India’s semiconductor push opens new opportunities for startups

For years, India’s semiconductor ambition was mostly a slide in a policy deck. What has changed is not the ambition but the evidence: physical facilities, commissioned lines, first silicon, and a supplier base beginning to form around them. From where I sit, the more useful shift is subtler than any single groundbreaking: the sector has crossed from announcement into execution, and execution is where the interesting company-building begins.

Consider the last six months. Micron’s Sanand assembly and test facility was inaugurated in February 2026.

Kaynes Semicon reached commercial production in March 2026, fourteen months after breaking ground. CG Semi began commercial output at Sanand in July 2026. In a sector where ecosystems take a decade, three plants already live in half-a-year. The Tata-PSMC fab at Dholera is the harder test – front-end fabrication is a different order of difficulty from packaging. But the pattern tells you something: the time from breaking ground to first output is compressing. The supporting muscles – construction, cleanroom fit-out, utilities, logistics, trained operators – are maturing faster than assumed.

That compression is what makes the demand numbers investable rather than aspirational. Deloitte projects India’s semiconductor consumption market reaching ~$120B by 2030 and ~$300B by 2035. However, India imports over 90% of those chips today. The whole opportunity sits in that gap. It isn’t a bet on inventing demand, but instead that domestic supply takes a share. It is also why the ₹1.25 lakh crore cleared for the second phase of the India Semiconductor Mission (ISM) matters more for the broader ecosystem: equipment, materials, indigenous IP, supply chains.

And, this is where the most durable businesses will be built. A fab is a landmark; the economy around it is the market. Specialty chemicals, semiconductor-grade gases, cleanroom consumables, precision tooling, pre-silicon verification, test infrastructure, custom chip vendors – each is a category India largely imports or lacks today, and each becomes standing domestic demand as capacity comes up. In a capital-intensive, qualification-gated industry, that is one of the more predictable demand structures anyone can underwrite over a decade – and once a material is designed into a proven line, displacing it is slow and costly, so suppliers who clear that bar early tend to hold the position for years.

Ask what pulls that demand, and the answer increasingly points one way: industrial AI. The next wave of AI isn’t in the cloud – it is moving onto factory floors, into robots, drones, machine-vision cameras and industrial IoT, each needing intelligence that runs locally on constrained edge silicon. That is a structural pull on exactly the chips India is starting to make, and it opens a set of adjacencies alongside them. The one I find most interesting is software: silicon becomes a finished product only once firmware, drivers, and optimised models sit on top of it. For a country with deep software talent, that is a space worth building in – and one reason we backed CraftifAI, a Bengaluru-based company automating that embedded layer across different chips. The point isn’t any single product; it is that whoever builds this layer shapes how quickly India’s new silicon reaches the field.

All of it is drawing out founders, and they are different from the ones we met three or four years ago. Increasingly, they are coming from inside the industry itself. Some are operators stepping out of companies like Applied Materials and MediaTek – we are talking to a lot of them. Others come from the labs: the SHAKTI processor programme at IIT Madras produced InCore and Mindgrove, and IISc’s entrepreneurship track produced Morphing Machines. The technical depth was always there; what’s new is that procurement cycles, qualification, yield and customer integration show up in the first conversation, not year three. Designing for manufacturability before the technology is proven is the strongest leading indicator I know in this sector.

This might not fit the reflexes of generalist venture capital. First, validation takes longer, capital sits in the ground before it works, and the milestones that matter are quiet: a qualification passed, a design-in won, an anchor customer who re-orders. That mismatch filters out a lot of capital – which is exactly why the capital that stays is what these businesses need. Secondly, talent isn’t new; a fifth of the world’s chip designers already work here. What’s new is everything around it: manufacturing, supply chain, and the connective tissue between them. That tissue doesn’t exist yet, so part of the work is convening – putting a startup in front of the fab that could qualify it, the corporate that could anchor its first order – that’s how this ecosystem compounds.

Which is the note I’d leave on. India’s semiconductor story has finally acquired physical proof, a widening ecosystem, a better generation of founders, and a demand pull that gives all of it a reason to exist. Multiply that across the adjacencies and the second-order effects become the real economic story: supplier depth, engineering work, export earnings, less exposure to a supply chain geopolitics has made fragile. India is expected to meet over 60% of its own chip demand by 2035, and the companies that get us there will be the ones that treated this as a long industrial buildout.

That’s a less exciting story than the headlines suggest. It is also a far more durable one. BW Disrupt