Rajeev Thakkar explains why FIIs are really leaving India | 1 Finance Magazine

ELI5/TLDR

Rajeev Thakkar runs India’s largest flexi-cap fund (PPFAS, ~₹1.3 lakh crore) and is sitting on roughly 20% cash while everyone asks why he isn’t buying the 10% dip. His answer: the dip was small, valuations weren’t cheap to begin with, and cash that earns 7% in a flat market hasn’t actually cost his investors anything. He thinks foreign money left India for one boring reason — prices ran ahead of earnings — not because India lacks an AI champion. And the realistic return to expect from Indian equities over the next decade is about 11-12% a year, in line with nominal GDP. Nothing exotic.

The Full Story

Why a flat market and a 20% cash pile aren’t contradictions

The interview opens with the question everyone on the internet wanted answered: the Nifty fell ~10%, Thakkar deployed over ₹4,000 crore in March, yet his fund still shows 19% cash. Why so timid?

His first point is mechanical and easy to miss. Cash percentage moves on its own without anyone trading. Hold 80% stocks and 20% cash; if stocks rise 10%, the stock slice swells and cash looks smaller even though you did nothing. The reverse happens in a fall. So the headline number bounces around regardless of conviction.

The deeper point is about the size of the fall and where it started.

“In global financial crisis 2008 2009 the fall at the index level was 60%. In COVID the fall was 40%. This time the fall was not that big to start with. That’s the first point. Second is our starting valuations were not cheap.”

A 10% dip from an expensive base is not a fire sale. He also reframes the supposed cost of holding cash. Over the last 18 months the Nifty 500 went nowhere, while debt instruments paid roughly 7% a year. Sitting in cash didn’t lose his investors anything — it quietly beat a flat market.

The FD complaint, and why equities are supposed to disappoint sometimes

Newer investors — the “class of 2020” who arrived after the post-COVID boom — are jittery because their statements aren’t going up every year. Some complain that equities aren’t even beating a bank fixed deposit. Thakkar’s counter is almost a logical proof:

“If equities beat FD on every time frame, every 6 months, every 12 months, FD will cease to exist as a product.”

The only reason equities pay more than debt over time is that they sometimes pay nothing, or less than nothing, for a stretch. The bad periods aren’t a malfunction. They’re the toll you pay for the higher long-run return. Investors who’ve lived through one full bull-and-bear cycle understand this; the ones who haven’t get spooked.

The AI panic, and the unglamorous math of IT services

Here’s the contrarian heart of the conversation. The market treats India’s lack of a homegrown AI model as a national tragedy. Thakkar shrugs. Frontier models are basically three — Google’s Gemini, OpenAI’s ChatGPT, Anthropic’s Claude — with a few runners-up like DeepSeek and Meta’s Llama. The UK has none. Japan has none. South Korea has none. Even Microsoft and Apple are partnering rather than building their own. So why single out India?

“Why are we so obsessed that we don’t have an Indian AI? A lot of developed countries don’t have their own frontier models.”

On the IT services companies the market is busy writing obituaries for, he offers a margin lesson worth internalizing. Suppose you sold a service for 100 at a cost of 80 — a profit of 20. AI now means you can only charge 90. Sounds like doom. But your cost also falls, say to 70, because your people are more productive and you need fewer of them on a project. Profit: still 20.

“People are looking at only one side of the equation saying there will be deflation in terms of the order values. But that deflation in order value is because of increase in productivity.”

And cheaper services unlock demand that didn’t exist before — projects that failed a cost-benefit test at the old price become viable at the new one. (Economists call this demand elasticity; the practical version is “when something gets cheaper, people buy more of it.”) IT services have always been cyclical — Y2K, the dot-com bust, the shift to cloud, then mobile, then SaaS. AI is just the next reinvention, not the funeral.

The real reason FIIs left: arithmetic, not narrative

Foreign investors pulled more than ₹3 lakh crore out over two years, and over ₹1 lakh crore in March 2026 alone. The popular story blames the missing AI play, the securities transaction tax, operational hassle. Thakkar deflates all of it:

“I think the valuations had run up ahead of the fundamentals. Beyond that, I don’t think there was a large issue.”

His clearest example: multinational subsidiaries listed in India that do 7-10% of their parent’s global business were trading at 80-90% of the parent’s global market cap. That’s a glaring distortion. Meanwhile cheaper markets beckoned — Taiwan with TSMC, South Korea with Samsung and its chipmakers, both riding the hardware side of the AI cycle. Money flows where it’s cheap and growing. He expects the gap to self-correct; nothing special needs fixing.

On taxes he makes a point most Indians don’t know: foreigners pay no estate duty investing in India, whereas an Indian who buys US stocks directly and then dies owes the US a 40% estate duty, plus 25% withholding on dividends. Every country has friction. Tax isn’t a deal-breaker; it just has to be priced in.

Look outside India — most of the world is

A statistic worth sitting with: India is about 4% of global market cap. The other 96% is everywhere else. The phone in your pocket runs iOS or Android — both American. Its chips are designed abroad. The apps (YouTube, WhatsApp, Instagram) are foreign. The blockbuster drugs (Ozempic, Mounjaro) are invented elsewhere. Ignoring 96% of the world’s listed companies is a choice, and Thakkar thinks it’s a poor one. Indian mutual funds have been frozen out of overseas investing for four years (the regulatory limit is maxed), but individuals can still send money abroad under the LRS route, and GIFT City funds are opening up to retail at ticket sizes as low as $500.

Cash flow is the dividing line for what he’ll value

Asked about gold, crypto, art, watches, wine, he draws a clean line. He’ll only call something an investment if it throws off cash flow you can model.

“If I had gold, I would not throw it away. But they don’t generate any inherent cash flow.”

Gold, silver, crypto, art — value purely on demand and supply, no framework to say what’s cheap. He simply has no view. Equities, REITs, InvITs, bonds, even aircraft leases and warehouses — those have estimable cash flows, so he can say what’s too expensive. PPFAS holds ~9.5% across two listed REITs, which he likes: ~6% yield plus some capital appreciation gets you comfortably double-digit. InvITs (roads, transmission lines, gas pipelines, telecom towers) he tells retail to avoid — too complex, and the headline yield is a trap because part of the “distribution” can be your own capital handed back.

The number to anchor on

For the next 5-10 years, what should a Nifty investor realistically expect? His framework is almost deflatingly simple: nominal GDP growth flows into corporate earnings, which flows into stock prices. Roughly 6-7% real growth plus ~5% inflation gives ~11-12% nominal. That’s the benchmark. Not 20%, not 30% — 11-12%.

His closing advice for turbulent times is two words. Asked whether investors should be disturbed or stay the course: “Sleep well.” Don’t over-weight geopolitics — today it’s Hormuz and Israel-Iran, a few years back it was Russia-Ukraine (still going, no longer headlines). Keep money you need in 1-3 years out of equities entirely, put long-term money into growth assets, and stop watching.

Key Takeaways

A fund’s cash percentage moves on its own with market prices — it rises in a fall and shrinks in a rally without any trading, so the headline number overstates the manager’s actual stance.
Holding cash isn’t free or costly in the abstract: over the last 18 months a flat Nifty meant cash earning ~7% in debt actually beat the index.
Equities can only out-earn FDs over the long run because they sometimes return nothing for a stretch — the drawdowns are the price of the premium, not a failure.
Only three real frontier AI models exist (Gemini, ChatGPT, Claude); most developed nations — UK, Japan, South Korea — have none, so India’s “missing AI” is unremarkable.
The margin defense for IT services: falling order values come from rising productivity, so lower revenue per project can be offset by lower cost per project — profit can hold.
Cheaper IT services unlock previously uneconomic projects (demand elasticity), so topline growth can return even if it’s not immediately visible.
IT services are cyclical by nature (Y2K → dot-com → cloud → mobile → SaaS → AI); each wave forced reinvention without killing the core function.
The actual driver of FII outflows was valuation overshoot, not the AI narrative or taxes — e.g. Indian-listed MNC subsidiaries doing 7-10% of parent revenue traded at 80-90% of parent global market cap.
Indians buying US stocks directly face a 40% US estate duty on death plus 25% dividend withholding; foreigners investing in India face neither — every market has its own tax friction.
India is ~4% of global market cap; the other 96% (phone OSes, apps, blockbuster drugs) is foreign — individuals can access it via LRS or GIFT City funds (now as low as $500 ticket).
Thakkar only values assets with modelable cash flow (equities, REITs, InvITs, bonds, aircraft leases); gold, crypto, art, wine he declines to have a view on.
REIT trap to avoid: don’t annualize a quarterly distribution and divide by price — part of the payout can be return of your own capital, so the “yield” is misleading.
PPFAS’s large-cap fund exploits arbitrages index funds legally can’t: buying the cheaper of two merging stocks (e.g. HDFC Ltd at a 4% discount to HDFC Bank), or front-running the forced last-day index rebalance.
The long-run equity return anchor: nominal GDP (~6-7% real + ~5% inflation ≈ 11-12%) is the realistic Nifty expectation, since earnings track GDP and prices track earnings.
In secular themes (internet, mobile), being first is often fatal — the early entrants went bankrupt; survivors bought after the hype crash made money. Wait for consolidation, management quality, and balance-sheet strength.

Claude’s Take

This is a high-signal interview precisely because Thakkar refuses to be interesting. Every dramatic narrative the questioner floats — AI doom, FII exodus, the dip you “must” buy — he answers with arithmetic and a shrug. That discipline is the product. PPFAS built a ₹1.3 lakh crore fund by being boring on purpose, and the interview is a clean window into that temperament.

The strongest segment is the IT margin argument, because it’s a genuine reframe most commentary misses: the market sees falling order values and prices in death, while ignoring that the same force cutting prices is cutting costs. Whether profits actually hold is an empirical question he can’t prove here — it’s a plausible thesis, not a demonstrated fact — but it’s the right question, and almost no one in the AI-panic discourse is asking it. Same with the FII point: “valuations ran ahead of fundamentals” is unglamorous and almost certainly more correct than the AI-shortage story the media prefers.

Where I’d add friction: the book is talking his own book. He’s overweight IT and REITs, so the bullish IT case and the REIT enthusiasm aren’t disinterested. The 11-12% nominal-GDP-equals-returns model is a reasonable anchor but glosses over the “leakages” he mentions in one breath and drops — equity dilution, multiple compression from an expensive start, and the fact that the entry valuation he himself flags as elevated can eat years of that return. An investor anchoring on 11-12% from today’s levels might be disappointed for exactly the reason he gives for FIIs leaving.

A 7 — substantive, honest, quotable, and refreshingly free of hype, but it’s a single fund manager’s measured worldview rather than a deeply original framework, and the conflicts of interest are real even if he’s transparent about his positions.