The Decoupling
In 2007, a popular idea was doing the rounds in markets. It was called decoupling.
The pitch went something like this. Sure, the US economy might slow down. But countries like India and China had their own engines of growth. So they would "largely remain unaffected" even if America caught a cold.
And for a while, it looked right. A study of that period found emerging markets were "somewhat insulated" from US turmoil right up to the summer of 2008.
Then Lehman Brothers collapsed in September 2008, and emerging markets suddenly "recoupled". The Sensex, above 21,000 in January 2008, fell below 8,000 by October. Foreign investors pulled out about $13 billion that year.
Why bring this up now? Because a very similar idea is doing the rounds today.
Let's imagine you run money for a big global fund in 2026.
For three years, you've ridden the AI boom: chipmakers in Taiwan and Korea, the American tech giants, all of it. And it's worked. The Nasdaq rose about 43% in 2023, 29% in 2024 and 20% in 2025 (Wikipedia).
But lately, you've been sleeping a little less. Chip stocks are roughly 20% below their June peak. And people keep asking awkward questions about how all those data centres will be paid for.
So you do what any careful person does before the monsoon. You buy an umbrella.
In investing, that umbrella is called a hedge. It's something you own not because you expect it to shine, but because it should hold up when your main bet gets soaked.
And for a lot of fund managers this year, the umbrella has a name.
India.
In September, JPMorgan called India's stock market "the world's largest, liquid AI hedge". It even reckoned that if global funds simply brought their India holdings back up to normal, around $115 billion could flow in.
Sounds reassuring, right?
Well, we went through a pile of official data from the RBI, the NSE and the statistics ministry to check. And this umbrella seems to have a few more holes than the brochure suggests.
To be fair, the logic behind the umbrella is pretty simple.
India doesn't have a TSMC or a Samsung. Its main stock index, the Nifty 50, is mostly banks, oil, telecom and cars. Financial companies alone make up about 37% of it (NSE). IT companies, the closest thing India has to an AI play, are just 7.5%.
So if AI stocks crash, the thinking goes, India shouldn't crash with them.
There's a second reason. India has its own money. Every month, millions of Indians put a fixed sum into mutual funds through SIPs (systematic investment plans). In September, SIPs brought in a record ₹32,490 crore. And Indian institutions like mutual funds and insurers now own 19.5% of NSE-listed companies, more than foreign investors' 15.1%.
In other words, even if foreigners head for the exit, there's a home crowd walking in.
And third, India is big. It makes up around 12% of the main emerging markets index. Most other markets that aren't AI-heavy, like Southeast Asia or Latin America, are too small to take in serious money.
Big, liquid, light on AI, and backed by a home crowd. Case closed, right?
Not quite.
The first thing you'd expect from an umbrella is that it keeps you dry. So let's see how this one has done in 2026.
The Nifty 50 fell 13.4% between January and September. Over roughly the same stretch, the MSCI Emerging Markets index, a basket of stocks from places like China, Korea, Taiwan and Brazil, rose about 18%.
And if you're a foreign investor, it's worse. You also lose when the rupee weakens, because your rupees buy fewer dollars when you take them home. The rupee is down more than 7% this year, and it hit a record low of 96.97 to the dollar in May. Put the two together, and by our rough math, the Nifty is down about 20% in dollar terms.
Foreign investors have noticed. They've pulled a record ₹3.12 lakh crore out of Indian stocks this year. That's nearly double what they took out in all of 2025.
Now, we know you'll say, hey, the AI trade has been booming all year. Of course India looks bad next to it. An umbrella isn't supposed to win on a sunny day.
Fair point. So let's ask a different question. What actually moved India this year?
The answer seems to lie in three things that have little to do with AI. Oil, American interest rates and the dollar.
On February 28, US and Israeli strikes on Iran effectively shut the Strait of Hormuz, a narrow stretch of sea that more than half of India's crude imports pass through. Brent crude jumped from $72 to about $100 a barrel within days.
And India imports nearly 90% of the crude oil it uses. So when oil gets pricier, India has to spend a lot more dollars to buy it. JPMorgan estimates that every $10 rise in the oil price widens India's current account deficit by about 0.5% of GDP. That deficit is, roughly, the gap between what India earns from the rest of the world and what it spends there.
Meanwhile, the yield on 10-year US government bonds went from about 3.95% in February to above 5% by October. When the safest bonds in the world pay over 5%, a lot of money doesn't feel the need to take a chance on emerging markets.
So yeah, nearly every big wave of foreign selling this year has been blamed on some mix of crude, the dollar and US yields. Not AI.
That's the first hole in the umbrella. You thought you were buying protection against AI. What you actually bought was a bet on oil and the rupee.
But here's the interesting part.
There was a stretch this year when the umbrella did seem to work. In July and August, foreign investors finally turned buyers, putting ₹20,200 crore and ₹29,631 crore into Indian stocks. The rupee steadied. And India's forex reserves hit a record $785.7 billion on September 4.
It was in mid-September, with that calm still in place, that JPMorgan made its "AI hedge" call.
So what was keeping things so calm? Well, a big part of it seems to have come from the RBI.
To understand this, we need a little context first.
By June, the rupee was under real pressure. So the RBI opened a special window. Between June 8 and September 30, banks could collect fresh dollar deposits from NRIs (non-resident Indians) for three to five years. They could then swap those dollars with the RBI, which took on the cost of protecting them against a falling rupee. That let banks offer NRIs up to about 7% on dollar deposits. That's a lot more than a dollar savings account pays in the US.
Think of it this way. Suppose a shop is struggling to bring in customers. So the owner starts offering a generous discount, paid for out of his own pocket. Sure enough, the shop fills up. But you wouldn't look at that crowd and conclude the shop is thriving on its own.
And this crowd was huge. According to the RBI's own numbers, NRI deposits brought in a net $119.2 billion between April and August this year. In the same months last year? Just $5.6 billion. In August alone, it was $83 billion.
Now, India's balance of payments is basically its ledger of all the money flowing in and out of the country. On paper, it showed a healthy $58 billion surplus for April to August. Take out those NRI deposits, and it turns into a deficit of about $61 billion.
In other words, the calm that made India look like a safe umbrella was, at least partly, borrowed.
And as the window wound down, foreign investors went back to selling. They pulled out ₹35,861 crore in September and another ₹52,100 crore in the first 17 days of October. Reserves fell about $60 billion in five weeks as the RBI sold dollars to defend the rupee. Even the RBI's first rate hike in nearly four years did little to stop the slide.
And there's a bill still to come. Japanese brokerage Nomura reckons the RBI's June measures have left it with about $200 billion of future dollar commitments. Those NRI deposits come due between 2029 and 2031.
But there's one more thing that could keep the umbrella from opening.
Say that $115 billion really does start flowing in. Who's selling to it?
As it turns out, plenty of people. Between April and September this year, Indian companies and their owners raised a record ₹2.43 lakh crore from the stock market through IPOs, share sales and placements. Only 47% of that was fresh money for the companies themselves. The rest was mostly existing owners, like promoters and private equity funds, cashing out.
And there's plenty more in the queue. Another ₹4.65 lakh crore of IPOs is either approved or waiting for approval.
Think of it like a mela. If more buyers show up but the stallholders keep wheeling out fresh stock, prices don't rise much. By our rough math, a year of selling at this pace would soak up about 45% of JPMorgan's $115 billion.
Okay, but some people make a deeper version of the argument.
It goes like this. AI tools are taking work that used to go to Indian IT companies. If the AI bubble bursts, companies will go back to hiring people to write and maintain their code. And who has the most of those people? India.
It's a neat theory. Except history suggests busts don't quite work that way.
Let's go back to the late 1990s. During the dot-com boom, companies laid far more fibre-optic cable than anyone needed. When the bubble burst, a lot of it sat dark and unused. And that cheap, overbuilt bandwidth is a big part of what made it possible to send work across the world to Bengaluru and Pune. Columnist Thomas Friedman called that overinvestment "a gift that keeps on giving".
In other words, the last tech bust left behind something cheap. And that cheap thing helped India.
Now ask yourself, what would an AI bust leave behind? Lots of data centres and chips that nobody wants to pay full price for.
We've already seen a preview. Renting an Nvidia H100, the workhorse AI chip, cost $7–10 an hour at its peak. By late 2025, as supply overshot demand, it was down to $2–4. And research group Epoch AI found that the price of a given level of AI performance falls anywhere from 9 to 900 times a year.
So an AI bust probably wouldn't make AI expensive. It would make it cheaper. And cheaper AI coding is competition for Indian IT, not a gift.
But there's one more problem. When the US economy stumbles, companies usually cut their tech budgets before they move them anywhere. In April–June 2009, after the global financial crisis, India's software exports fell 11.5% from a year earlier. And the US still buys about 54% of India's software exports.
And even if some of that work does come back to India, it may not go where you'd think.
A growing share of it goes to GCCs (global capability centres). These are the in-house India offices that foreign banks, retailers and carmakers set up for themselves. Their revenue jumped from $64.6 billion in FY24 to $98.4 billion in FY26, roughly 23% a year by our math. Meanwhile, TCS, India's biggest listed IT company, grew revenue just 2.8% year on year in its latest quarter.
Here's the catch. GCCs belong to their foreign parents, so you can't buy them on the NSE. The work might come back to India without ever coming back to Indian stocks.
And here's the irony. India isn't that light on AI any more. Reliance has committed about $110 billion over seven years to giant AI data centres. Adani has pledged $100 billion. Google, Microsoft and Amazon have together promised about $67.5 billion. Even TCS now gets $3.1 billion a year from AI work, over 10% of its revenue.
And the foreign "hedge" money? It's been buying AI plays too. Half of the ten best performers in the Nifty 500 this year are indirect AI bets, like optical-fibre maker HFCL and generator maker TD Power Systems, with gains of 123% to 287%.
So if big tech cuts its AI spending, India's power, cable and data-centre stocks are right in the firing line. That's an odd thing for a hedge to hold.
Put all of this together and you notice something odd.
India actually pays for its oil with code. In FY26, the country bought $337 billion more goods from the world than it sold, and $120 billion of that gap was oil. What filled most of the hole? About $217 billion of net services exports, led by IT, and $144 billion sent home by Indians working abroad. More and more of that money now comes from white-collar Indians in the US and the UK.
So think about the two extremes.
If AI works brilliantly, cheaper code squeezes India's IT earnings and white-collar jobs, at home and abroad. That's the very money paying for the oil. If AI busts badly and drags America into a recession, foreign money flees emerging markets, just like it did in 2008.
India seems to do best somewhere in the middle, when AI simply muddles along. But a real hedge is supposed to pay off in the extremes. India looks like it could lose in both.
Still, one year is just one year. So we went back and checked history.
Since 2000, the Nasdaq, America's tech-heavy index, has had five really bad years, when it fell 15% or more: 2000, 2001, 2002, 2008 and 2022. In those years, the Nifty did better than the Nasdaq four times out of five (Wikipedia).
That's the hedge case, and it's real.
But doing better than a crashing market isn't the same as keeping you dry. The Nifty actually ended those years in the green only twice, in 2002 and 2022. And in 2008, the worst of the lot, it fell 51.8% while the Nasdaq fell 40.5%.
The difference seems to come down to one thing. When the tech bust came without a US recession, like in 2000, 2002 and 2022, India held up relatively well. When it came with one, like in 2001 and 2008, India got soaked too.
So we tried putting rough odds on the next 12 months using these base rates.
On 16 occasions since 1971, the Nasdaq has had three straight up years, as it did in 2023–25. The next year turned out bad 3 times, so about 19%. And five of the Nasdaq's eight bad years since 1971 overlapped a US recession (NBER).
Multiply those through, and the "orderly bust where India protects you" comes out at about 7%. A bust with a recession comes out at about 12%. The other 81% is a world where the AI trade keeps going, and India's fate hangs mostly on whether oil comes down. History treats that as a coin flip: after big oil spikes since 1988, prices fell the next year only half the time (FRED).
These are small samples, so please treat the numbers as rough. But the shape is hard to ignore.
Now, that doesn't mean India is a bad place to invest. Far from it.
The home crowd is real, and fewer people are stopping their SIPs. Foreign investors own the smallest share of Indian stocks in 17 years, so there's less left for them to sell. Company earnings are picking up. And at about 19 times earnings, the Nifty is a lot cheaper than it used to be.
India even rallies on some of the AI market's worst days. On October 13, when Korea's market fell 7.4% in a chip rout, Indian IT stocks rose as much as 6%.
So India may well be a good long-term bet. It's just that "hedge" is a stronger word than the data seems to support.
But you're probably wondering, what does all this mean for you?
Well, remember the home crowd? The SIPs that are supposed to catch the market when foreigners leave? That money mostly comes out of monthly salaries.
And that's where AI seems to be quietly showing up in India. Not in the stock indices, but in jobs.
Unemployment among young people in cities, aged 15 to 29, rose to 19% in September, up from 17.5% in May. TCS ended September with about 15,000 fewer employees than it had in June 2025, even as its AI business grew. And India's tech industry grew revenue by 6.1% in FY26, but its headcount by just 2.3%.
So if AI keeps squeezing white-collar hiring, it may not just hurt IT stocks. Over time, it could thin out the very SIP money that's supposed to make India safe. To be clear, that's our reading, not something the data proves yet. But it's worth keeping an eye on.
Maybe, then, the better way to think about India isn't as an umbrella against AI. It's a bet on oil, the rupee, and the paycheques of India's salaried class.
Whether that bet holds up the next time it really pours is something we'll only find out when the rain comes. We hope it does.
Until dust settles…
If this story helped you look at the "India as an AI hedge" idea a little differently, share it with your friends, family, or even strangers on WhatsApp, LinkedIn and X.
Disclaimer: This story is not investment advice. The examples and data shared here should not be taken as recommendations to buy or sell any asset. Investments are subject to market risk, and past performance does not guarantee future returns. Always do your own research or consult a qualified financial advisor before making investment decisions.