Innovation Is Geographically Rigged

The first version of this essay had a much harsher thesis:

Indians are not allowed to innovate.

Not by law. Not because someone explicitly stops us. And definitely not because Indians are somehow less intelligent.

We are stopped, or at least slowed down, by something much harder to identify: our environment.

I live in Kanpur. When I compare what it feels like to build from a city like this with what I imagine it feels like to build in San Francisco, the difference is not just access to money or better offices. It is the kind of conversations you are allowed to have without first having to justify yourself.

In one environment, saying that you want to build a new kind of AI company can immediately lead to questions about the product, the model, the market and the speed at which you can ship it.

In another, the idea must first survive a more basic interrogation. Is this a real career? How much will it pay? Why would you leave a stable job? Has somebody already done it? What happens if it fails?

These are not foolish questions. In a country where failure is expensive and family security matters, they are often rational. But rational questions, repeated often enough, can still produce a conservative culture.

This does not mean everyone in San Francisco is imaginative or everyone in Kanpur is cautious. It means one place gives unusual ambition more social permission than the other.

Innovation begins long before somebody builds a product. It begins when an environment allows a strange idea to be discussed seriously.

Innovation is an ecosystem pretending to be an individual achievement

We like to tell stories about lone geniuses.

A founder has an idea. An engineer invents something in a garage. A college dropout sees the future before everyone else does.

These stories are attractive because they make innovation look like a personality trait. Find an intelligent and sufficiently rebellious person, give them a laptop, and eventually they will produce the future.

But innovation does not work like that.

It is a supply chain. It needs universities doing research, governments willing to finance uncertain work, investors who can lose money, companies willing to become early customers, skilled people who can change jobs, laws that make ownership legible, and peers who treat building as a normal thing to do.

The geography of innovation makes this visible. According to the World Intellectual Property Organization, the world's top 100 innovation clusters account for roughly 70 percent of international patent filings and venture-capital activity, along with about half of all scientific publications. The San Jose and San Francisco cluster ranks first in innovation intensity. [1]

If innovation were only a measure of individual intelligence, it would not be this geographically concentrated.

The person matters. But the system changes the probability of what that person can become.

This is why I think innovation is geopolitically driven.

Countries decide who gets access to research funding, advanced chips, energy, laboratories, defence contracts, universities, immigration pathways and large pools of capital. Those decisions determine where frontier technology can realistically be built.

Even Silicon Valley was not created by a few people spontaneously deciding to think differently. Stanford benefited from enormous federal research and defence funding after the Second World War. The surrounding industrial ecosystem grew through military contracts, university research, immigration, private capital and decades of institutional support. A National Academies history of Stanford and Silicon Valley puts the role of federal investment plainly: neither the university nor the region could have grown as strong as quickly without it. [2]

The mythology is private. The foundation was geopolitical.

Where my own argument goes too far

My instinct was to say that India can never be at the forefront of new technology. That Indians in India are capable only of reproducing what somebody else has already created.

It is a satisfying sentence when you are frustrated.

It is also wrong.

India has four clusters among the world's top 100: Bengaluru, Delhi, Mumbai and Chennai. WIPO describes India as an innovation overperformer relative to its inputs and one of the fastest climbers in its rankings since 2013. [3]

India clearly can innovate. The harder and more useful question is why that ability is converted into frontier technology so inconsistently.

India's own policy research identifies the structural gap. A 2026 NITI Aayog report puts Indian research and development spending at approximately 0.65 percent of GDP, compared with roughly 3.5 percent in the United States, 2.4 percent in China and 4.5 percent in South Korea. It also points to low researcher density, administrative friction and limited institutional capacity. [4]

That is not an intelligence problem. It is a conversion problem.

India produces an extraordinary amount of human capability. What it struggles to produce is an environment that compounds that capability.

An Indian does not suddenly become more intelligent after boarding a flight to San Francisco. What changes is the probability landscape around them. They enter a place with a greater concentration of capital, potential co-founders, specialised talent, early customers and people who will entertain an idea before demanding proof that it is safe.

The person may be the same. The possible versions of that person are not.

Tier-II and Tier-III India pays a higher innovation tax

The situation is not uniform inside India either.

Someone building in Bengaluru already operates closer to capital, talent and startup culture than someone doing the same thing from a smaller city. A person in a Tier-II or Tier-III city has to assemble much more of their environment manually.

They find their peers online. They teach themselves from videos, documentation and repositories. They explain unconventional career choices to people who may never have seen them work. They build without knowing whether the problem is interesting to anybody outside their immediate circle. Even discovering the right question can take longer when the relevant conversations are not happening around you.

I think of this as an innovation tax.

San Francisco subsidises unusual ambition. Bengaluru taxes it less heavily than most Indian cities. Tier-II and Tier-III India often makes you pay upfront.

There is evidence that this is changing. The Indian government says roughly half of all DPIIT-recognised startups now originate from Tier-II and Tier-III cities. That is meaningful. It shows that entrepreneurship is no longer restricted entirely to a handful of metros. But a startup registration is not the same thing as access to frontier research, patient capital or a dense technological network. [5]

Participation is spreading faster than power.

Copying, wrappers and the misunderstood middle

Part of the problem is that we use the word innovation too loosely.

Invention creates a new technological primitive. Innovation turns a primitive into something useful. Adaptation makes it work for a particular market. Distribution ensures that it reaches people.

These are different forms of work, but startup culture often treats only the first one as intellectually respectable.

We saw the same debate with AI wrappers. A company would build a product on top of an existing large language model, and people would dismiss it as "just a wrapper." Some of those products were shallow. Others turned a general capability into a useful workflow, solved a real problem and built successful businesses.

Users rarely care which layer of the stack contains the most impressive research. They care whether the complete product works.

In that sense, I think the internet is the foundational model on top of which an entire generation of Indian builders is creating wrappers.

That is not an insult.

Every technology is built on another technology. An AI application depends on a model. The model depends on chips, data, research and energy. A fintech product depends on banking rails. An online business depends on protocols, payment systems, logistics and distribution networks that it did not invent.

Building on top of something is not the opposite of innovation. It is how most innovation happens.

India may not invent every foundational model or manufacture every advanced chip. It can still create enormous value by taking global technological primitives and making them useful under Indian constraints: lower purchasing power, multiple languages, inconsistent infrastructure, enormous scale and radically different user behaviour.

Sometimes the layer itself becomes foundational. UPI is the obvious counterexample to the idea that India can only copy. It is interoperable digital public infrastructure built for the particular complexity and scale of the Indian market. By 2025, it was processing more than 19 billion transactions a month and had become the world's largest real-time payment system by volume, according to the International Monetary Fund. [6]

That is not a copy of Silicon Valley. It is a different kind of frontier.

At the same time, we should not use application-layer success as an excuse to abandon foundational research. Making a useful product with somebody else's model is valuable. Creating the next model is valuable too. A serious country should want the capacity to do both.

The internet changes the odds, not the starting conditions

For someone in a smaller Indian city, the internet is more than a communication network.

It is a university, a laboratory, a professional network, a distribution channel and occasionally an escape route.

A builder in Kanpur can access the same open-source repository as someone in California. They can call the same API, read the same paper, contribute to the same project and sell to a customer they may never meet. That would have been impossible for almost every generation before ours.

But the internet has not made geography irrelevant.

Research using worker-location and patent data in Silicon Valley found substantial knowledge spillovers from face-to-face meetings, including seemingly accidental interactions. The NBER paper gives evidence to something builders already understand intuitively: sometimes the valuable part of being in the right place is hearing the right sentence from somebody you did not know you needed to meet. [7]

The internet reduces information distance. It does not completely eliminate social, financial or institutional distance.

Still, it gives us a way to participate.

An Indian builder does not have to wait for the country to develop the complete infrastructure of San Francisco before creating anything. We can use what already exists. We can build for our own problems, sell into global markets, form online communities and gradually create local pockets in which ambitious conversations become normal.

The internet is not a level playing field. But it is an opening in the wall.

The thinker-builder trap

I keep coming back to the difference between a thinker and a builder.

A thinker can develop a beautiful explanation of the world. A builder has to collide that explanation with reality.

The market can reject it. The code can fail. Users can behave differently from the theory. Building creates feedback, and feedback is what prevents intelligence from becoming self-admiration.

This does not mean thinking is useless. In fact, as AI makes execution cheaper, analytical judgment becomes more valuable. When almost anyone can generate code, copy a landing page or automate a workflow, the scarce ability is knowing what deserves to exist, how the pieces connect and what consequences will follow.

The real distinction is not thinker versus builder.

It is thinking that enters reality versus thinking that never has to risk being wrong.

Thinking without building can become performance. Building without thinking becomes noise. The people who matter will increasingly be able to do both.

That is why I struggle with people who respond to every technological shift with abstract declarations about human nature. They say they choose humans over technology, as if the two are opposing teams and one of them must win.

Technology is a human creation. It contains human decisions, incentives and values, only with greater leverage.

For the rest of my life, and probably throughout the lives of most people my age, technology will remain one of the main forces reorganising work, culture, politics, relationships and power. AI will not be the last major technological shift we experience. There will be advances in robotics, biology, energy, computing and fields we do not yet have names for.

You do not have to worship that future. But refusing to understand it does not make you more human. It only gives you less influence over what it becomes.

And no, Marcus Aurelius was not wrong for spending his life thinking about character, duty and mortality. He used the intellectual tools available to him to understand his time.

But quoting Marcus Aurelius while refusing to understand the defining machinery of our own time is not depth. It is nostalgia dressed as wisdom.

We still need philosophy. The more powerful technology becomes, the more urgently we need to ask what it should do, who it should serve and what must never be outsourced to it. But those values have to enter our products, companies, policies and code. Otherwise, they are decoration.

The next important thinker may also need to be a builder.

Permission can be built too

I do not know whether India will lead the next foundational technological wave.

Right now, the capital, compute, laboratories and institutions required to build at the frontier remain heavily concentrated. The 2026 Stanford AI Index reports that American private AI investment reached $285.9 billion in 2025. That scale of capital changes which experiments can even be attempted. [8]

So it would be dishonest to say that a sufficiently motivated person with a laptop can simply ignore geography.

But it is equally dishonest to turn a structural disadvantage into a permanent statement about Indian capability.

India does not lack intelligent people. It lacks enough environments in which intelligent people can be strange, speculative and wrong for long enough to make something original.

That can change.

Every person who builds from a smaller city makes building slightly more legible to the next person. Every serious online community reduces the isolation. Every company that hires remotely redistributes access. Every founder who solves an Indian problem without waiting for foreign validation expands the set of ambitions that appear reasonable.

We do not need to recreate San Francisco everywhere. We need more rooms in Kanpur, Indore, Jaipur, Kochi, Patna and hundreds of other cities where asking "What are you building?" feels as ordinary as asking "Where do you work?"

My original sentence was that Indians are not allowed to innovate.

The more accurate version is this:

Indians are not incapable of innovation. We are operating in environments that charge a higher price for unusual ambition.

The internet helps us pay that price. Building helps us reduce it for the person who comes next.

San Francisco's most valuable product may not be a company or a technology. It may be permission.

Our job is to manufacture more of it wherever we are.

Innovation has an address. The internet gives us a chance to change it.