Rethinking Make in India in the age of AI

For most of industrial history, software ate everything except the factory floor. Digital transformation swept through finance, media, retail, and services, but the physical act of bending metal, mixing chemicals, For most of industrial history, software ate everything except the factory floor. Digital transformation swept through finance, media, retail, and services, but the physical act of bending metal, mixing chemicals, and assembling products remained stubbornly analog. That era is ending. In 2026, the physical world is finally becoming programmable. And this shift will determine whether Make In India becomes an unstoppable economic force or remains a policy aspiration with impressive PowerPoint decks.

Five civilizations, one collision

AI is not hitting a homogeneous global manufacturing surface. It is hitting five deeply different industrial civilizations simultaneously, each with its own philosophy of making things.

Germany is building the cathedral, a complete operating system for intelligent factories where digital twins don’t merely mirror physical plants but prescribe their behaviour, testing thousands of scenarios faster than real-time and pushing optimised parameters back to the floor. Japan is evolving its philosophy of monozukuri, the art of making things, into what we might call kinetic intelligence, where robots learn from master craftsmen through demonstration and haptics, then generalise across tasks. China is pursuing saturation, flooding thousands of factories with AI simultaneously through state-directed coordination at a velocity no market economy can match. America is running the platform play, owning the compute layer, the simulation layer, and the foundation models, then selling shovels to everyone who digs.

And India? India’s honest keyword is bifurcation. A handful of world-class digital factories coexisting with a vast MSME base that hasn’t crossed the digital threshold. Manufacturing’s share of GDP stands at 14%, against a target of 25%. The execution gap is real.

But here is the counterintuitive thesis: India’s apparent weakness could be its greatest structural advantage.

The architecture India should actually build

Science fiction writer William Gibson never imagined centralised mega-factories. He imagined distributed networks of small fabrication shops, connected and coordinated by intelligence. This is precisely the architecture that AI now makes viable.

Consider the economics. Before AI, manufacturing brutally punished small runs. The optimal batch size, the quantity at which per-unit cost bottoms out, was always well above one. That floor is now dropping toward one. AI-driven generative design takes a plain-language description of what you need and produces not just a shape but a manufacturable design complete with toolpaths and material specifications, in hours rather than weeks. The sequential pipeline of design-simulate-prototype-test-redesign-tool-up-produce is collapsing into a near-simultaneous loop.

Now apply this to India’s ten million MSMEs. A network of thousands of small manufacturers, each equipped with AI-optimised production scheduling, connected through intelligent logistics, sharing demand signals and capacity in real time, this mesh could outperform a centralised mega-factory on responsiveness while approaching it on unit cost. India should not be replicating the mega-factory model. India should be building the world’s first AI-coordinated manufacturing mesh, turning its fragmented base from a liability into an intelligent, distributed production network.

The Make in India initiative and the Production Linked Incentive schemes provide the policy scaffolding. AI provides the connective tissue that the scaffolding has always lacked.

The hard problem nobody is talking about

But beneath the optimism lurks a problem that most commentary ignores: verification.

When AI generates a novel alloy composition, an optimised factory layout, or a reconfigured supply chain, someone must verify it actually works. The physicist Stanislaw Lem wrote an entire novel, Solaris, about encountering intelligence whose outputs are purposeful but incomprehensible to humans. This is already happening in AI-driven materials discovery: algorithms propose molecular structures that perform beautifully, but no scientist can fully explain why.

Manufacturing guided by intelligence we don’t fully understand demands a new kind of governance infrastructure. Statistical process control doesn’t become obsolete when AI arrives, it becomes the essential verification layer that tells you whether the AI’s optimisations are working or quietly drifting into failure modes. The organisations that build this governance capability, rigorous, quantitative evaluation of what AI proposes before it hits the production floor, will capture the value. The rest will be flying blind with sophisticated autopilots.

This is India’s deepest opportunity. Not just adopting AI in manufacturing, but building the human and institutional infrastructure to verify, calibrate, and govern AI-driven production at scale. A nation of engineers and analysts, properly organised, is precisely the resource this moment demands.

The real race

The race that matters is not about who adopts AI fastest. China will win that race through sheer state-directed velocity. The race India can win is about who builds the most intelligent architecture for distributed, verified, continuously improving manufacturing.

The physical world is programmable now. The question is whether India writes the program or spends the next decade running someone else’s code.