Atoms Must Split
Note: AI as helping tool, thoughts are mine.
There is a useful question that economics tends to postpone until it becomes impossible to avoid: what, exactly, limits how much a civilisation can produce?
We usually answer this in abstractions. Capital. Labour. Productivity. Technology. Institutions. Human capital. These are perfectly respectable answers, and economists have spent centuries refining them. But every abstraction eventually has to cash its cheque in the physical world. Capital must become a machine or a building or a network. Labour must eat. Technology must run on hardware. A factory must have electricity. A semiconductor fab must have water, chemicals, clean rooms and extraordinary quantities of power. Even the most apparently immaterial parts of the modern economy ultimately terminate in steel, silicon, concrete and electrons.
This is where the argument in Build Baby Build naturally leads. If prosperity is fundamentally a question of productive capacity, then the next question is not merely how to increase productive capacity through better institutions or more capital. It is to ask what the physical ceiling on that capacity actually is.
And once you ask that question seriously, you eventually arrive at energy.
That is a more profound question than it first appears.
The old argument was about production
It is fashionable to remember Adam Smith as the patron saint of markets and Karl Marx as his ideological opposite. The historical relationship is more interesting.
Smith begins The Wealth of Nations with production. His first chapter is not about stock markets, entrepreneurship or even self-interest. It is about the division of labour, which he calls the greatest improvement in the productive powers of labour. He observes something that now seems obvious but was revolutionary in its implications: a person embedded in a sophisticated productive system can accomplish vastly more than the same person working in isolation. The pin factory is not impressive because any individual worker is extraordinary, it is impressive because specialisation, machinery, accumulated knowledge and coordination turn ordinary human effort into extraordinary output.
This is perhaps the most important insight in classical economics. Prosperity is not principally a consequence of people working harder. It is a consequence of people becoming capable of producing more with the same hour of human life.
Marx, despite drawing radically different political conclusions, understood the importance of the same underlying phenomenon. His analysis of the "productive forces" was concerned with the development of labour, technology and the means of production, and with the way those forces interact with the social relations surrounding production. Marx thought capitalism would eventually generate contradictions between increasingly powerful productive forces and the relations through which those forces were organised. Whatever one thinks of his conclusions, he identified something worth preserving from the analysis: the productive apparatus of society is itself an object of historical development.
This is where the argument becomes interesting for someone who is neither a Marxist nor a doctrinaire laissez-faire economist.
Smith and Marx disagreed about ownership, class, markets, capital and the political organisation of society. They did not live in disagreement with the existence of factories, machines, specialisation and technology. Both were trying, in different ways, to understand a world in which human beings had begun to organise production on an unprecedented scale.
The political argument was over who controls the productive apparatus and how its output is distributed.
The deeper economic fact was that the productive apparatus itself was becoming enormously more powerful.
That distinction has survived the ideology.
A socialist economy still needs steel.
A capitalist economy still needs steel.
The ownership of the steel mill can be changed by law. The quantity of steel that can be produced cannot be changed by political decree.
A government can nationalise a power plant. It cannot nationalise the laws of thermodynamics.
That sounds almost trivial, but it is an important corrective to a great deal of political thinking. Political systems determine many things about production, particularly incentives, allocation, ownership and coordination. They do not abolish the physical conditions under which production occurs.
The Soviet Union could command factories into existence. It could not command aluminium into existence without the electricity required to refine it. China could reorganise its economic institutions. It still needed coal, steel, ports, machines, engineers and energy to become the manufacturing giant it is today.
The productive forces have their own stubborn reality.
And that reality is becoming more important as civilisation becomes more complex.
Wealth is ultimately physical
Imagine that every bank account in India doubled tomorrow morning while absolutely nothing else changed.
The country would appear richer.
It would not actually possess twice as many houses, twice as much food, twice as much electricity, twice as many doctors, twice as much steel or twice as much computing capacity. People would simply have greater monetary claims on an unchanged stock of goods and services.
That distinction explains a surprising amount of economic confusion.
Money is a coordination mechanism. It is a claim on real resources. Wealth itself consists in the things those resources can produce and the capabilities embodied in the people, machines and institutions doing the producing.
If the supply of housing is fixed while purchasing power rises, houses become more expensive. If the supply of electricity is fixed while industry expands, electricity becomes more valuable. If the number of engineers is fixed while investment surges, engineers become more expensive. If semiconductor supply is constrained while demand for computation explodes, the scarce chips become extraordinarily valuable.
There is no monetary trick that resolves this.
At some point you have to make more things.
This is why I increasingly think of economic development as a problem of constraint removal.
A primitive society is constrained by human muscle. The domestication of animals removes part of the constraint. Agriculture removes another. The steam engine removes another. Electrification removes another. Mechanisation removes another. Computing removes another. Automation removes another.
Every technological revolution is, in some sense, a decision to stop accepting a particular constraint as permanent.
This is why economic growth is so much more interesting than the redistribution of existing output. Redistribution determines who receives the products of the machine. Growth changes the machine itself.
And the machine is what interests me.
The frontier keeps moving
Suppose a country doubles its steel production.
That sounds like a victory until you realise what immediately happens. Steel becomes cheaper relative to what it was before, making bridges, ships, railways, buildings, machinery and factories more economical. Those activities expand. Their expansion creates new bottlenecks. Perhaps now electricity is scarce. Perhaps ports are inadequate. Perhaps skilled engineers are scarce. Perhaps copper becomes expensive. Perhaps the transmission grid cannot handle the additional industrial load.
Solving a constraint does not end scarcity.
It moves scarcity somewhere else.
This is one of the beautiful and slightly cruel properties of development. The richer an economy becomes, the more sophisticated its constraints become. A poor society worries about obtaining enough food. A middle-income society worries about housing, transport and electricity. A highly industrial society worries about semiconductor capacity, grid stability, advanced materials, specialised talent, compute and energy density.
The problem does not disappear.
The frontier moves.
The task of civilisation is therefore not to reach some mythical state in which constraints cease to exist. It is to become extraordinarily good at discovering the next constraint and then attacking it.
This is why I am suspicious of economic philosophies built entirely around optimisation.
Optimisation assumes that the system you are optimising is fundamentally the system you want.
Civilisation advances when somebody decides that the system itself can be changed.
The horse carriage can be optimised indefinitely. Eventually someone builds the automobile.
The factory can be made more efficient. Eventually someone invents an entirely different manufacturing process.
The power grid can be optimised. Eventually someone asks whether the country simply needs vastly more generation.
At some point, efficiency reaches diminishing returns and abundance becomes the more interesting objective.
That distinction matters enormously for energy.
Energy is deeper than most constraints
Almost everything else in the economy is downstream of energy.
This is not a metaphor.
Steel requires energy. Cement requires energy. Aluminium requires enormous amounts of energy. Chemicals require energy. Fertiliser requires energy. Transportation requires energy. Refrigeration requires energy. Water treatment requires energy. Data centres require energy. Semiconductor manufacturing requires energy. The physical infrastructure supporting AI requires energy.
The more sophisticated the economy becomes, the more elaborate its energy conversion machinery becomes.
This is why the idea of an "immaterial economy" is misleading. Software may have extraordinarily high value relative to its physical mass, but software is executed somewhere. Somewhere there is a processor, memory, networking equipment, cooling, a building and a power supply.
The cloud is simply someone else's electricity bill.
And the more intelligence we put into machines, the larger that bill becomes.
India is therefore approaching an interesting moment. It wants to industrialise, urbanise, electrify transport, expand manufacturing, build semiconductor capacity, deploy AI at scale and raise the living standards of hundreds of millions of people. These objectives are not independent. They all draw upon the same underlying physical system.
Electricity demand is consequently not some background variable that can be assumed away. The International Energy Agency expects India's electricity demand to continue growing rapidly through 2030, driven by industrial activity, cooling, electrification and other factors.
The question then becomes uncomfortable in its simplicity.
Where does all the energy come from?
You can answer this with renewables, storage, transmission, hydro, gas, coal, nuclear or some combination of them. The engineering answer will almost certainly be a combination.
But the strategic question is different.
Do we want India to become extremely good at managing energy scarcity, or do we want India to become extremely good at producing abundant energy?
I know which country I would rather live in.
Abundance is a different economic regime
There is a subtle difference between making a scarce resource more efficiently allocated and making the resource abundant enough that allocation ceases to dominate the decision.
Suppose electricity is scarce. A factory owner has to worry about peak prices, outages and whether the grid can support expansion. A data-centre operator has to negotiate for capacity. A household has to decide how much cooling it can afford. The economy spends enormous intellectual and institutional effort deciding who gets access to a constrained resource.
Now imagine that reliable electricity becomes sufficiently abundant.
The economic question changes.
The factory owner begins asking what additional production capacity is profitable. The data-centre operator begins asking how much computation customers will buy. The chemical company begins considering processes that were previously too energy-intensive. A city can expand without treating every additional megawatt as a crisis.
The important change is not that electricity becomes free.
It is that energy stops being the first-order constraint on so many decisions.
That is what abundance means.
We sometimes use the word casually, but genuine abundance is a radical economic phenomenon. If something becomes sufficiently plentiful, behaviours that were previously irrational become rational.
Cheap computation changed software.
Cheap storage changed the internet.
Cheap transportation changed global trade.
Cheap electricity could change industry.
This is why the energy question deserves to be treated as a question about the shape of the future economy rather than merely the energy ministry's portfolio.
And this is where the atom enters
India currently has 8.78 GW of installed nuclear capacity, representing about 3.1% of electricity generation. The government's roadmap aims to reach 100 GW by 2047, with capacity expected to rise to roughly 22 GW by 2031–32 and further expansion thereafter. The plan envisages roughly 54 GW through NPCIL and the remaining capacity through other public-sector entities, state governments, private companies and joint ventures.
The number itself is interesting.
The institutional change may be more interesting.
The SHANTI Act has opened the nuclear sector to wider private participation, while the government is pursuing indigenous small modular reactors, including a 220 MWe Bharat Small Modular Reactor and a 55 MWe design, with at least five indigenous SMRs targeted for operationalisation by 2033.
This matters because nuclear power is not simply another method of generating electricity. A serious nuclear programme is a programme of industrial competence.
It requires materials science, precision manufacturing, control systems, heavy engineering, construction capability, quality assurance, specialised human capital and regulatory sophistication. It forces a country to become competent at things that are difficult, expensive and unforgiving.
The reactor is therefore only the visible object.
Behind it is an ecosystem.
And ecosystems compound.
A country that builds one sophisticated machine learns something from building it. A country that builds ten develops suppliers. A country that builds a hundred develops institutions. Universities begin producing specialists. Companies learn where the difficult parts are. Engineers become managers. Managers become founders. Suppliers move up the value chain. Standards become domestic knowledge rather than imported paperwork.
This is how industrial capability becomes self-reinforcing.
The value of the first reactor is therefore partly measured in megawatts.
The value of the hundredth reactor will also be measured in what the country learned while building the first ninety-nine.
Capitalism's great trick was experimentation
This brings us back to the old argument.
The most impressive feature of capitalism is not that it permits people to become rich. Plenty of political systems have permitted elites to become rich.
Its extraordinary achievement is that it creates a mechanism for decentralised experimentation.
Thousands of people can attempt things simultaneously.
Most will fail.
Capital gets destroyed.
Companies disappear.
Ideas are discarded.
Workers move.
Customers refuse products.
Competitors copy whatever works.
Prices transmit information no central committee possesses in its entirety.
This is an ugly process. It produces inequality, failure, waste and spectacular mistakes. But it also produces discovery.
A central planner has a difficult problem: before allocating resources, somebody must decide which future is worth investing in.
A market does not need to know.
It can finance many futures and allow reality to eliminate most of them.
This is particularly powerful when technology is changing rapidly, because nobody actually knows what the optimal future looks like.
The state therefore has an important role, but it is not necessarily to predict the future.
It can build things whose benefits are broad enough that private capital cannot capture them entirely: roads, grids, ports, universities, basic science, defence infrastructure, nuclear programmes.
Then it can allow private capital to explore the possibilities created by those foundations.
The state builds the substrate.
Entrepreneurs search the space above it.
That arrangement is much more interesting than the stale argument about whether the state or the market should "run the economy."
A modern economy needs both.
The question is what each is good at.
Marx's question survives too
There is another reason Marx remains relevant to this discussion.
He asked what happens when the productive forces of society become incompatible with the social relations surrounding them.
I disagree profoundly with much of the political machinery Marx built around that observation. History has not been kind to the idea that abolishing private capital automatically produces a more prosperous society, and the twentieth century provided rather expensive evidence that centralised ownership does not eliminate the problems of information, incentives and coordination.
But the underlying question is still alive.
What happens when technology changes the productive capacity of society faster than institutions change with it?
We are going to encounter this repeatedly.
AI may alter cognitive labour.
Robotics may alter physical labour.
Nuclear and advanced energy technologies may alter the cost structure of industry.
Biotechnology may alter medicine and agriculture.
Manufacturing automation may alter the economics of geographical labour arbitrage.
The political question will not disappear.
It will become harder.
Because abundance creates distributional questions of its own.
If machines can produce extraordinary quantities of goods with relatively little human labour, who owns the machines becomes more important, not less.
If energy becomes extraordinarily cheap, who controls the infrastructure becomes strategically important.
If AI makes individual workers vastly more productive, the ownership of the productive systems around those workers becomes consequential.
Capitalism solves the problem of discovering productive opportunities remarkably well.
It does not automatically solve the political problem of distributing the resulting power.
That is a genuine problem.
But there is a sequence here that I think matters.
It is difficult to have an intelligent argument about distribution when production itself remains desperately constrained.
You cannot divide an abundance that has not yet been produced.
So I would rather build the abundance first and argue about its distribution from a position of strength.
India has a peculiar opportunity
India is still sufficiently poor that its development problem is primarily one of increasing productive capacity.
That is an enormous advantage.
We do not have to invent the problem of post-scarcity economics yet.
We have roads to build, electricity to generate, cities to expand, factories to construct, ports to modernise, engineers to train and technologies to domesticate.
There is something almost luxurious about having such a large number of obvious things left to build.
The danger is that we become satisfied with managing scarcity rather than eliminating it.
A society that has spent generations being poor becomes extremely skilled at frugality. It learns to make things last. It learns to repair rather than replace. It learns to stretch resources. These are useful traits for a household.
They are not sufficient traits for a civilisation aspiring to become a major industrial power.
A country of 1.4 billion people cannot thrift its way into prosperity.
At some point, the denominator has to change.
More electricity per person.
More capital per worker.
More machines per factory.
More computation per engineer.
More transport capacity per citizen.
More floor space per household.
More productive land.
More scientific equipment.
More energy.
The objective should be to make the country so productive that the scarcity mindset becomes progressively obsolete.
That requires an almost uncomfortable appetite for construction.
There is a hidden hierarchy here
Think of the modern economy as a stack.
At the top are the things we notice: applications, companies, products, services, AI models, financial markets.
Below them are factories, servers, warehouses, transport networks and communications systems.
Below those are machines, materials and industrial processes.
Below those is electricity.
Below electricity is primary energy.
And beneath all of it sits physics.
The further down the stack you go, the more fundamental the constraint becomes.
You can write better software without building a new road.
You cannot run the software without hardware.
You cannot manufacture the hardware without industrial processes.
You cannot operate the industrial processes without energy.
You cannot increase energy production indefinitely without dealing with the physical properties of the technologies used to generate it.
Eventually the abstraction ends.
You meet the atom.
That is why I think the nuclear question is much larger than nuclear power.
It is a question about whether India is willing to invest in the deepest layers of its productive stack.
A semiconductor fab is impressive.
A domestic semiconductor ecosystem is more impressive.
A nuclear reactor is impressive.
A domestic nuclear industry capable of designing, financing, manufacturing and deploying reactors at scale is more impressive.
A large data centre is impressive.
A country capable of supplying the energy, chips, networks and engineers required to operate thousands of them is more impressive.
The pattern is the same everywhere.
Capability matters more than the individual asset.
We should stop being afraid of abundance
There is a peculiar tendency in wealthy societies to discuss consumption as though consumption itself were the enemy.
Of course waste is bad. Pollution is real. Externalities matter. Resources are finite. Environmental constraints are not imaginary.
But there is a danger in turning those observations into a philosophy of permanent scarcity.
A billion people becoming richer will consume more.
That is not a moral failure.
It is the objective.
The correct technological response to rising demand is not necessarily to demand less forever. It is to become better at supplying what people want with fewer external costs.
If people want cooling, invent better cooling.
If they want mobility, build cleaner mobility.
If they want electricity, generate more electricity.
If they want computation, build more efficient computation.
If they want food, develop better agricultural systems.
If they want water, develop better desalination.
The history of civilisation is largely the history of replacing constraints with technology.
We should continue doing that.
The environmental argument for abundance is therefore stronger than the romantic argument for austerity.
A wealthy society can afford sophisticated environmental controls, cleaner infrastructure, better waste treatment, higher-quality buildings and expensive research.
A poor society is much more likely to burn whatever is immediately available.
Abundance creates room for better choices.
The next century belongs to constraint removers
This is ultimately what I take from the nuclear announcement, and from Build Baby Build.
The interesting question for India is not whether the government can announce enough schemes.
It is whether the country can develop an institutional culture that continuously removes constraints.
If electricity is scarce, build generation.
If transmission is scarce, build transmission.
If engineers are scarce, train engineers.
If capital is scarce, deepen capital markets.
If manufacturing is weak, build industrial ecosystems.
If technology is imported, learn to manufacture it.
If manufacturing is expensive, automate it.
If computation is expensive, build compute.
If compute consumes too much energy, solve the energy problem.
Then repeat.
This is how a civilisation compounds.
Not by discovering one perfect policy.
By becoming unusually good at solving the next bottleneck.
The nuclear programme is interesting because it attacks one of the deepest bottlenecks available to us. India's stated objective of reaching 100 GW by 2047, combined with private participation, indigenous reactor development and SMRs, is therefore worth watching not merely as an energy programme but as a test of whether India can develop the institutional and industrial competence required for large-scale, technically demanding projects.
If it succeeds, the electricity will be the obvious outcome.
The less obvious outcome will be everything built because that electricity exists.
Factories that would otherwise not be economical.
Data centres that would otherwise be constrained.
Industrial processes that would otherwise remain marginal.
Cities that can support greater density.
Research that becomes affordable.
Businesses that can assume reliable power as a given rather than a strategic concern.
And eventually, perhaps, a generation of engineers who grow up believing that building difficult physical systems is simply what India does.
That is the real prize.
The reactor is merely the machine that makes the next set of machines possible.
Which brings us back to the original question.
What happens when you start thinking seriously about the physical limits of productive capacity?
You stop thinking about development as a contest over how efficiently to divide scarcity and start thinking about how aggressively to remove the constraints that create scarcity in the first place.
You start looking at roads differently.
At factories differently.
At semiconductors differently.
At energy differently.
At nuclear power differently.
You begin to see that the history of economic growth is, to a remarkable extent, the history of humans discovering that some constraint they had accepted as permanent was merely technological.
The fire gave us heat.
The steam engine gave us mechanical power.
Electricity gave us an entirely new industrial substrate.
Computers gave us cheap information processing.
AI is beginning to give us cheap cognitive labour.
The next great constraint will eventually be attacked too.
Perhaps it will be fusion.
Perhaps advanced geothermal.
Perhaps something nobody has named yet.
It doesn't matter.
The objective is not to worship a technology.
The objective is to become the kind of civilisation that can exploit the next one.
For India, that means building an economy with enough capital, energy, infrastructure, technical talent and institutional competence that the frontier keeps moving outward faster than scarcity can close in.
That is a much larger ambition than becoming merely richer.
It is becoming more capable.
And capability is the one form of national wealth that compounds into almost everything else.
So yes, atoms must split.
Not because nuclear power is the destination.
Because the civilisation that refuses to attack its deepest constraints eventually becomes very good at living within them.
I would rather see India become very good at breaking them.