India needs the ‘fabs’ for its own long-term interest

Recent headlines and self-congratulatory government advertisements have been gushing about India setting up a semiconductor plant (or fab) at Dholera in Gujarat. A fab has been New Delhi’s dream for a while now. What most headlines don’t talk about, and what the ads hide, say critics, are that by the time these fabs get operational, the chips (semiconductors) churned out will be ‘old technology’. Their case is that by the time this facility goes operational, it will begin by producing 28 nanometer (nm) chips. And by the time it acquires the capabilities to produce 26 nm chips, the world would have moved on, and the chips Dholera produces might be obsolete. It is pertinent then to investigate if these critics have a point.

Chip technology is like fashion—trends move fast. Right now, major chipmakers such as Samsung and TSMC make 5 nm chips. By 2026, cutting edge will likely be 3 nm or even 1 nm. To get a sense of what that means, the strand of a human hair is roughly 75,000 nm thick and a Red Blood Cell is 8,000 nm in diameter.

It is on such smaller and sleeker chips that the world’s most innovative companies pack the technology to handle complex tasks at lighting speed. So, on a fashion ramp, India’s 28 nm chips, by comparison, would be like a bulky winter coats that are a few seasons behind—functional at best, but not the kinds which can set the ramp on fire. This, most critics believe, is the dampener on India’s ambitious plan to establish domestic semiconductor “fabs”.

Add to this the breakneck speed at which this industry moves. This is why the alternative doing the rounds is that it makes sense for India—it is to bypass investing in building the ‘foundry’ or semiconductor ecosystem and morph to become the chip designing hub of the world.

As a New York-based investor of Indian origin pointed out, “Over 80% of the world’s chip designers are of Indian origin.” What makes this case stronger still are the number of engineering graduates India creates every year which is in excess of a staggering 1.5 million students graduate every year with a majority of them having studied computer sciences and electronics. Not all of them get placed, but if chip designs are going the way they are, there is a case for what the New York investor says.

But let’s get pragmatic here: Semiconductors are the new oil, and there is a geopolitical battle being fought over who controls it. What we also know is that the Americans are wooing India to protect their strategic interests. Their relationships with China has been going south and India is a good hedge. This takes us to a question this column had asked last year: why doesn’t India have its own fabs? It is in India’s geopolitical interests in the longer term to start work on that.

To do that, you’ve got to start someplace. Starting with established technology allows for faster learning and avoids the risks of cutting-edge ventures. It’s like training wheels on a bicycle – a necessary step before tackling high-performance models. “Yes. We had to start somewhere and frankly expecting India’s semiconductor value chain to directly start from the hi-tech 10 nm and below would be unrealistic,” says Navkendar Singh, a technology analyst at the research firm International Data Corporation (IDC). So, what about these out-dated chips? Where does that go? “The legacy technology and many non-smartphone categories such as automobiles, TVs, displays, memory and other similar technologies still use higher nm silicon,” Singh says. He doesn’t see that changing for a while.

When looked at from this perspective, the ‘training wheels’ phase makes sense and falls into place. But the world of technology is unforgiving – competitors won’t wait for India to catch up. The East Asian giants Taiwan and South Korea are constantly innovating and pushing the boundaries of miniaturization. And now that India has decided to get into the game, it might as well prep for a marathon, run the race, and shut the critics out. India needs the fabs in its longer term interests.