India Semiconductor Timeline 2026–2030: Fabs, OSAT Plants, Tata, Micron & the Chip Manufacturing Race
India's semiconductor timeline 2026: Micron, Kaynes and CG Semi are producing chips, the Tata-PSMC Dholera fab is under construction, plus Semicon 2.0.
For years, India’s semiconductor strength sat in the part of the industry most people never see: chip design. That is starting to change. In 2026, Micron opened an assembly-and-test plant at Sanand, Kaynes Semicon began production, and CG Semi entered commercial production — three plants packaging and testing chips on Indian soil. Meanwhile Tata Electronics is building a much harder piece: a 300 mm wafer fabrication plant at Dholera with Taiwan’s PSMC. Packaging a chip in India is a real manufacturing milestone. Fabricating the silicon wafer itself is a different level of difficulty. This India semiconductor timeline tracks both, from the 2021 policy blueprint to the 2030 test of whether the ecosystem holds together.

Fact-checked: 28 August 2026. Investment figures are approved project values from Government of India and company statements, not money already spent. Every future milestone is labelled target, expected or reported. This is not investment advice.
🧠 The 60-second answer
In 2026 India moved from promising chips to producing them. Micron (28 February 2026), Kaynes Semicon (31 March 2026) and CG Semi (4 July 2026) all began assembly, packaging and test operations in Gujarat — real semiconductor manufacturing, but not wafer fabrication. The Tata–PSMC fab at Dholera, India’s first pure-play silicon foundry, is under construction with first wafers targeted for late 2026 and commercial-scale output reported for around 2028. The government has approved about 10 manufacturing units (expanded toward a dozen) worth a cumulative ₹1.64 lakh crore, and in July 2026 cleared Semicon 2.0, a ₹1,27,500 crore programme aimed at deepening the ecosystem — equipment, materials, design IP and R&D — rather than only subsidising fabs.
India’s chip push: key questions
What India’s semiconductor story looks like in 2026
- 2026 is the production year. Three plants — Micron, Kaynes, CG Semi — began commercial assembly and test operations, replacing future-tense promises with physical output.
- Packaging is not fabrication. Micron, Kaynes and CG Semi are OSAT/ATMP plants. They take fabricated wafers or dies and cut, bond, package and test them. The silicon is still made abroad.
- The wafer-fab test is still ahead. Tata–PSMC Dholera is India’s first pure-play foundry, under construction. First wafers are a late-2026 target; commercial-scale output is reported for around 2028, and the starting node may be 90 nm before 28 nm.
- PSMC is not TSMC. Tata’s fab partner is Powerchip Semiconductor Manufacturing Corporation, a Taiwanese foundry specialising in mature nodes — a different company from TSMC.
- Semicon 2.0 changes the emphasis. The ₹1,27,500 crore phase approved in July 2026 targets equipment, materials, design IP, R&D and skills — the parts of the stack a single fab subsidy does not build.
- The map is wider than Gujarat. Assam (Tata OSAT), Uttar Pradesh (HCL–Foxconn display drivers), Odisha (SiCSem silicon carbide, 3D Glass packaging), Punjab (CDIL) and Andhra Pradesh (ASIP) all have approved units.
- 28 nm is not obsolete. Mature nodes run most automotive, power-management, microcontroller and industrial chips, where reliability and long product life matter more than transistor size.
- Investment figures are approvals, not spending. The ₹1.64 lakh crore cumulative number mixes private investment, central incentives and state support across projects at very different stages.
- The Adani–Tower fab is a proposal, not an approved plant. Maharashtra cleared it at state level in 2024; it never received central approval, and Adani paused talks with Tower in 2025.
- Capacity numbers measure different things. 50,000 wafer starts a month is not 50,000 chips. 48 million packaged units a day is a different metric again. They cannot be ranked against each other.
India’s 2026 semiconductor production breakthrough
Three backend plants entered production in five months. The next test is wafer fabrication.
All three 2026 milestones are backend manufacturing — packaging and testing. None of them fabricates a silicon wafer.
How a chip is actually made
The simplified journey from a design file to a device — and where Indian projects sit.
India today is strong at the top of this chain (design) and building fast at the bottom (packaging and test). The middle step — wafer fabrication — is the hard gap Dholera is meant to close.
Fab vs OSAT vs ATMP: what’s the difference?
Fab (fabrication plant)
A fab starts with blank silicon wafers and builds the transistor structures and integrated circuits directly onto them, layer by layer, using photolithography, thin-film deposition, etching and doping. This is the most capital-intensive and technically demanding step. Example: Tata–PSMC, Dholera. SiCSem in Odisha is a fab too, but for silicon carbide rather than silicon logic.
OSAT (Outsourced Semiconductor Assembly and Test)
An OSAT facility receives fabricated wafers or dies and performs the backend: wafer dicing, die attach, wire bonding or flip-chip, encapsulation and final testing. It is real semiconductor manufacturing, with its own yield and reliability challenges, but it does not create the circuits. Examples: CG Semi, Kaynes Semicon, Tata Assam.
ATMP (Assembly, Testing, Marking and Packaging)
ATMP is the term Indian policy documents often use for the same backend activity, sometimes with the marking (laser-etched part numbers) step called out. Example: Micron Sanand, which packages and tests memory. Functionally, ATMP and OSAT describe the same part of the chain.
The key point: OSAT and ATMP production is genuine manufacturing and a real milestone for India. It is still not the same thing as fabricating the wafer.
India semiconductor timeline: 2021 → 2030
Newest first. Forward items are labelled target or expected.
2026
Cabinet approves Semicon 2.0
The government approved the second phase of the India Semiconductor Mission with an outlay of ₹1,27,500 crore. Its stated pillars span chip design and IP, equipment and materials (machinery, specialty chemicals, industrial gases), fabs (silicon, compound, discrete and display), ATMP and advanced packaging, R&D aimed beyond 28 nm, and talent. The FY2026–27 Union Budget had earlier set aside ₹1,000 crore toward launching the phase — an annual provision, separate from the total programme outlay.
2026
CG Semi enters commercial production
The CG Semi OSAT facility began commercial production at its G1 line. The joint venture pairs CG Power (Murugappa Group) with Renesas Electronics of Japan and Stars Microelectronics of Thailand. Approved project value is about ₹7,584 crore, with planned capacity building toward roughly 15 million chips a day across G1 and G2, for automotive, industrial and consumer applications.
ASML equipment deal; two more units approved
Tata Electronics signed a lithography-equipment agreement with ASML (16 May 2026) for the Dholera fab — a reminder that even a domestic fab depends on the global tool base. Around the same time the government approved two further units: Crystal Matrix (compound semiconductors and displays, Dholera) and Suchi Semicon (discrete-device OSAT, Surat). SiCSem also signed its firm fiscal-support agreement with the mission (11 May 2026).
2026
Dholera fab gets its SEZ notification
The Centre formally notified a 66-hectare Special Economic Zone in the Dholera Special Investment Region, with Tata Semiconductor as developer and around 21,000 jobs cited. Construction of the fab continued through 2026. Reports that the plant is “half complete” with a late-2026 pilot are project projections, not independently verified.
2026
Kaynes Semicon starts production
The Kaynes Semicon OSAT plant at Sanand was inaugurated and began production, about 14 months after groundbreaking. Approved value is around ₹3,307 crore. The plant makes intelligent power modules and packaged devices for EV, industrial and consumer use, ramping from roughly 1.5 million units a day toward a planned peak near 6.3 million a day.
2026
Micron opens the Sanand assembly-and-test facility
Micron inaugurated India’s first semiconductor assembly and test facility. The plant converts DRAM and NAND wafers fabricated elsewhere in Micron’s global network into finished memory and storage products. Micron expects tens of millions of units assembled and tested at Sanand in 2026, rising to hundreds of millions in 2027. The total project structure is about ₹22,516 crore (around $2.75 billion as announced in 2023), combining Micron investment with central and Gujarat government support.
2026
HCL–Foxconn breaks ground in Uttar Pradesh
Foundation work began on the HCL–Foxconn unit near Jewar, approved in May 2025 at about ₹3,706 crore. It is an ATMP plant focused on display driver chips for phones, laptops, TVs, wearables and vehicles — not a logic fab. Planned capacity is around 20,000 wafers a month, or tens of millions of packaged units a month. Operations are targeted for 2027.
Four more units approved: Odisha, Punjab, Andhra Pradesh
The government approved four projects: SiCSem (Bhubaneswar, Odisha — India’s first commercial silicon carbide fab plus packaging, ~₹2,067 crore, with UK partner Clas-SiC); 3D Glass Solutions (Odisha — glass-substrate advanced packaging, ~₹1,943 crore); CDIL (Mohali, Punjab — discrete power-semiconductor expansion); and ASIP (Andhra Pradesh — system-in-package assembly, with Korean partner APACT).
→25
India moves from policy to construction
In February and September 2024 the Cabinet approved the first big wave: the Tata–PSMC Dholera fab, the Tata assembly-and-test plant in Assam, the CG Power OSAT and, in September, Kaynes. In May 2025 it approved HCL–Foxconn. Together these projects account for the bulk of the ₹1.64 lakh crore cumulative figure.
India Semiconductor Mission launched
The government launched the India Semiconductor Mission with a ₹76,000 crore outlay, offering fiscal support for fabs, compound-semiconductor and packaging units, and a display fab, alongside a design-linked incentive scheme. Early proposals stalled — a Vedanta–Foxconn plan collapsed — before the 2024 approvals gave the programme real projects.
2027–2028: the wafer-fab test
Forward-looking. Dates are company and government targets and can move.
The next phase is about Tata Dholera and the other fab projects moving from construction to qualified production. Tata and government messaging points to first or trial wafers by late 2026 and a commercial ramp across 2027–2028. Some industry reporting is more cautious: that the fab may start at 90 nm rather than 28 nm and that commercial-scale output slips to around mid-2028. Both are targets, not facts. The measurable questions for this window are equipment installation, process qualification, customer qualification and yield — not ribbon-cuttings. SiCSem’s silicon carbide fab in Odisha is targeting first wafers around the end of 2026 on a similar “target” basis.
What would ecosystem maturity look like by 2030?
Measured in output, not announcements.
2030 scorecard — what to measure
- Number of fabs actually in sustained commercial production, and at what yield
- Total wafer starts per month across all operating fabs
- Number of OSAT and ATMP plants at or near rated capacity
- Advanced-packaging capability — whether any Indian plant runs 2.5D or 3D integration in volume
- Indian-designed chips actually in production, not just taped out
- Local suppliers of specialty gases, chemicals, photoresists and equipment services
- Export customers who have qualified Indian-made parts
- Repeat private investment — second units, expansions, new entrants without fresh subsidy
- A working fabless startup ecosystem shipping products, not prototypes
Notice what is not on that list: a specific process node. Semicon 2.0 can strengthen equipment, materials, design and packaging without any approved roadmap committing India to sub-20 nm production by 2030 — and no such commitment currently exists.
India semiconductor plant tracker — August 2026
Approved units under the India Semiconductor Mission. Values are approved project cost, not money spent.
| Company / JV | Location | Type | Focus | Approved ₹ cr | Status |
|---|---|---|---|---|---|
| Micron | Sanand, Gujarat | ATMP | DRAM / NAND memory packaging & test | 22,516 | 🟢 Commercial production (Feb 2026) |
| Kaynes Semicon | Sanand, Gujarat | OSAT | Power modules, auto / industrial | 3,307 | 🟢 Commercial production (Mar 2026) |
| CG Semi | Sanand, Gujarat | OSAT | Auto / industrial / consumer packaging | 7,584 | 🟢 Commercial production (Jul 2026) |
| Tata–PSMC | Dholera, Gujarat | Fab (300 mm) | 28–110 nm logic, PMIC, display, MCU | 91,526 | 🟠 Under construction |
| Tata (TSAT) | Morigaon, Assam | OSAT / ATMP | Auto, EV, communications, consumer | 27,120 | 🟠 Under construction |
| HCL–Foxconn | Jewar / YEIDA, UP | ATMP | Display driver ICs | 3,706 | 🟠 Under construction |
| SiCSem | Bhubaneswar, Odisha | Fab + ATMP | Silicon carbide (compound semi) | 2,066 | 🟠 Approved, early build |
| 3D Glass Solutions | Bhubaneswar, Odisha | Advanced packaging | Glass substrates, 3D integration | 1,943 | 🟣 Approved |
| CDIL | Mohali, Punjab | Discrete | Power semiconductors (Si & SiC) | ~117 | 🟣 Approved (expansion) |
| ASIP | Andhra Pradesh | OSAT | System-in-package assembly | ~480 | 🟣 Approved |
| Crystal Matrix | Dholera, Gujarat | Compound fab + ATMP | GaN, mini / micro-LED | part of 3,936 | 🟣 Approved (May 2026) |
| Suchi Semicon | Surat, Gujarat | OSAT | Discrete devices | part of 3,936 | 🟣 Approved (May 2026) |
🟢 commercial production 🟠 under construction 🟣 approved / not yet building CDIL and ASIP values are approximate; the four August-2025 units total about ₹4,594 crore.
Major proposals not equivalent to approved plants
| Project | Location | Reported value | Central ISM approval? | Status |
|---|---|---|---|---|
| Adani – Tower Semiconductor | Panvel, Maharashtra | ~₹83,947 crore | No — appraisal stage only | State-cleared 2024; Adani paused talks with Tower in 2025; no revival confirmed |
This distinction matters. A state cabinet clearance and a memorandum of understanding are not the same as central approval under the semiconductor programme, and this project has never had the latter. It should not be counted among India’s approved fabs or described as under construction.
Tata–PSMC Dholera: India’s biggest fab bet
The Tata Electronics–PSMC plant at Dholera, Gujarat, is India’s first pure-play semiconductor foundry — a fab that manufactures chips designed by customers rather than selling its own branded products. The customer designs the chip; the foundry makes it.
Technology partner: Powerchip Semiconductor Manufacturing Corporation (PSMC) of Taiwan, a mature-node specialist. This is not TSMC — a different Taiwanese company with a similar-sounding name. Wafer size: 300 mm (12-inch). Planned capacity: up to 50,000 wafers per month. Technology range: 28 nm to 110 nm, covering power-management ICs, display drivers, microcontrollers and high-performance-computing logic, for automotive, computing, communications and data-storage customers. The approved project value is ₹91,526 crore (Tata’s own wording is “up to ₹91,000 crore”), roughly $11 billion.
Status, August 2026: under construction. A 66-hectare SEZ was notified in April 2026 and an ASML lithography-equipment agreement signed in May. Claims that the fab is around half built with a late-2026 pilot are project projections. First wafers are a target for late 2026; industry reporting suggests the fab may open at 90 nm and reach commercial scale around 2028. Until wafer production actually begins, the correct status is under construction, not operating.
Why Assam matters to India’s chip strategy
The Tata Semiconductor Assembly and Test facility at Morigaon (Jagiroad), Assam, is an OSAT / ATMP plant — assembly, packaging and test — not a wafer fab. Approved value is about ₹27,120 crore, with a planned full-scale capacity of 48 million packaged units a day — 4.8 crore units — serving automotive, EV, communications and consumer markets. That figure is a target for full ramp, not current output.
An earlier government roadmap expected Phase 1 to begin around April 2026. As of August 2026, the minister’s language is that the plant is “likely to begin operations within FY2026–27” — commissioning has not been confirmed as complete, and there is no verified first-chip event. The strategic value is regional: it puts a large, skilled-manufacturing employer in the north-east and diversifies the packaging base beyond Gujarat.
Beyond Gujarat: the wider semiconductor map
HCL–Foxconn, Jewar
India’s first display-driver-focused ATMP plant, near the new Noida International Airport. Approved May 2025 at ~₹3,706 crore; construction started February 2026; operations targeted 2027.
SiCSem + 3D Glass, Bhubaneswar
SiCSem is India’s first commercial silicon carbide fab plus packaging (~₹2,067 crore, UK partner Clas-SiC). 3D Glass Solutions builds glass-substrate advanced packaging for RF, defence, HPC and photonics.
CDIL, Mohali
An expansion of Continental Device India’s existing discrete-semiconductor plant, adding MOSFETs, IGBTs, Schottky diodes and transistors in both silicon and silicon carbide.
ASIP
A system-in-package assembly unit with South Korean partner APACT, approved August 2025, adding another backend node in a new state.
Semicon 1.0 vs Semicon 2.0
What the second phase changes.
| Semicon 1.0 (2021) | Semicon 2.0 (2026) | |
|---|---|---|
| Main aim | Create an initial manufacturing base | Deepen the ecosystem around it |
| Outlay | ₹76,000 crore | ₹1,27,500 crore |
| Manufacturing | Fabs and packaging units | Silicon, compound, discrete and display fabs; advanced packaging |
| Design | Design-linked incentive scheme | Stronger push on full-stack Indian chip IP |
| Equipment & materials | Not a focus | Named pillar — machinery, chemicals, gases |
| R&D | Limited | Frontier tech beyond 28 nm; global institution tie-ups |
| Talent | University programmes | Industry-led research and training centres |
The strategic logic: subsidising one more fab adds capacity, but does not create the tool vendors, gas suppliers, materials makers and IP houses a real semiconductor cluster runs on. Phase 2 is aimed at those gaps.
India vs Taiwan, Korea, China and the US
Different starting points, not a league table.
| Factor | India | Taiwan | South Korea | China | United States |
|---|---|---|---|---|---|
| Leading-edge fabs | Not yet | Strong | Strong | Developing under constraints | Strong design, expanding fabs |
| Memory manufacturing | Packaging only | Limited | Global leader | Expanding | Micron |
| OSAT / packaging | Rapidly expanding | Mature | Strong | Strong | Expanding |
| Chip design | Strong | Strong | Strong | Strong | Global leader |
| Equipment & materials | Emerging | Deep supply base | Strong | Expanding | Strong |
A more useful question than “can India replace Taiwan?” is “can India become a second major manufacturing base?” Supply chains tend to diversify rather than relocate wholesale, driven by geopolitics, resilience, subsidies, customer diversification, talent and cost — not by a single “China plus one” slogan.
Isn’t 28 nm old technology?
No — not for most of what chips actually do. Leading-edge AI processors use much smaller nodes, but 28 nm-class technology runs the majority of automotive, power-management, microcontroller, communications and industrial chips, where a qualified, reliable, long-lived process matters more than transistor count. “Older than the leading edge” is not the same as “obsolete”. The better term is mature node, not “legacy”.
Will Dholera make Nvidia-class AI GPUs?
There is no basis to assume that. Frontier AI accelerators depend on sub-5 nm fabrication and advanced 2.5D/3D packaging ecosystems that India does not have. Dholera’s announced 28–110 nm range targets different but economically important markets. India’s realistic path toward AI silicon runs through design and domestic IP, then packaging and chiplets, then specialised accelerators — and possibly more advanced fabrication later. It does not jump straight from OSAT to GPUs.
What a semiconductor fab needs
Why water and power matter
Fabs repeatedly rinse wafers with ultra-pure water — municipal water treated to remove almost all ions, particles and organics — so a fab needs both a large water source and its own purification plant. It also needs stable, high-quality electricity: a power dip measured in milliseconds can ruin a batch of wafers, so fabs build in redundancy and backup. Project-specific consumption figures should come from environmental filings, not generic “millions of litres” claims.
Capacity numbers: read them carefully
What the numbers mean
- 50,000 wafers/month at Dholera is a wafer-start figure — each wafer holds many chips
- 48 million units/day at Tata Assam is a packaged-unit figure at full ramp
- “Commercial production started” means a line is running, not that it is at rated output
- Cumulative ₹1.64 lakh crore is approved investment across all stages
What not to do with them
- Don’t treat 50,000 wafers as 50,000 chips
- Don’t compare wafers/month against chips/day — different manufacturing stages
- Don’t rank “Assam is X times bigger than Dholera” from mismatched units
- Don’t read approved investment as money already spent
Is India making “Made in India chips”?
Yes, with a caveat about what the phrase means. For Micron, the wafer is fabricated abroad and the chip is assembled and tested in India. For a future Dholera chip, the silicon wafer itself is intended to be fabricated in India. Both are legitimate uses of “Made in India”, but they describe different depths of manufacturing. And no country is fully self-sufficient across design, EDA tools, equipment, materials, fabs, memory and packaging — the realistic goal is greater supply-chain resilience and domestic capability, not complete “chip sovereignty”.
Things worth knowing
- PSMC (Tata’s fab partner) and TSMC are different companies — do not confuse them.
- Micron, Kaynes and CG Semi are the three plants in commercial production; all three are backend, not fabs.
- The India Semiconductor Mission began in 2021 with ₹76,000 crore; Semicon 2.0 adds ₹1,27,500 crore.
- The FY2026–27 Budget’s ₹1,000 crore line for the mission is an annual provision, not the programme total.
- Silicon carbide (SiCSem) is a compound semiconductor for EVs and power electronics, not a silicon-logic fab.
- Even a domestic fab depends on ASML, Applied Materials, Lam Research, Tokyo Electron and KLA for tools.
- The Adani–Tower Maharashtra fab has never had central approval under the semiconductor programme.
How AiTimeline tracks semiconductor projects
We use a fixed set of status labels so a ribbon-cutting is not mistaken for a mature industry. Proposed: announced only. Approved: formal Government of India approval. Under construction: physical site work underway. Pilot: a limited production or test line. Commercial production: company or government confirms commercial operations. Full ramp: rated production reached or formally reported. For each facility we separate announcement date, approval date, construction start, pilot and commercial-production date — they are different milestones and are often reported as if interchangeable.
Frequently asked questions
The harder milestones come after the ribbon-cutting
A ribbon-cutting ceremony is not the same as a mature semiconductor industry. The difficult milestones come afterward: equipment has to run reliably, production yields have to improve, customers have to qualify the chips, suppliers have to locate nearby, and factories have to keep producing at commercial scale for years.
That is why India’s semiconductor story should now be measured less by the number of announcements and more by what is actually coming off production lines.
India’s chip story has moved from announcements to production
India’s semiconductor push entered a different phase in 2026. Micron began assembly and test operations in Sanand. Kaynes moved into production. CG Semi began commercial output. Those milestones matter because they replace future-tense promises with physical production.
But the most technically difficult step is still ahead. India has to bring large-scale wafer fabs into reliable commercial operation, develop local materials and equipment suppliers, raise manufacturing yields and build deeper advanced-packaging capability. Tata’s Dholera fab will be one of the most important industrial tests of the decade. If it succeeds, India will no longer be only a place where chips are designed and packaged — it will also be a place where the silicon itself is manufactured at commercial scale.
2026 proved India can start producing semiconductor products. The next test is whether it can build an entire chip ecosystem around them.
Explore related timelines
⚠️ How AiTimeline tracks India’s semiconductor build-out
We classify each project as proposed, approved, under construction, pilot, commercial production or full ramp, and update status as facts develop. Investment figures are approved project values from Government of India and company statements, not money already spent, and mix private capital with central incentives and state support. This article is editorial and AI-assisted, compiled from public sources including PIB and Cabinet releases, PMIndia, the India Semiconductor Mission, Micron, Tata Electronics, CG Power, Kaynes Technology, Reuters, Business Standard, The Economic Times and TrendForce. It is not investment advice. Information may contain inaccuracies; corrections are made as the projects progress.
Sources & further reading
Every dated entry above was checked against these references. Last reviewed 28 August 2026.
- Cabinet approves Semicon 2.0 - PMIndia
- Cabinet approves Semicon 2.0 - Press Information Bureau
- Micron Celebrates Opening of India's First Semiconductor Assembly and Test Facility
- PM inaugurates the Kaynes Semicon plant at Sanand, Gujarat - PMIndia
- CG Semi starts commercial production at G1 OSAT facility in Sanand - Business Standard
- Tata Electronics and PSMC agreement to build India's first semiconductor fab
- India Semiconductor Mission
- Tata's India fab reportedly to launch at 90nm, commercial production delayed to mid-2028 - TrendForce