India's Semiconductor Opportunity: Can India Become a Global Chip Hub?
India now packages chips commercially and designs some of the world's most advanced ones. Whether it becomes a chip hub depends on the layers it does not yet own.
Summary
- By September 2026 three Indian semiconductor units are in commercial production, all in assembly and test, while the first commercial fab at Dholera now targets mid-2028.
- India's real advantage is design talent, about a fifth of the global chip design workforce, but most of that work is done for foreign companies that own the resulting IP.
- Our view: India can become an indispensable node in packaging, design services and mature-node chips, and investors should back the layers around the fabs rather than wait for them.
On 28 February 2026 Micron's assembly and test plant at Sanand began commercial production, and its first memory modules went to Dell. Kaynes Semicon followed at the end of March. CG Semi announced commercial production in July. After years of announcements, India now has factories shipping semiconductor products to paying customers.
That is real progress, and it is also the easy part. The harder question for anyone putting money into semiconductor India is which parts of the chain the country will actually own, and on what timeline.
Where India stands, layer by layer
India's semiconductor position by layer of the value chain
| Layer | Status in September 2026 | Our assessment |
|---|---|---|
| Chip design | About 20% of the global design workforce; Qualcomm taped out a 2nm design from India in Feb 2026 | World class in skill, thin in owned IP |
| Design startups | 24 startups in the DLI scheme; 16 tape-outs and 6 chips fabricated | Early, promising, capital-light |
| Assembly and test | Micron, Kaynes and CG Semi in commercial production | Real and scaling |
| Wafer fabrication | Tata-PSMC Dholera under construction; commercial operations targeted mid-2028 | The test that matters most |
| Materials and equipment | Largely imported | The deepest gap |
The design paradox
India's strongest card is one that rarely makes headlines. The IT ministry told Parliament in March 2026 that the country hosts nearly 20% of the world's chip design workforce. Qualcomm, Intel, AMD, Nvidia and MediaTek all run large engineering centres in Bengaluru, Hyderabad and Chennai.
In February 2026 Qualcomm announced it had completed the tape-out of a 2nm design using its Indian teams. Chips at that node are made only by a few foreign foundries, led by TSMC in Taiwan. That is the paradox in one example: the thinking happens in India, the manufacturing happens elsewhere, and the patents, margins and customer relationships sit with a company headquartered in California.
India rents out its best chip engineers and keeps very little of what they create.
Fixing that is harder than building a factory. It needs Indian-owned companies that design chips, sell them under their own name and survive long enough to earn back years of development cost.
The first signs of owned IP
The Design Linked Incentive scheme is the government's attempt at that problem. It pays part of a startup's design costs and gives access to expensive design software and prototyping. According to PIB, 24 startups are in the programme, and supported firms have completed 16 tape-outs and fabricated six chips. Separately, the ChipIN centre at C-DAC gives more than 300 academic institutions access to the same tools.
These numbers are small, and that is fine. A chip product company takes five to ten years to reach scale. Saankhya Labs in Bengaluru, which built software-defined radio chipsets and was bought by Tata group's Tejas Networks in 2022, shows the path: design in India, sell a product rather than engineering hours, and end up as a strategic asset for a larger Indian group.
Manufacturing: packaging first, fab later
India chose to enter manufacturing through assembly and test, and we think that was right. Packaging plants cost a fraction of a fab, the technology can be licensed, and they reach production in a few years. They also build the operating habits a fab needs: clean-room discipline, yield management and qualification with global customers.
The wafer fab is a different order of difficulty. Tata Electronics and Taiwan's PSMC are building a ₹91,000 crore fab at Dholera, and the government notified a special economic zone for it in April 2026. In July, Tata said the plant would start with 55nm and 90nm processes before moving to 28nm, and the IT minister put commercial operations at mid-2028. That is later than the late-2026 date that was widely reported, and at older nodes than many assumed.
We do not read that as failure. Mature nodes are what cars, power electronics and industrial equipment use, and every first fab in a new country has slipped. But investors should price Dholera on a 2028 revenue start, not on press-release dates. The full cluster picture is in our piece on Micron Sanand and the Gujarat chip cluster.
Why global companies are looking now
The timing is not luck. US export controls on chips and chipmaking tools to China, the pandemic shortages and repeated worries about Taiwan have pushed every large chip company to spread its supply base. India offers engineers, a large home market and state support on a scale few alternatives match. That is why Micron, PSMC and Renesas, through its partnership with CG, chose to commit here rather than only announce interest.
The gaps that still decide the outcome
- Materials and equipment. Specialty gases, chemicals, wafers and tools are almost entirely imported. ISM 2.0, cleared in July 2026 with an outlay of ₹1,27,500 crore, now targets suppliers of these inputs for the first time.
- Operating experience. A fab runs on thousands of engineers who have done it before. India has design engineers in depth and process engineers in hundreds, not thousands.
- Customers for Indian chips. Designing a chip is one hurdle; getting an automotive or telecom buyer to qualify it is a slower one. Many DLI startups now face the second hurdle.
- Patient capital. Chip startups cash for years before revenue. Indian venture funds have little history with that profile.
So can India become a chip hub?
Not in the Taiwan sense, and not this decade. Taiwan's position took forty years and a supplier base that cannot be copied on a policy timeline. But the better question is whether India can be one of the few places the global industry cannot do without in specific layers. On that, we think the answer is yes for packaging and test, yes for design services, and possible for mature-node fabrication if Dholera runs well.
The domestic market helps. IESA and Counterpoint project India's own semiconductor consumption at around $103 billion by 2030, and that demand gives local suppliers a base that export-only hubs never had. The mission behind all this is set out in our note on the India Semiconductor Mission.
What to do if you are an investor or founder
- Back the layers around the fab. Packaging, test, clean-room services, specialty chemicals, gases and precision components have shorter build times and clearer customers than a fab.
- Look for owned IP. Among design companies, prefer those selling a chip or licensable block over those billing engineering hours to foreign clients.
- Model the delay. Any plan tied to Dholera output should assume a 2028 start and a slow ramp.
- Check subsidy dependence. Central support covers up to half of approved project costs for ISM units. Ask what the look like without it, because the next phase of support will not be as generous.
Before committing capital, a proper due diligence process should test all four points against the company's contracts and capacity, not its deck. For founders building in this space, our deep-tech practice and VC and PE advisory teams work on exactly this kind of raise.
How useful was this article?
One tap. It tells us what to write more of.
About the author
SRF Capital Studio
The next one
Get what we publish next, by email.
Working notes on raising, borrowing, protecting, growing and structuring capital in India. One email a week at most, and you can leave any time.
We use your address only to send this. See our privacy policy.
We store your address to send you these emails and nothing else. See our privacy policy.
Related reading
