There is a number making the rounds in global fund management circles that should unsettle every Indian equity strategist. Just 229 companies now account for 48% of all listed equity value on this planet. Thirteen of them are worth more than a trillion dollars each. The club is not growing. It is consolidating. And India, with its 1.4 billion people and decade-long capital markets deepening story, is not in it.
The AI cycle has done what no policy document or ratings upgrade ever could: it has redrawn the map of Asian capital.
Cast your mind back to October 2020. China sat at 43.2% of the MSCI Emerging Markets index. India was the rising star, touching 20% briefly in mid-2024. By April 2026, the picture looks almost unrecognisable. Taiwan: 24.8%. Korea: 18.7%. China: 23%. India: 11.9%.
Taiwan and Korea now outweigh China and India combined.
That did not happen because of demographics or diplomatic pivots. It happened because Taiwan makes the chips the world cannot build AI without, and Korea makes the memory that runs it. One theme, two countries, and everything else got swept aside.
The Return Numbers Tell You Everything
If you need proof that capital follows one story at a time, look at twelve-month local market returns. Korea’s KOSPI is up 191%. Taiwan is up 99%. The Nikkei is up 67%. The S&P 500 is up 31%. And India? The Sensex is down 3%. The Nifty, down 0.4%.
In USD total return terms, the humiliation is sharper. KOSPI delivered 73%. Taiwan delivered 42%. India’s Nifty was down 12.2%.
Nifty was down 54% relative to KOSPI in dollar terms over twelve months. Not 54 basis points. Fifty-four percent.
India did not underperform because something broke. It underperformed because it had nothing to offer the trade that dominated 2025 and 2026. No advanced semiconductor ecosystem, no AI memory exports, no data centre supply chain. When the world’s allocators were building exposure to the AI hardware stack, India was simply not in the conversation.
Two Countries Are Carrying an Entire Region
Jefferies data on MSCI Asia Pacific ex-Japan is worth sitting with. The index is forecast to deliver 53% earnings growth in 2026. That number sounds like a regional boom. It is not. Decompose it, and Korea alone contributes 38.2 percentage points of that 53. Taiwan adds 6.1. China 4.2. India 1.7.
Strip Korea and Taiwan out entirely, and regional earnings growth falls to 8.8%.
The rest of Asia, including India, China, ASEAN, and Australia, is collectively growing at roughly the pace of a mid-cycle developed market. The Asian emerging market premium is almost entirely a two-country story. And those two countries are called Korea and Taiwan.
The US is experiencing the same compression. S&P 500 EPS estimates for 2026 have been upgraded by 6% in the last three months. But remove AI and commodities, and the revision was 0.3%. The rest of the index is growing at 11%. The overall index, which includes the AI names, is growing at 19%. The gap is not a rounding error. It is the entire thesis.
Memory Is Not Boring Anymore
Everyone knows SK Hynix. The Korean memory giant became the symbol of the AI hardware rally. But here is what most people missed: even SK Hynix was not the best performer in Asia’s mega cap universe over the past year.
Kioxia Holdings, Japan’s pure-play NAND flash company, returned 1,797% in twelve months. SK Square, the holding company that sits above SK Hynix, returned 1,069%. Zhongji Innolight-A, a Chinese optical connectivity supplier powering 800G data centre links, returned 897%.
Since September 2025, SanDisk is up 2,658%. Kioxia is up 1,906%.
Why NAND? Because the market is working through a realisation that took longer than it should have. Agentic AI is not purely a compute story. Inference requires memory. Agents that run continuously, remember context, and act autonomously require persistent storage at a scale that nobody had modelled for. NAND is no longer the boring infrastructure at the back of the server rack. It has become strategic inventory. And the world does not have enough of it.
The PEG Paradox Nobody Is Talking About
AI stocks look expensive. That is the received wisdom. On a headline price-to-earnings basis, the numbers are large and visible. But on a PEG basis, comparing what you pay against what you get in growth, AI is actually the cheapest sector in the S&P 500, sitting at just 0.6x. Consumer Staples trade at 3.3x. Healthcare at 1.5x. Industrials at 2.0x. The broad S&P 500 sits at 1.1x.
Jefferies estimates that AI is driving close to 80% of S&P 500 year-to-date returns. Their AI basket is already up 30%. And it is still the cheapest growth in the building.
This is the paradox the consensus keeps missing. Something can look optically expensive and still be fundamentally cheap, if the growth underneath it is large enough and fast enough. On current numbers, AI passes that test comfortably.
When Two Companies Move a Nation’s Tax Base
The Seoul story has moved well beyond markets. Samsung and SK Hynix’s combined income tax expense is projected to rise from 11.8 trillion won in 2025 to 146 trillion won in 2026, and potentially 211 trillion won in 2027. Total Korean corporate tax revenue across the entire economy last year was 84.6 trillion won.
Two companies are heading toward paying nearly two and a half times what every Korean corporation combined paid twelve months ago.
That is not a stock market story. That is a national fiscal event. The Korean government is about to become structurally dependent on the earnings of two semiconductor companies.
Korean retail investors have noticed, and then some. Margin loans are at record highs. The retail chase of Samsung and Hynix has the texture of a momentum-fuelled frenzy, the kind that always looks rational right up until it does not.
The Cracks Worth Watching
Samsung’s labour situation is deteriorating. Strike action is being threatened, talks have collapsed, and the company is reportedly already cutting production ahead of the planned action. Here is the telling detail: Samsung’s share price went up on the news. The market interpreted a potential strike as a supply-tightening event and bought it.
When a labour crisis is read as bullish by equity markets, the momentum has become detached from fundamentals.
On the supply side, China is moving fast. CXMT, the country’s leading DRAM maker, is planning a Shanghai IPO in Q3 2026 at approximately US$4.2 billion. YMTC, the leading NAND flash producer, is reportedly considering a domestic listing in Q4 at a valuation between 200 and 300 billion renminbi. Chinese integrated circuit exports in April surged nearly 100% year-on-year to a record US$31.1 billion. Year-to-date, chip exports are up 83.7%.
The Chinese are coming. They always do, eventually.
India’s Turn Is Being Built in the Silence
India sat out this cycle. That is simply true, and there is no version of the numbers that says otherwise. But the reason it sat out is structural, not fundamental. India has no semiconductor export base, no NAND flash champions, no optical interconnect suppliers feeding data centres in Virginia and Singapore. The AI hardware cycle had no natural hook into the Indian economy.
What India does have is something that chip capacity cannot replicate on a five-year horizon: a domestic economy still deepening its financial system, a manufacturing thesis gathering genuine momentum under the production-linked incentive framework, a demographic profile decades from peak consumption, and a capital markets infrastructure that is steadily becoming more accessible to global institutional allocators.
The markets being left behind today are not broken. They are simply the anti-trade.
And when the memory cycle consolidates, as every cycle does, the rotation back toward economies with durable structural stories is likely to be rapid and disorderly. The biggest money in markets is rarely made by chasing the current extreme. It is made by identifying, patiently and early, what comes after it.
India’s moment in this story is not this quarter. But it is being set up quietly, precisely because it has been so thoroughly overlooked.
Extremes always rotate. The question is whether you are positioned before or after.
