India’s $100
billion problem.

Executive argument

India built a development model around selling human work to the world.

Foreign clients pay us. We pay salaries. Those salaries pay for homes, restaurants, schools, everything around them. The dollars help pay for the things we buy from outside India.

Now the client can buy intelligence without buying the Indian engineer's time.

And when we say, fine, we will move people into factories, there is another machine waiting there. Robots. The same AI getting better at software is getting better at controlling physical work.

We can lose both ladders before we own what replaces them.

Labour arbitrage collapses when a machine does the work for less. India cannot defend its export engine by offering cheaper people to a buyer that no longer needs their labour.

The missing $100 billion is annual. It does not politely happen once and leave. Reserves buy time. Debt buys time. Selling assets buys time. None of them rebuilds the income engine.

If we do not replace the earnings or secure financing, we buy less from the world. That hits fuel, equipment, households, credit, property and jobs.

My conclusion: India is fucked unless we do something radical.

The danger is becoming a colony again: foreign suppliers controlling the machines, foreign financiers funding the imports, foreign owners collecting the income, and Indians losing the power to decide.

A country can keep its flag and lose the power to say no.

Build the replacement income engine. Own the productive capacity. Keep the power to say no.

1. Premises and boundaries

This paper runs one scenario: annual Indian IT export receipts fall from $200 billion to $100 billion. AI keeps improving fast. Robotics improves with it. Automated production gets good enough to eat into India's cheap-labour advantage.

The figures below separate dated observations from the inputs to this planning case. Capital-flow cases, assumed competitor costs and exchange-rate/oil tables illustrate mechanisms and sensitivities; they do not report future outcomes.

Take fast AI and robotics progress seriously. Do not assume it at the start, then quietly switch to slow automation when the consequences become uncomfortable.

Robot intelligence is not the whole cost of a robot. Hardware, power, maintenance, equipment life, utilisation, finance, safety and factory redesign still matter. Smarter software does not make all of those free.

The scenario here is that those constraints get solved well enough for automated foreign production to weaken our labour advantage. It is not a claim that every robot's delivered cost falls at the same exponential rate as its intelligence improves.

This is a national income problem. The loss of export earnings strips dollars from the economy that pays for imports.

The adjustment does not require every reserve to disappear or every rupee transaction to stop. RBI can stop defending the currency before reserves hit zero. Households can keep buying domestic goods while losing income and access to essential imports.

People can still trade domestic food, housing and services in rupees after export income falls. Depreciation does not make every rupee worthless. Literal zero purchasing power would require another collapse in production, institutions or the monetary system.

If work moves from an Indian IT company to a foreign company's office in India, Indian wages and service exports can survive. If the foreign client automates the work abroad, the income leaves India.

Those are different outcomes. Count them properly.

Industry revenue is not export receipts. Export receipts are not net foreign-exchange earnings. Neither is the same as domestic value added or shareholder profit.

For the calculations below, $100 billion is the additional annual external gap after any savings in overseas IT delivery costs. If we lose $100 billion of gross revenue but also stop spending money abroad to deliver it, the net hole is smaller.

The response directions in this paper are proposals. This is not an announcement of a fund, an investment product or an operating programme.

2. India's import bill and the export offset

Here is what we buy from the world.

The government's April 2026 release puts FY2025-26 goods imports at $774.98 billion and estimated services imports at $204.42 billion.

Total: $979.40 billion.

Against that, goods exports were $441.78 billion and estimated services exports were $418.31 billion.

The goods deficit was $333.19 billion. The services surplus covered $213.89 billion of it. The combined trade deficit was still $119.30 billion.

Services are doing real work in this equation. They are helping pay for the physical things we do not produce enough of.

These are the April release numbers. March services were estimated and the figures can be revised. They are not being passed off as a final revised September series.

The goods bill

1 April 2025 to 31 March 2026. US$ billion, rounded to two decimals. Remaining goods is calculated from the unrounded source figures.

Services imports sit on top of this table. Their category-level breakdown is outside this paper.

Keep imports and everything else unchanged. Remove $100 billion of service exports. The combined trade deficit goes from $119.30 billion to $219.30 billion.

That is not the current-account deficit. Remittances and primary income still have to be counted.

The lost income is about 10.2% of our entire import bill. If we close the hole only by buying less abroad, that is a serious cut.

And imports are not all disposable consumption. Imported oil, stones, components and machines also support exports. Cut those and you can cut the income they help earn.

The saving is the net foreign-exchange effect. Not whatever impressive number you can pull from a customs table.

3. What a recurring $100 billion gap means

$8.33 billion a month.

$274 million a day.

Every year.

There is no single invoice arriving at the finance ministry. A source of dollars disappears from the system. Importers still need dollars. Banks have to find them somewhere.

Exporters. Remittances. Investors. Lenders. Other banks. RBI.

When one large source shrinks, another source has to grow, spending has to fall, or reserves have to cover the difference.

That is the entire bridge: more net export earnings, more net transfers or foreign income, lower imports, more net capital inflows, or reserve drawdown.

Existing remittances and existing capital inflows already help fund the existing position. You cannot count them again to fill a new hole. And incoming money is only half the story. Foreign exits, outward investment and debt repayments count too.

An unchanged $100 billion annual loss means $300 billion of missing receipts over three years. $1 trillion over ten.

These are missing receipts. Reserve drawdown is the balance after replacement earnings, financing and spending adjustments.

Reserves buy time. They do not earn next year's income.

Reserves are foreign purchasing power we accumulated earlier.

If RBI supplies the entire missing $100 billion each year, we move the loss into the reserve stock, before other flows and valuation changes.

The simple bridge is usable reserves divided by annual drawdown. But "usable" matters. Liquidity needs, external obligations, reserve composition and intervention commitments all affect it.

A reserve countdown requires a dated measure of usable reserves and external obligations. That measure has not been established here.

Debt is future income spent early.

Borrow $100 billion every year for ten years. At an illustrative 5% annual interest rate, that leaves $1 trillion of principal and a $50 billion annual interest run-rate on that stock.

That illustration leaves out repayments, compounding and exchange-rate changes. Even on that stripped-down basis, the point is obvious. You now have another external bill.

Borrowing to build a replacement income engine can make sense. Borrowing forever because the old engine died is a different business.

Foreign equity does not come with a fixed principal repayment schedule. It comes with claims on future profits.

A new factory can create new external earnings. Selling an existing asset mainly gives us money now and changes who owns its future income. Both have to be judged after the money paid to foreign owners.

Funding the transition and selling the furniture are not the same thing.

4. The exchange rate and the oil channel

USD/INR is how many rupees buy one dollar.

Moneycontrol reported an opening rate of Rs 96.05/$ on 29 September 2026 at 09:03 IST. The Economic Times reported Brent futures at $107 a barrel in a report updated at 08:11 IST that day.

These observations are from different times on 29 September 2026. India’s contracted crude price is a separate measure.

A $100 billion loss does not magically give us one future exchange rate. Capital flows, RBI intervention, import demand, export supply, inflation and expectations all affect it. A precise rupee forecast without those inputs is astrology with a spreadsheet.

But if other flows do not fill the hole, the pressure has a direction. Dollars get harder to obtain. Our imported purchasing power falls.

An Indian IT collapse does not automatically raise oil's dollar price. A weaker rupee raises what the same dollar-priced oil costs us.

Rupees per barrel = dollars per barrel × rupees per dollar.

Keep oil at $107. Move the rupee to Rs 120/$. The rupee crude cost rises 24.9%. We get the same barrel. It costs us more rupees.

At fixed volumes, the dollar import bill has not changed. Retail fuel prices also include refining, freight, distribution, taxes, subsidies and margins.

Someone carries the domestic cost: consumers, government or suppliers. Moving it between them does not produce a single extra dollar.

Petroleum-product exports were $53.91 billion. Against petroleum imports of $173.95 billion, the net petroleum trade bill was about $120.04 billion.

Raise both import and export prices by 20%, hold quantities fixed, and the net bill grows by about $24.01 billion.

A separate dollar-price shock enlarges the external hole through the net petroleum bill.

5. Why robotic China blocks the easy manufacturing answer

"We will move everyone into manufacturing."

With what advantage?

IFR's World Robotics 2025 release records roughly 295,000 industrial robot installations in China in 2024. India installed 9,100.

China accounted for 54% of global installations. It had more than two million robots operating. Chinese manufacturers supplied 57% of China's domestic installation market.

They are building the machines and using the machines.

These installation figures establish the existing industrial gap. Future robot costs and manufacturing output are separate from this measured installation series.

The advantage shifts to productive capital, equipment suppliers, power, utilisation, process knowledge, logistics, finance and buyers.

Cheap wages matter less when the labour hours required approach zero.

We are competing with automated systems. Not just Chinese workers. Other advanced economies can automate closer to their own customers too. China is not the only country with this option.

Making the rupee cheaper does not make our factories better.

Take a hypothetical Indian product costing $100. Domestic costs are $80. Imported inputs are $20. Start at Rs 96.05/$.

Keep domestic rupee costs, imported dollar costs, quality and everything else fixed. To match a Chinese dollar unit cost C, the required exchange rate E is:

80 × (96.05 / E) + 20 = C

E = 80 × 96.05 / (C − 20)

The equation isolates the exchange-rate effect at fixed domestic costs and imported dollar costs.

The imported $20 does not get cheaper because our currency falls. It creates a floor. If imported inputs are 50% instead, matching an $85 competitor needs Rs 137.21/$.

Domestic inflation, higher finance costs or a weaker Chinese currency can eat into the advantage further.

So yes, depreciation can preserve some sales. By making Indian income worth less internationally. It can do that repeatedly and still fail to close the physical productivity gap.

Getting poorer until we are competitive is not a development strategy.

And matching a toy unit cost does not win an order. Quality, delivery, capacity, certification and customer access still matter.

India can import robots. We should not pretend imported equipment is forbidden. It can improve productivity.

Imported robots cost foreign currency upfront. Automated output does not rebuild the labour-income ladder that automation destroys.

Judge each project over its lifetime:

Additional export receipts + avoided imports − imported equipment and inputs − maintenance − foreign fees − financing payments − foreign profit claims.

It needs positive net external value and positive domestic economic value. Do not count the same saving twice.

6. The domestic cascade

The IT worker's salary is someone else's revenue.

When receipts fall, there is less money for salaries, hiring, contractors, profits and taxes. Revenue and employment do not fall in a neat one-to-one ratio. Output per worker changes. This paper does not invent a national layoff count.

And the damage is bigger than people already employed. It includes the entry-level jobs that never get created.

Households do not wait for the final layoff notice. They cut spending when their expected future earnings weaken.

Then exposed housing and office markets lose demand. Developers, retailers, transport operators, schools and local services lose business.

A borrower can lose income and collateral value together. Banks can tighten credit before defaults show up. These channels feed each other. They do not wait politely for their turn.

At the national level, a weaker rupee makes imported inputs more expensive while weaker incomes reduce demand.

Cut machinery, components, energy or fertiliser imports and we can damage today's output and tomorrow's productivity.

Less demand. Higher selected costs. Both at once.

A national GDP growth number can sit above severe damage in the households and regions taking the hit.

The middle-class ladder breaks.

The danger is lower real household income and less access to imported or import-dependent necessities and productive goods.

A broken route into the middle class. Persistent impoverishment. Less ability to buy what we need to produce. That is already enough.

7. Dependence on the US or China

A colony does not need to look like the last colony.

The fear behind "slaves" is losing practical sovereignty. Keeping the formal right to choose while losing the ability to afford the choice.

Look at the dependencies:

• Foreign compute and model access.
• Chinese or other foreign industrial equipment and components.
• Foreign financing.
• Foreign ownership of income-producing assets.
• Foreign buyers deciding whether our exports have a market.

Any dependency gets more dangerous when there is no alternative supplier, no replacement income and not enough reserves to survive an interruption.

Creditors and investors get bargaining power through financing terms. Concentrated suppliers get bargaining power because we cannot switch. A country that cannot pay for essential imports has fewer choices, however sovereign it looks on paper.

Concentrated suppliers, financiers and owners gain power when India cannot switch or replace the income. Domestic capacity and Indian-owned foreign assets build alternatives. Practical sovereignty is the power to say no.

Measure sovereignty by the production and essential imports India can sustain when a supplier or financier says no.

8. What a radical response must accomplish

There are two jobs. Do not confuse them.

Earn or save foreign currency.

Keep households able to buy things when human labour earns less.

A profitable automated factory can solve the first and leave the second untouched. Domestic AI access can improve productivity without earning foreign currency. Owning assets only pays if those assets produce a real surplus after costs.

These are the response directions:

Build income the world will pay for.

Fund automated production and other exports where we have an actual competitive advantage. Real customers. Competitive delivered cost. Positive net foreign-exchange economics.

A gross export target is not enough. Show what remains in India after the bills.

Stop structurally needing so much from outside.

Energy substitution. Productive domestic supply. Recycling.

Prioritise what saves recurring foreign currency after imported equipment and inputs. Make the economics work. A tariff is not proof that we saved anything.

Own productive capital and the income it generates.

Indian ownership of viable domestic and foreign assets can earn beyond wages.

Foreign assets cost money today. Their income and sustainable realised gains can fund recurring spending tomorrow. Buying an asset is not the same as already having its future income.

Make productivity gains reach households.

If labour income falls, people need a funded route to purchasing power from capital income and tax receipts.

A dividend with no profitable assets behind it is a slogan. A transfer with no tax base behind it is a promise somebody still has to pay for.

Use external financing to build the bridge out.

Measure projects by additional net external income, resilience, domestic benefit and time to cash flow.

Reserves and debt can fund the transition. They cannot stand in for the result.

Radical means changing the income and ownership structure fast enough to keep our choices.

This paper does not claim that a particular fund, subsidy, industrial programme or transaction has already been designed or agreed.

9. Decision, evidence, and next action

My judgment is that India faces national economic failure and subordination if we do not respond radically to AI and robotics together.

This is not a mild slowdown thesis.

Lose labour-based export income. Fail to compete in automated production. The external accounts force an adjustment: financing claims, cuts to imports, lower domestic incomes and weaker sovereignty.

An economy that loses its income engine pays for the failure through claims on its future and cuts to life today. Delay gives the owners of its machines and financing more power.

The biggest unanswered question is simple:

Can we build or acquire enough net external earning capacity before the missing IT income forces damaging cuts to consumption and investment?

The next analysis needs an annual external-account bridge. Dated usable reserves and obligations. Net IT losses. Other exports. Import price and volume responses. Net capital flows. A small set of replacement projects.

Every exchange rate must be an explicit stress input or a model result with its assumptions shown. No invented precision.

The response must prove four things:

Replacement foreign earnings that actually stay in India. Imports falling because we produce substitutes, not because we are too poor to buy them. Indian automated producers winning sustained orders at positive net external returns. Broad household income that stays funded as demand for labour falls.

An AI announcement is not that evidence. A robot demo is not that evidence. A gross FDI number is not that evidence.

Show the income engine working.

Conclusion

The old plan was to sell our brains. The fallback was to sell our labour.

AI can take the first. Robotics can take the second.

Under this scenario, we cannot assume displaced software workers will find a comparable export-income ladder in factories. Both labour advantages can disappear.

If we do not build replacement income and ownership, the adjustment still happens. Through more claims on our future earnings and less purchasing power today.

We need to pay for essentials and the imports that help us produce. Keep household income alive. Keep enough bargaining power to make our own decisions.

That is what radical action is for.

Build the machines. Own the income. Keep the country able to say no.

Sources & status

Developed on 30 September 2026. Published 1 October 2026. Revised 2 October 2026.

Annual trade figures come from the April 2026 release and can be revised. Market quotes retain their observation dates.

• Ministry of Commerce: FY2025-26 goods and services trade, 15 April 2026.
• Ministry of Commerce: March 2026 commodity tables, export page 1 and import page 3.
• International Federation of Robotics: World Robotics 2025, reporting 2024 installations.
• Moneycontrol: USD/INR opening quote, 29 September 2026, 09:03 IST.
• Economic Times: Brent futures quote, 29 September 2026, updated 08:11 IST.

Thesis by Milind Sonpal, developed with AI assistance.