Raja Manickam grew up in the township of BHEL, Trichy. As a kid, he was inspired by RC Bhargava, who was then the Director of BHEL Trichy, and later on went to become the founding Managing Director of Maruti Udyog. Raja loved physics in his school days. Did well in JEE and went to IIT Kharagpur, where he graduated in Electrical and Communication Engineering. He wanted to join IBM, which was the numero uno in recruiters in his time. He got interviewed there but they did not make him an offer.
However, he did get offers from Fairchild Semiconductors and Texas Instruments. The two had very different reputations. TI was a place where you lived a pressure cooker-ed life. In contrast, Fairchild had a country club atmosphere. Raja chose the hard path and joined TI. Most freshers aim to start off in the field of design. Next in the pecking order are fields like production and maintenance. At the bottom of the heap is testing and quality. Raja started off as a DRAM memory test engineer.
Testing is important in the industry as an IC is expected to have a life of 30 years and work reliably during that time. Yield is important and a couple of percentage points difference can lead to huge changes in market share. The biggest plus point of the TI job was the opportunity to work on the shop floor. TI was vertically integrated, they made their own wire bonders and testing machines. Those days, you even drafted the etching images by hand.
After a 4 year stint at TI, Raja decided to enjoy the country club at Fairchild, where he ended up heading the logic devices unit. Fairchild got bought over by National Semiconductor. His new bosses exiled him for penance in Penang, Malaysia. He headed the engineering centre over there. Penang taught him to understand Southeast Asian cultures and values. He became the telecom business head at National Semiconductor and headed the telecom business for them, interacting with the likes of C-DoT.
His next assignment got him to Silicon valley and a switch to sales. The entrepreneurship bug bit him in 2004. He had always wanted to come back and set up a manufacturing unit in India. However in India, in 2004, as it is today, it’s very difficult to set up anything in manufacturing. Raja zeroed in on services and along with two of his friends, founded www.tessolve.com, an engineering services company. Today Tessolve has a revenue of $ 100 m and has raised a lot of venture capital funding. In the two decades that Raja ran Tessolve, the business saw a lot of ups and downs. It came close to bankruptcy 3 times in these two decades. In 2016, a majority stake was divested to the Hero group. Veerapan and Srini, the other two co-founders, continued to work with Tessolve, but Raja moved on. The biggest achievement of Tessolve, as per Raja, was to build a test centre in Chennai. This showed the global semiconductor industry that chips could be tested in India.
The semiconductor industry is worth $ 500 billion today and is set to become a trillion dollar market in 10 years. The Chinese semiconductor market is already 30% of the United States market size. In about 10 years time it is expected to match US values. Their aim is to be self-sufficient. Raja feels that In the chip industry, no country, including Taiwan can be self sufficient. An average internal combustion engine premium car has $ 3000 worth of chips in it. For electric vehicles the proportion doubles. Incidentally, 50% of all auto IC packaging is done in Malaysia. Chip fabrication is at an inflection point. Alternatives to silicon are being explored. This is the time where a lot of original engineering thinking is required. Indian talent is what drives the cheap industry. 50% of the team at Qualcomm is of Indian origin.
After leaving Tessolve, Raja did some freelance consulting. He came to Tata’s notice during a McKinsey contract – and was soon invited to head Tata Electronics OSAT (OutSource Assemble and Test) project. 2 years atTata made him realise that the IAS and the TAS folks are not easy guys to work with. Decision making by committee was not the Raja style of work, Raja wanted to be in control and he wanted to move fast. The manufacturing keeda had still not left Raja. So he decided next to get his own chips to the market with https://www.ivpsemi.com/, albeit in a fab-less way.
Why fab-less? People who have built mega fabs are now figuring out how to use the capacity. Nvidia, recently ranked as the world’s most valuable company, is fabless. There are hundreds of chip manufacturing companies in Taiwan but few outside of it. There has been a gradual shift in the semiconductor market. Earlier, you had a few numbers of chips, but great volumes in each of these designs. The era of the super chip which can host all types of applications in all types of situations is over. We now see a large diversity of custom chips, each being produced at relatively low volumes. Custom chips are more cost effective.
After solar PV, most silicon is used to fabricate memory chips. Next in volume order is computing, analog, sensors and passives. Digital design today is very tool driven, so the fundamentals are being lost. He recommends that during the early years semiconductor engineers should have a career focus on analogue. Analogue required significantly low power.
Here’s some wikipedia gyan on analogue integrated circuits.The voltage and current in analog ICs vary continuously over time. In contrast, digital chips only assign meaning to voltages or currents at discrete levels. In addition to transistors, analog chips often include a larger number of passive elements like capacitors and resistors. Analog chips are required for wideband signals, high-power applications, and transducer interfaces. Research and industry in this specialty continues to grow and prosper. Some examples of long-lived and well-known analog chips are the 741 operational amplifier, and the 555 timer IC.
The cutting edge technology is the 2 to 3 nanometer chip which can be made in very few places. TSMC is the leader here. A fab for such a kind of nanometers cost 10 billion dollars to build. Silicon carbide could be the next big thing in semicon materials. It works well in high voltage ICs. It can be operated in a higher temperature range, Designing DC based appliances for solar applications could be the next wave. Silicon carbide will represent engineering challenges as it is a new material. With lab grown diamonds becoming cheaper, diamond based substrates, which are good for thermal management and conductivity, will also become more popular.
Wafers are being made in Japan, Taiwan and the US. A lot of the chemicals still come from China. In fact there are some gases used in deposition which are made only by the Chinese. Chemical availability is not a problem, purity is. The price of a wafer is proportional to the area of the wafer. 12 inch is a norm now. Nvidia chips are among the largest in the market. Only 3-5 Nvidia chips fit into a 12-in wafer. The future is going to be in distributed manufacturing and smaller fabs. Raja feels that a lot of these custom chips will not go for large wafer sizes. When Raja started his career 5 inch was the standard wafer size. The cost of lithography grows disproportionately with an increase in wafer radius. The moot question: do we really need large diameter wafers?
Raja’s crystal ball guessing tells him that 4-5 companies will dominate the global computing chip market.. It gets most of its chips made by the likes of TSMC. Incidentally Jensen Huang is also from Taiwan and is a close friend of Morris Chang, the TSMC founder. Morris spent his formative years at Texas instruments. TSMC is very tech heavy. A similar number will be the big gorillas in the memory space. But that still leaves a big market to be captured in spaces like photonics and sensors.
Raja;s latest company, iVP semiconductor makes components, chips and modules for the EV, renewables, and automation industry. In India, we are constrained in manufacturing because of our complex input export laws and the lack of availability of specialty chemicals. When one thinks of semiconductors, it is the ICs that come first to mind, but there’s also a lot of high volume non glamorous stuff like MOSFETs. The semiconductor success stories in India have been in simpler stuff – like LEDs and solar PVs. There is a lot of semiconductor that goes into building inverters for solar. Fast chargers. DC operated BLDC motors in air conditioners have cut down powered usage by 50%. Shortage of air conditioner today because of shortage of chips. BLDC motors need power electronics to drive them. A big growth area that Raja sees is going to be in solid state sensors.
Why Tata is supplying metal casing from Hosur to Apple is a big mystery. They are also into PCBs and anodizing. Tatas have to realise that they can’t do everything themselves. Instead of working as a supply chain partner, Tata’s would be better off doing basic manufacturing. The need to add silicon to their steel portfolio. Joint ventures with the Japanese and the Korean firms will help India. India will never be self-sufficient. No country can be. But IC technology in computing and memory is important for our security. We should make RISC5 and memory chips. MSMEs are going to play a very important role in the Indian semiconductor sector. Most people working in fabs tend to be migrants. The Philippines in Taiwan and the Indonesians in Malaysia. Manufacturing got affected during Covid because all of this work force went home. One idea that Raja has in mind is why not export labour from India to fabs in Taiwan and Malaysia adding to the diversity of the labour pool.
Even if we build a fab in India, Intel and ST micro will not come and use our facilities here. We will have to create our own demand. So semiconductor startups are important. The fun part is that in the smartphone that you are holding in your hand, 73% of chips are legacy: power management and boosters. These don’t require 3 nm chips. In some cases even microns are okay. The West is not interested in selling any chip where the cost is less than 5 dollars. So the Indian chip industry will need to figure out markets for sub $ 5 chips.
The Indian semiconductor market is about $ 15 billion. Indians will need to come up with frugal solutions in the semiconductors space. We have to make products which are 90% as good as Nvidia but at much lower costs. Fab requires a lot of water. Microprocessor manufacturing has about 200 steps. Washing is required at every step for removing chemicals. Today 90% of the water can be recycled if you can do the capex in water. In the US, fabs are coming up in places like Phoenix, Arizona which sits right in the middle of a desert. Imagine the air conditioning required for such a fab. But then it’s always politics first and economics later when it comes to manufacturing. In a level playing field India would have been better off than China. But what tilts the balance in China’s favour is subsidized capital. Interest rates for entrepreneurs can touch almost 0.
Most of the Indian semiconductor talent is from the southern part of the country. There are more chip designers in India than in the US. Today the entire design cycle is being completed in India. We are also developing the tools which are required in chip design. Except Taiwan there is no other country that has this advantage, not even China. The industry will be very low on employment.
Maybe more job creation can happen in the equipment supply for fabrication plants. Work has already started. There is an MSME in Coimbatore which is making deposition equipment and exporting to the US. 40% of its value add is in India. Semiconductors will never be run by the business types, it requires entrepreneurs with a tech mindset. A tinkering mindset is important, we need to know how to work with our hands. Indians will also have to become good team players to find success in this space.