ASML-Tata: India's First Chip Factory Is Real, But Is It Enough?

When India's Prime Minister stood beside the Dutch Prime Minister in the Netherlands on May 16, 2026, to witness Tata Electronics and ASML sign their semiconductor MoU, every Indian tech outlet called it a "historic masterstroke." And to be fair, it is a landmark moment. India is building its first commercial front-end chip fab. But here is the part that the headlines skipped: while the world's most advanced chips are now being made at 2nm (think of it as fitting 50 billion transistors on a chip the size of your fingernail), India's Dholera fab will manufacture chips at 28nm to 110nm; nodes that were mainstream between 2001 and 2011. 

So does this deal actually put India on the semiconductor map, or are we celebrating catching a train that left the station a decade ago?

Let's break down what the deal actually is, who it matters more to, why 28nm is not the disaster the critics say but also not the revolution the cheerleaders claim, and most importantly what India's real chip play looks like if we stop trying to fight yesterday's war.

The ASML-Tata Deal, Decoded

On May 16, 2026, Tata Electronics and ASML, the Dutch company that makes the machines that make every advanced chip on earth, signed a Memorandum of Understanding (MoU) to set up India's first commercial 300mm semiconductor fabrication plant in Dholera, Gujarat.

Here is what "300mm" actually means: 300mm refers to the size of the silicon wafer (imagine a large dinner plate made of pure silicon) on which chips are printed. It is not the chip's capability level. Both TSMC's cutting-edge 2nm fabs and Tata's upcoming fab use 300mm wafers. The real metric that defines a chip's power is the process node and Tata's fab will operate at 28nm to 110nm nodes, using ASML's DUV (Deep Ultraviolet) lithography tools.

The fab will be built in Dholera with a planned investment of $11 billion (~₹91,000 crore), targeting 50,000 wafer starts per month. Tata has also partnered with Taiwan's Powerchip Semiconductor Manufacturing Corporation (PSMC) for process technology. 

How Big Is This, Really? For ASML vs. For India

This question has two very different answers depending on whose side of the table you sit on.

For ASML, this deal opens a new geography and creates long-term service revenue, but at €32.7 billion in annual sales, India is a rounding error today. The real ASML story is EUV: their machines, priced at $250-380 million per unit, are the sole reason sub-7nm chips exist. 

SK Hynix alone committed $8 billion to ASML in a single EUV purchase in early 2026. The Tata deal uses ASML's older DUV tools, not EUV. Think of it this way: ASML is the luxury watchmaker who just agreed to also supply affordable watches for a new market. It is a win, but it does not move the needle much.

For India, however, this is seismic. The country has never had a front-end semiconductor fab. This is a zero-to-one moment, not a one-to-ten moment. The significance is more about building an ecosystem, training engineers, and proving to the world that "Made in India chips" is not just a slogan.

28nm vs. 2nm: What Are We Actually Talking About?

Think of chip nodes like camera megapixels, more is generally better for performance, but an 8MP camera is still perfectly useful for WhatsApp. The question is: what job are you trying to do?

Source: SIA, WaferPro, Silicon Analysts, SEMI Fab Outlook Report 2024

The 28nm market is not dead, it is actually booming. Every electric vehicle needs dozens of automotive-grade chips. Every 5G base station uses 28nm components. India's own EV push (targeting 30% EV penetration by 2030) and 5G rollout (500 million subscribers projected by 2030) create real, immediate domestic demand for exactly these chips.

But let us not confuse "viable" with "world-leading." The chips running AI infrastructure, chips like NVIDIA's H100, Apple's M4, Qualcomm's next-gen Snapdragon, are manufactured at 3nm and 4nm today, with 2nm ramping up at TSMC and Samsung as of late 2025. India's fab will not touch those categories for the foreseeable future. The gap is not just technological, it is also economic. A 2nm fab requires EUV machines at $380 million per unit, and ASML's EUV tools are export-controlled, meaning India cannot even buy them today.

Can This Deal Put India on the Semiconductor Map?

The honest answer: partially, and over a long, long timeline.

India's semiconductor market was valued at approximately $60 billion in 2025 and is projected to reach $100-180 billion by 2030-2034, growing at a CAGR of 12-15%. But here is the structural problem that one $11 billion fab cannot solve alone.

Source: Endiya Partners IESA Summit 2026 Report; Mordor Intelligence; IndexBox; SEMI

India has always been extraordinarily good at chip design. ARM opened a 2nm design office in Bengaluru in September 2025. Intel employs over 13,000 engineers in India. Qualcomm's 2nm tape-out at TSMC in February 2026 involved Indian design teams. The talent exists. What has been missing is the physical manufacturing capability and the Tata-ASML deal starts to address that, at the 28nm level.

But to genuinely displace anyone from the global semiconductor supply chain at the leading edge? India would need EUV access, a $50-100 billion multi-generational fab investment program, and another 10-15 years of uninterrupted execution. TSMC spent 40 years and hundreds of billions of dollars getting to where it is today.

The Jio Moment for Chips: Should India Just Leapfrog?

Here is a thought experiment worth sitting with.

In 2014, India had roughly 9% rural internet penetration and a landline network that covered a fraction of the country. Instead of spending decades and billions building landline copper infrastructure the way developed nations had, India effectively skipped it. Reliance Jio launched in 2016, offered 4G internet free for months, and within four years became the world's largest mobile network. Today, India has the cheapest mobile data on earth as 1GB costs less than a vada pav and with over 1.16 billion mobile subscribers.

India did not compete with AT&T or Deutsche Telekom on their terms. It found a different game and dominated it.

The same strategic question applies to semiconductors: should India spend the next 15 years trying to catch TSMC at 2nm, a race that China, with $100 billion in government spending, is still losing or should it find the equivalent of the "4G leapfrog" in chips?

The answer lies in three specific bets that are already quietly underway.

Bet 1: Own the RISC-V Design Stack

RISC-V is an open-source chip architecture, think of it as the Linux of processors. No licensing fees to ARM. No geopolitical restrictions. Indian startups are already here as Mindgrove Technologies (IIT Madras) is building RISC-V chips for edge AI at 30% lower cost than international alternatives. 

CalligoTech launched TUNGA, hailed as the world's first Posit-enabled RISC-V CPU for AI. These chips can be designed in India and manufactured at TSMC or Samsung, because India does not need to own the fab to own the IP.

Bet 2: Corner the Edge AI Chip Market

The next wave of AI is not in data centers, it is at the edge. Your phone, your car, your security camera, your factory floor. These applications need efficient, low-power AI chips at 28nm to 7nm nodes, not the 2nm bleeding edge. India's domestic demand with 500 million 5G subscribers, 154.2 million smartphones shipped in 2025, and a fast-growing EV market, creates the world's second-largest testing ground for exactly this category of chips.

Bet 3: Become the World's OSAT Capital

Chip packaging, testing, and assembly, collectively called OSAT, is a $60+ billion global market and does not require EUV machines. Micron's ATMP facility in Sanand, Gujarat was inaugurated by PM Modi on February 28, 2026 and is already shipping memory to Dell and HP. CG Power's OSAT pilot launched in August 2025, handling half a million chips per day. India can realistically be to chip packaging what Vietnam is to electronics assembly; fast, scalable, and cost-competitive.

Source: India Chip Moment Report (March 2026); IESA Vision Summit 2026; IoT Analytics; Reuters

What India Actually Needs to Do Beyond the Headlines

The Tata-ASML deal is necessary but not sufficient. Data from the 2026 industry landscape points to four non-negotiable building blocks:

1. Move from design to IP ownership: India designs chips but rarely owns the resulting IP. The DLI scheme has approved 23 chip design projects and given 72 companies access to EDA tools. That is a start, but the goal must be filing patents and licensing IP globally, the way ARM built a $60 billion business by designing CPUs it never manufactured.
2. Build a full-stack talent pipeline: India has software engineers in abundance but a severe shortage in analog design, physical layout, and chip verification, the disciplines that turn a design into working silicon. IIT Madras's SHAKTI project and ARM's Bengaluru design center are bright spots, but the pipeline needs to be 10x larger.
3. Target $50 billion in cumulative fab investment by 2035: The Tata fab at $11 billion is step one. SEMI data shows China is to spend $100 billion on 300mm fab equipment between 2025-2027 alone. India needs more fabs across nodes, more OSAT facilities, and more packaging plants to have any meaningful supply-chain weight globally.
4. Secure technology access agreements for 7nm: Today's DLI and PLI schemes are unlocking 28nm. The next policy objective must be negotiating technology-transfer frameworks for 7nm, potentially through the US-India iCET (Initiative on Critical and Emerging Technologies) semiconductor cooperation track, which was reinforced in early 2026.

The Verdict

The Tata-ASML deal is real, meaningful, and historic, but a  28nm fab is a foundation, not a finish line. The global semiconductor industry is operating at 2nm at the frontier, and the gap is measured not just in nanometers but in decades of compounding technological investment.

The smarter play for India is not to spend the next 20 years trying to catch TSMC, a game with no realistic finish line. It is to do what India did with telephones: skip the copper wires, go straight to 4G, and build the cheapest, most penetrated mobile internet on earth. In chips, that equivalent is owning the design IP, dominating the OSAT segment, and positioning Indian AI chip companies to ride the edge-computing wave before the West and China have locked it up.

The Dholera fab is India's "buying land in a new city" moment. Whether India builds a trading post or a metropolis on it depends on the ten decisions that come after this one.