its-chipmaking-mission-to-stay-ahead-in-the-world-order-intel-icon-vinod-dham/primearticleshow/91322426.cms
Comentarios