It is especially sad to see a market you were leading and to see countries that you were comfortably ahead of 12 months ago pass you by. This is what has happened to India in 2026 twice during the same month. First Taiwan and now South Korea have left India behind to become the sixth largest equity market in the world. The statistics are stark: South Korea's total market capitalisation has jumped 86% this year to $5 trillion, while India's has fallen to $4.8 trillion. The turnaround would be bad enough in itself. But most of those claiming it as a surprise have completely misunderstood the unfolding saga.

This was not a market event! It was an industrial policy reckoning 30 years in the making.

In New Delhi, no one was ready for the hardware supercycle.

Driven by the strong demand for high-bandwidth memory chips that power AI servers and accelerators, SK Hynix shares have surged nearly 250% since the beginning of 2026. The company has established itself as a major Nvidia partner and is at the heart of the AI supply chain. Samsung had done the same two weeks prior to that milestone. Overall, SK Hynix is the third Asian company to value over $1 trillion, after Samsung and TSMC, which remain the most valuable company in Asia at more than $2 trillion.

It's not a story of a stock rally. It's a story of the reality of countries that invested 30 years in developing physical manufacturing capability for semiconductors, and then suddenly became the centre of the most capital-intensive technology transition in a generation. Altogether, Samsung, SK Hynix, and Micron are presently worth more than $3 trillion, and it's evident that memory chipmakers have become the heart of the AI economy. Why is that? It's because SK Hynix's HBM capacity for full-year 2026 is fully booked, and shortages are anticipated for 2027. With physical restrictions in supply, and all the major AI companies on earth bidding for the same chips, the power is all with the chip maker. There is no Indian manufacturer in the race.

Samsung has also stepped up its HBM production, with the first commercial HBM4 chips expected to be delivered in early 2026, with the company looking to catch up with the competition in high-value AI memory. The company is set to invest vast amounts of capital exceeding 110 trillion won ($73 billion) on AI chips in 2026 to strengthen its dominance in the field. That number ($73 billion in one year) is actually bigger than most countries' semiconductor pursuits. It's not as if South Korea's two giants are the only ones on the wave; they are the power behind the wave.

The decades-long divergence

What's being asked in the financial worlds these days is what happened to India? The truth is that they made different decisions not bad ones at the time, but decisions that created capacity in the wrong areas for this specific time.

The dominance of South Korea and Taiwan over semiconductor manufacturing is not a recent development. Both economies invested in decades in the development of export-oriented semiconductor industries with state support. Governments collaborated closely with domestic corporations, enabling companies to produce technologically sophisticated chips on an international scale. Those efforts eventually gave birth to a global winner like Samsung, SK Hynix and TSMC.

In contrast, India developed a powerful software and engineering services industry. The country provided the brains to design the chips which were made in Hsinchu and assembled in Seoul. It made sense, it was profitable, and for two decades it seemed to be the more sensible option, the less capital-intensive, the more often they could iterate, the more services their margins would improve. Indian engineers make around 20% of the world's semiconductor designers working at Qualcomm, Intel, AMD, Nvidia and MediaTek design centres in Bengaluru, Chennai and Hyderabad and Qualcomm has already completed 2nm tape-outs designed entirely in its Indian design centres, with fabrication at TSMC Taiwan.

Well that one sentence tells you all you need to know. In India they make some of the most sophisticated chips of the world. It then sends the blueprints to Taiwan where it has them made.

Indian technology companies are getting into the AI services and software development fray, but the largest winners in global markets are semiconductor manufacturers that provide the hardware that runs AI. The AI investment cycle is for physical infrastructure, not software architectures; fabrication plants, packaging lines, testing facilities. If Nvidia requires chips to be manufactured, it turns to TSMC. SK Hynix is data centre's Memory Partner. That call list doesn't include India, and that’s reflected in the stock market in real time.

The capital flight text

That the market-cap ranking change were only a relative story would make it easier to swallow. It is not. In 2026, FPIs had a record outflow of ₹2.2 lakh crore. India's near-term earnings growth will be below that of the commodity and tech-heavy emerging markets, and India in its entirety has not yet been an obviously big beneficiary of the AI and semiconductor cycle, which has been propelling global capital allocation, Kotak Institutional Equities strategists stated.

The foreign portfolio investors are not wagering on the Indian economy. They're wagering that in today's market cycle, the high returns are coming from the companies building the actual hardware of AI. India's weightage in financials, consumer and service companies, which gain from GDP growth and domestic consumption, is quite high, and does not reflect the AI infrastructure premium. The Nifty 50 could finish 2026 lower, its first annual loss since 2015, a Reuters poll recently suggested. It has been exacerbated by the rupee, which has been trading near ₹95 which means foreign investors' returns will be further hit, and India's equity story will become less appealing in dollar terms.

On the macro level, it's a structural change with regard to the allocation of global capital in tech cycles. Great technology has been associated with previous leadership cycles. Japan was the leader in electronics manufacturing, the USA was the leader in the internet revolution and China was the leader in infrastructure and manufacturing. Today's cycle is being driven by AI hardware. South Korea and Taiwan are the beneficiaries as they are located at the epicenter of semiconductor manufacturing, while India is more vulnerable to software services and local consumption.

Can late-mover status be cured?

The truth is partly and not fast. On 16 May 2026, Tata Electronics and ASML, Netherlands, inked a strategic deal to aid in establishing India's first front-end semiconductor manufacturing plant in Gujarat. As part of the deal, ASML will supply cutting-edge technology and equipment for Tata Electronics' proposed 300mm wafer fabrication facility under construction in Dholera, which reportedly will produce chips for automotive electronics, mobile devices, industrial applications, and AI, and is estimated to cost $11 billion.

The Union Budget 2026-27 has earmarked ₹8,000 crore for the semiconductor mission the biggest ever in a single year since the mission's inception, along with the unveiling of ISM 2.0, which has a semiconductor equipment and material manufacturing, advanced design and indigenous IP focus. The 13 approved semiconductor projects are now up and running or underway throughout the country.

But the market is rightly discounting these aspirations quite a bit. The involvement of India's partners Taiwan's TSMC and South Korea's Samsung is still not expected. Even if the Dholera fab is on schedule, the facility will make chips at legacy process nodes, not the bleeding-edge market for AI accelerators that sees the valuation premiums, serving automotive, consumer and industrial markets. Those projects will likely take years to significantly affect the benchmark equity indices or make semiconductor giants like Samsung or TSMC, analysts warn. The market is simply paying off the investments of South Korea and Taiwan, which have been made over decades.

Fabrication capacity isn't something that can be rushed. Thirty years and the whole support of the Taiwanese state have propelled TSMC to its present position. Samsung's memory advantage was established over a series of booms and busts, investing when prices were low, taking losses when other manufacturers cut back, and coming out on top each time, with larger market shares. That is where India is starting that journey, and those numbers are trillions of dollars and tens of billions of dollars every year being reinvested by the incumbents to stay ahead of the curve.

In essence, what India has to offer.

All this is not to say that India's market is broken. This has been a bad time to be holding stocks in India and anyone who says otherwise had better be honest and say so, instead of pretending it's a minor squall.

India continues to be one of the fastest growing major economies in the world. Its stock exchange is far more diversified than many AI exchanges, having a decent percentage of financials, consumer companies, industrials and services. It is India's very diversity that renders it less attractive to AI-capital allocators, but also makes it more resilient when semiconductor cycles turn. They will turn as they always do.

The underlying structural investment thesis in India holds: a young population, growing consumption, financial inclusion and a pool of engineers that the world relies on. That 20% of the global chip designers is no consolation prize, it's a real asset that if combined with fabrication capacity over the next ten years could bring about something lasting. The investment in Dholera, the partnership with ASML, the design start-up ecosystem those are the keystones of an answer, although the answer may be 15 years away.

It is not a bad thing that we are made uncomfortable by South Korea and Taiwan, because it's a good thing. It exposes a hole that Indian industrial policy has left for the last three decades: that software services, no matter how good, aren't as valuable to the equity markets as hardware manufacturing during a technology supercycle. That's just been driven home by the AI craze and quantified in market cap numbers that are no longer deniable.

This month, the race is not over. When the policy was decided, implicitly, incrementally, over the course of 20 years, that India would be the world's chip designer and not the chip maker, it was lost. That was an appropriate decision then. In a world where the next trillion dollars of market value is being generated in a fabrication plant in Hsinchu and not a software campus in Bengaluru, it simply does not work.