The demographic buyback solution: Why a lower fertility rate may help India

Rather than a crisis, India's falling fertility could boost per capita prosperity if AI-driven productivity offsets a shrinking workforce

In India, we get our knickers in a twist over the silliest of things. Stray dogs. A gymkhana club. Cockroaches.  

Probably the only sensible thing that’s got us worked up in recent months is the news that India’s total fertility rate (TFR) is now below the replacement rate. “India will get old before it gets rich” and “Our demographic dividend is no more”, this dirge trended for several days.  

Viewed conventionally — and economics is conventional beyond reproach — this is apparent Armageddon. For a country whose major export has been manpower and which has been producing more babies than anyone else, what could be worse news? 

Except that it’s not. 

The Indian stock bear market must be really soporific because the usual India fund management choir that breaks out singing “All global bad news is good news for India”, has slumbered through this sole bullish news to have come out of India in the last couple of years.  

For decades, the global economic priesthood has prostrated at the altar of something called the demographic dividend.  

The gospel is simple: If your country is pumping out babies faster than fake news on Indian WhatsApp, you are mathematically guaranteed a golden age of economic dominance.  

Conversely, if your fertility rate plunges, you are a Titanic heading into an iceberg.  

It sounds logically airtight. 

It also happens to be so dated, so 2020. And so completely wrong.  

It is a 19th-20th century construct applied to a 21st-century silicon reality. It misses the todayness of evolution and progress, especially since consumer-level AI rode in 3-4 years ago.  

A few months ago, we wrote in this paper that the way to look at America was not to think of it as a country, but as a giant 

platform with millions of profitable customers. It lured all sellers by giving them zero-cost entry (low tariffs). And now that everybody is hooked, new charges (higher tariffs) start creeping in.  

It made the tariff war logic relatable. Everyone got it instantly.  

It is time we applied some real-world corporate finance to population science. 

Think of per capita gross domestic product (GDP), as a loose equivalent of earnings per share (EPS). Allow us some economic licence here, okay? 

There are two ways a company can grow EPS: Grow revenue, profit and, thereby, EPS.  

But there is another way. When you cannot deliver through the first way, you simply shrink the denominator — the number of outstanding shares — which, then, even if your revenue and profit are not growing a lot, can still keep increasing because the same profit is now getting divided by a smaller number of shares.  

In corporate finance, this is called a share buyback.  

This is pure financial engineering: Even if net profit remains flat or grows moderately, the EPS increases because those earnings are now divided over far fewer shares. 

Think AutoZone: Negligible profit growth for years but 20 per cent EPS growth because of aggressive buybacks. The stock price went from $250 to $3,000. Apple did the same between 2015 and 2019. IBM. Oracle. This is exalted company.  

For centuries, growing the economic pie required a moderately curvilinear model: More bodies, more growth. 

However, now, when a smart country's population shrinks, it replaces a warm body with a subscription or an automaton. 

N Chandrasekaran, chairman, Tata Group, recently said that in TCS, they are going to have an equal number of human and AI agents working in the near future.  

Expand this to a country level and you get the drift.  

The total ‘shares’ (the population denominator) decrease, while total productivity (the numerator) via agents expands.

The result? The macroeconomic equivalent of EPS—GDP per capita—compounds exquisitely. 

Crucially, top-line GDP growth is not going to stand still while populations shrink.  AI and robotics are accelerating baseline national-level productivity quite dramatically. Estimates put an additional 1-1.5 per cent productivity growth in America because of AI. Something even better is happening in China. 

We are no fans of lofty Excel projections (It’s polite not to talk about that seminal Excel exercise called Brics 25 years back, stunning in its linear flatness. We now know better, but some Excel hallucinations help. As you will see soon. 

PwC research says AI is projected to inject up to $15.7 trillion into the global economy by 2030, with a staggering $6.6 trillion driven directly by labour productivity gains.  

Findings from McKinsey indicate that generative and agentic AI architectures will add between $2.6 trillion and $4.4 trillion annually in pure structural value across global industries.  

An AI market bubble collapse won’t change anything. The infrastructure will deliver. Just as the internet has done despite the dotcom crash.  

When you compound these accelerated productivity gains over 50 to 100 years, the traditional demographic doom 

narrative vanishes. 

Year 2100 projections 

Exhibit: The 2100 Wealth Ranking 

India’s national TFR has officially breached the replacement mark, dropping to roughly 2.0, with progressive southern states plunging all the way down to 1.3.  This, to us, merits uncorking a Dom Perignon Plenitude. 

India needs to (and will) reverse compound its (population) share count, in order to positively compound its EPS, i.e., per capita GDP. 

It means India will avoid the trap of having to perpetually dilute its wealth across an exploding denominator. 

Let’s assume India's population peaks around 1.65 billion in the 2050s and then naturally steps down to a leaner, meaner 1 billion by 2100. Over the next 75 years, let's assume India’s real long-term GDP growth averages a conservative 3.5 per cent, allowing for a natural cooling as the economy matures, but structurally elevated by AI integration in its services and manufacturing sectors. 

By 2100, India’s aggregate GDP would reach roughly $52 trillion in real terms. Divide that by 1 billion ‘shares’, instead of a bloated, over-capitalised 1.65 billion share count. 

India’s per capita GDP surges from today’s $2,675 to $47,000. 

India knifes through the middle-income trap entirely, not by breeding more population ‘shares’ i.e. mouths, but by driving massive productivity per human through automated systems, divided over a far lower head count.  

This is important: the per capita wealth differential between the United States and India compresses dramatically over this period, dropping from a 34x advantage in 2025 to just a 9x gap by 2100. Concurrently, the economic spread between China and India follows a similar, narrowing trajectory, with China’s per capita premium over India shrinking from a 5x multiple to just 3x.  

This reduction in the power gap will have massive geo-political implications. 

Macro-EPS growth beats top-line GDP growth vanity. 

The traditional demographic dividend logic was perfectly valid in the 19th and 20th centuries, when economic output was a linear function of muscle tissue and factory shifts. If you didn’t have the warm bodies, your factories remained cold. 

But today, a single engineer running an AI-driven automation layer can generate the economic output that used to require an entire village. 

Think about the implications:  If the ‘demographic buyback’ did not happen, India’s per capita GDP in 2100 would be $20,000. And even that is iffy. 

The implications of all of these are seismic: the concept of output gap — i.e., actual GDP minus potential GDP — may well vanish. 

By replacing human labour with software agents and humanoid robots, a nation can expand its potential GDP vertically without adding a single human worker. 

Even as Japan, South Korea, China and India’s populations shrink, their automated capacity will ensure they never hit a structural capacity wall. 

They can sustain a massive, non-inflationary negative output gap where supply capability is endlessly abundant, prices stay low, and per-capita wealth skyrockets. AI transforms potential GDP from a function of headcount into a function of compute.

These creatures will not require state and parental spending till they become productive by age 21. They will be productive instantly for small subscriptions.  

They will not need to be trained. 

They will not have the uneven quality of the human demographic dividend. Just producing millions of below-average children is a solution worse than the problem it seeks to solve.  

The government’s resources on critical items like poverty alleviation, food grain, health, nutrition, social security, etc., will decline dramatically as a percentage of GDP because there are far fewer mouths to feed.  

The pressure on resources, natural and monetary, will ease dramatically.  

The resultant population will be far more prosperous and have the spending power to drive consumption almost effortlessly. 

Think premiumisation across the board: Fewer cars, but higher-end cars, for example.  

In fact, the entire construct of GDP itself will change, insofar as the GVA is concerned: Generating one unit of GDP will require far less resources than today. The GVA delta may well surprise us:

Mathematically, Gross value added (GVA): GVA = gross value of output minus intermediate consumption

Gross domestic product (GDP) from GVA standpoint = GVA plus product taxes minus product subsidies 

The impact on our fisc will be leviathan: Tax collections will rise, freebies and subsidies will drop. Infra spending will decrease because you need a lot less for a billion people vs 1.7 billion.  

The economy’s  ‘free cash flow’ will increase, along with the EPS increase. That’s hypnagogic — deep dream state, that is. 

Mandatory caveats apply: India needs to actually make the AI and robotics bit happen through policy. Just gazing at this article’s exhibit won’t manifest it. (At this point we should also add that the exhibit meets the key test of a good forecast: That it should be so far out into the future that nobody remembers it then. Because very few reading this piece now are likely to be around in 2100). 

The path will be non-linear. In the interim, unemployment will be the cause of social unrest. Jantar Mantar will be booked solid right up to 2090. 

But we need to last it out till 2100, because this Y2100K bug is a good, warm, cuddly bug. 

There....we knew this would bring Cheshire cat smiles back on the faces of Indian asset managers. 

Expect ‘Vision 2100 SIPs’ soon. Expect ‘Sensex at 10 cr by 2100’ forecasts even sooner.

---

Shankar Sharma is a prominent investor and founder of GQ FinXray, an AI company. Amit Bhartia is founder of Delorean Partners, a US fund, and former emerging markets fund manager at GMO, a US fund house. Shlok Rathod provided the 

data analytics