India REITs Explained: How Real Estate Investing Works?

ELI5/TLDR

A REIT is a way to own a slice of expensive commercial real estate without buying a whole building. It pools rent from many properties, is forced by law to pay out 90% of its income, and gets taxed only once. This interview is with the CEO of Capital Land India Trust — Singapore-listed, but it owns India’s first REIT and a 32-year-old portfolio of IT parks, factories, warehouses and, increasingly, data centres. The pitch: a bond-like yield (5.5–6.5%) backed by buildings you can actually see, with the AI data-centre boom adding a growth kicker.

The Full Story

What a REIT actually is, stripped down

The guest, Gauri Shankar Nagabhushanam, gives the cleanest definition you’ll hear. A real estate investment trust is “nothing but a pooling vehicle for earning regular income by investing in real estate.” It buys completed, rented buildings, collects the rent, and hands it back to investors. The structure matters because it solves a tax problem: assets sit inside separate companies (SPVs), often across jurisdictions, but the income gets taxed only once on its way out — sometimes in the trust, sometimes in the investor’s hands, never both.

The thing to hold onto: a REIT is not an operating company.

“The business of the company is not to have growth in the company and reinvest capital… The purpose of the company is to collect and distribute. That’s why most REIT regulations insist that 90% of the income you collect you have to mandatorily distribute.”

That 90% payout rule is the whole personality of the asset. With no retained earnings to reinvest, a REIT that wants to grow has to go raise fresh debt or equity. So mentally it sits closer to a bond — a fixed coupon — except the coupon comes from a pile of tenant leases instead of a single borrower. It’s an inflation hedge too, because rents tend to escalate with inflation.

Why Singapore is the centre of gravity

Singapore is the REIT capital of Asia: 12% of the entire stock market’s value sits in REITs. The reason is demographic. An aging population wants safe retirement income, and a Singapore bank deposit pays a miserable 0.5–1.5%. A REIT pays 5.5–6.5% — a 400-basis-point gap — on assets people feel they understand. You can see the building. You know it’s collecting rent. That visibility makes the risk easy to underwrite, at least psychologically.

The India backstory (a phone call between Tata and Singapore)

Capital Land’s India history is older than most people realise and turns on a single phone call. India’s first IT park — International Tech Park Bangalore in Whitefield — came out of a request from India’s PM to Singapore’s PM: help us build an industrial park like the one you built for the Chinese. The Singapore official tasked with it knew exactly one person in India well enough to call — his old classmate, Ratan Tata. The result was a joint venture (40% Capital Land, 40% Tata, 20% Karnataka government) that broke ground in 1994 and went live in 1999. The trust listed in 2007. Capital Land now has $8–9 billion invested in India and wants to double it in four to five years.

The three pools of capital

Worth understanding because it explains why a REIT looks “boring” by design. Capital Land runs money in three buckets sorted by risk, not by asset type:

Balance sheet — the riskiest, longest-gestation projects (10–15 year government-to-government ventures). Being wound down.
Private funds — development-stage assets, co-invested with sovereign wealth funds and global institutions. You buy land, get approvals, build.
The listed REIT — only stabilised, income-producing assets. “REITs are meant to distribute income.”

So an office tower under construction sits in a private fund; once it’s full of tenants and throwing off rent, it migrates into the REIT. The REIT is the calm, finished end of the pipeline.

The data-centre detour

The most interesting stretch. Capital Land got into data centres in 2021, mid-Covid, out of fear. Offices were 4–5% occupied and nobody knew if anyone would come back.

“That existential crisis made us diversify into data centres because… if the world is not going to be physical, it’s going to be digital.”

There were no finished data centres to buy in India, so a REIT that exists to avoid development had to break its own rule and build them. The economics are unusual. He splits a data centre’s cost into three layers:

Core and shell — just the building. Costs roughly $1 of $9.
Powered shell — building plus all the power infrastructure (a data centre is a “power guzzler”). Adds another $2.
The IT load — the tenant’s own racks, chips and GPUs. The remaining $6.

Capital Land plays in “colocation” — it provides and maintains the power, cooling, backups and security, while the tenant brings the chips. The clever part: the tenant spends roughly twice what the landlord does. Once a hyperscaler has sunk a billion dollars of its own capex into your building, it is not leaving. That stickiness, plus tenants with the deepest pockets on earth, is why data-centre valuations have gone “mad” in the West. India is wildly under-supplied — 1 MW of data centre per million people, against 220 MW in Singapore, ~10 in China, ~50 in the US. A genuine sunrise industry.

The discipline: data centres won’t take over the portfolio. The target mix is two-thirds steady-state (offices, IT parks), one-third “new economy” (data centres, industrial). Big enough to scale up fast if needed, small enough that a tech shift can’t sink the ship.

Who owns it, and can an Indian buy in

Half the units are held by Singapore “mom and pops” — retirement money chasing that 6.5–7% Sing-dollar yield against ~1% at the bank. The rest is institutions and global emerging-market funds. An Indian retail investor can buy, subject to the cap on overseas investment, and gets two extras: an appreciating currency (the Sing dollar climbs even against the US dollar) and a tax regime where the distributed income and capital gains are both untaxed in Singapore. He also confirms a pure India listing is coming — “just a matter of time” — as they’ve already done in Malaysia, China and Japan.

What the portfolio reads about the economy

As a landlord to factories and warehouses, he has a useful vantage point. Electronics, pharma and automotive are the strong industrial tenants. If you’re holding an iPhone, odds are good it was assembled in one of their facilities (a Foxconn-type tenant, though he doesn’t name it). The gap he keeps circling back to: India lands the headline investment — a semiconductor fab, a giant data centre — but the supplier network underneath it doesn’t exist yet. Sub-assemblies still get imported from China. Building that ecosystem is slow, partly because India isn’t a “cookie-cutter” command economy; every state runs its own policy. But states are now actively courting FDI, which helps.

The honest warning to investors

His main worry is investors confusing two different things: a REIT play versus a real-estate-operations play. A REIT is a risk-adjusted, current-yield product with modest growth. It is not a capital-appreciation bet — that’s an operational game. He’s wary of “total return” pitches that quietly lean on appreciation to make up for thin income. The other pitch: diversification. Owning one office block leaves you exposed if your single tenant walks; a professionally managed REIT with a leasing team across many markets re-lets far faster.

Closing wisdom

After 25 years, his northstar is anti-greed. Real estate has too many moving agencies — if one fails you can’t simply restart; unwinding a stuck land deal can take five or six years. So: keep things simple, don’t overextend, pay slightly more for something clean and sleep well rather than chase the optimistic outcome. “Half the money is made going in at the right time at the right price and with the right structure.”

Key Takeaways

A REIT is legally forced to distribute 90% of its income, which is why it behaves like a bond, not a growth stock — to grow it must raise fresh capital.
REIT income is taxed only once across all jurisdictions; that single-tax efficiency is the structural reason the vehicle exists.
Singapore is Asia’s REIT hub — 12% of its market cap — driven by retirees swapping ~1% deposits for ~6% REIT yields.
Capital Land’s India roots trace to a 1994 Tata–Singapore–Karnataka JV that built India’s first IT park (ITPB, Bangalore).
It runs three capital pools by risk: balance sheet (riskiest dev), private funds (development), listed REIT (only finished, income-producing assets).
By choice, Capital Land does no residential in India — only office/IT parks, industrial, logistics and data centres.
Data-centre cost splits roughly 1 : 2 : 6 across core-and-shell : power infrastructure : tenant’s IT load — so the tenant outspends the landlord ~2x, making leases extremely sticky.
India has ~1 MW of data centre per million people vs 220 in Singapore, ~10 in China, ~50 in the US — a structural undersupply.
Target portfolio mix: two-thirds steady-state (offices/IT parks), one-third new-economy (data centres/industrial).
Indians can invest (subject to the overseas-investment cap), gaining an appreciating Sing-dollar exposure and tax-free distributions in Singapore.
The trust never sold an asset between 2007 and 2025; its first divestments (one office, one industrial, plus a 20% data-centre stake) came in 2025–26.
The author’s caution: a REIT is a current-yield product, not a capital-appreciation play — don’t confuse the two.
India’s manufacturing bottleneck is the missing supplier/sub-assembly ecosystem beneath the headline mega-investments.

Claude’s Take

This is a clean, jargon-light explainer wrapped in a corporate interview, and the value is lopsided toward the first fifteen minutes. The REIT-101 section — the 90% payout rule, the single-tax mechanic, why it’s bond-like, the Singapore demographics — is genuinely the best plain-English version of this you’ll get, and the Tata phone-call origin story is a nice bit of texture.

The data-centre segment is the real signal: the 1:2:6 capex split and the “tenant sinks more capex than the landlord, so they’re stuck” insight is a sharp, durable mental model for why hyperscaler leases are considered so safe. The undersupply numbers (1 MW per million vs 220 in Singapore) are striking, though they’re his framing and worth a pinch of salt — per-capita data-centre comparisons flatter whoever’s selling the growth story.

The BS-filter caveat: this is the CEO of the trust talking, on a friendly show, with zero pushback on valuation, leverage, or what happens to those “very safe” data-centre yields if the AI capex cycle turns. The “AI will only enlarge the pie” passage is the standard landlord’s optimism — he has every incentive to believe his tenants are growing, not shrinking. And his own admission that even he can’t explain Asian data-centre valuations (“we’re still trying to figure out why they’re paying crazy valuation”) is the most honest line in the interview, and arguably the one to remember.

A 7: high clarity, real conceptual takeaways, but it’s an interview with a seller and never gets adversarial. Useful for understanding the instrument; not a substitute for a skeptical look at any specific REIT.