Micron's $2.75 billion Sanand facility opened in February 2026 and has already shipped made-in-India memory to Dell. Tata Electronics' much larger Dholera fab — the one actually meant to fabricate chips rather than assemble and test ones made elsewhere — is still under construction. Both are real. They are not the same kind of real, and the difference is the whole story.
India's "semiconductor moment" gets reported as a single storyline, but it is really two different projects on two different timelines wearing the same headline. One of them is finished and shipping product. The other is a genuine, capital-intensive, multi-year construction project that will not tell us whether it worked until well after this letter is written. Conflating the two — treating "India now makes chips" as a settled fact — is the most common and most consequential error in coverage of this sector right now.
Micron's Sanand facility, a $2.75 billion investment jointly backed by Micron and Indian government incentives, formally opened on February 28, 2026, with more than 500,000 square feet of cleanroom space in its first phase. It is already producing: Micron's own guidance points to tens of millions of chips assembled and tested in 2026, scaling to hundreds of millions in 2027, and the company marked the opening by shipping its first made-in-India memory modules to Dell for laptops manufactured in India for the Indian market. This is real, live, revenue-generating activity today, not a projection.
Sanand is an assembly and test facility — what the industry calls OSAT (outsourced semiconductor assembly and test) — not a fabrication plant. It takes finished silicon wafers made elsewhere (Micron's own fabs are in the US, Taiwan, Japan, and Singapore) and packages, tests, and modules them into finished DRAM and NAND products. That is genuinely valuable, labour-intensive, supply-chain-critical work, and it is the first link most countries build before attempting fabrication itself. But "India assembles and tests memory chips" and "India fabricates semiconductors" are different claims, and a lot of headline coverage this year has quietly let the first one stand in for the second.
Tata Electronics' Dholera fab, built in partnership with Taiwan's Powerchip Semiconductor Manufacturing Corporation (PSMC), is the project actually attempting fabrication — turning raw silicon wafers into finished chips through the full lithography-and-etching process. The numbers here are larger and the technology genuinely harder: a stated ₹91,000 crore (roughly $11 billion) investment, capacity planned for up to 50,000 wafers a month, starting at the 28-nanometre node and advancing toward 22-nanometre. Multiple reports point to a 2026 target for the first chip to roll off the line, though the framing across sources is consistently "may" or "targets," not "has." Construction reporting through mid-2026 described the project as past the halfway point, not complete.
| Facility | Type | Status | Scale |
|---|---|---|---|
| Micron Sanand | Assembly & test (OSAT) | Open, shipping product | $2.75B, 500,000+ sq ft phase 1 |
| Tata-PSMC Dholera | Wafer fabrication | Under construction | ₹91,000cr, up to 50,000 wafers/month |
It is also worth being precise about what Dholera will make even once it is running: 28nm and 22nm are trailing-edge nodes by global standards, well behind the 3-5nm processes TSMC and Samsung use for the most advanced AI and smartphone chips. That is not a criticism of the project — trailing-edge chips run the vast majority of the world's cars, appliances, and industrial electronics, and building deep expertise there is a sound, sequenced strategy rather than a failed attempt at the leading edge. But it means Dholera's eventual output competes in a different, less glamorous market than the AI-chip headlines around India's semiconductor ambitions sometimes imply.
India has one working link in the semiconductor chain and one under construction. Investors and policymakers who treat those as the same achievement will be surprised, in either direction, by how the next eighteen months actually unfold.
This is the least satisfying part of the letter to write, and the most important to be honest about: there is no clean, direct, listed pure-play on either Sanand or Dholera available to a retail investor today. Micron Technology itself is a US-listed global memory giant for which the India facility is a modest, if strategically important, piece of a much larger business — a way to gain some exposure to the theme, not a bet on India semiconductors specifically. Tata Electronics is a subsidiary of Tata Sons and is not independently listed. The more honest way to get exposure is indirect: capital-equipment and engineering firms supplying cleanroom construction and fabrication tooling, ancillary electronics and materials companies in the Gujarat semiconductor corridor, and — for investors willing to hold at the conglomerate level — Tata Group-linked listed entities that stand to benefit from the broader ecosystem Dholera anchors, understanding that any single one of those is a diluted, indirect way to express this specific thesis.
Fabrication is unforgiving of delay and cost overrun in a way assembly and test is not — yield ramps, equipment installation, and process qualification at a brand-new fab have derailed or badly delayed comparable projects elsewhere, and Dholera's "2026" target should be treated as aspirational until wafers are actually shipping in volume. India's semiconductor policy also remains subsidy-dependent; the Production Linked Incentive framework has been generous, and any future tightening of that support would change project economics for both facilities. Finally, global memory pricing is cyclical, and Micron's own India guidance is tied to a broader DRAM/NAND market that has been through sharp swings before.
India's semiconductor build-out is real, but it is two different projects at two different stages of maturity, and the sector deserves to be evaluated that way rather than as a single undifferentiated "India chips" story. Sanand is a genuine, shipping success in assembly and test; Dholera is a genuine, still-unproven attempt at fabrication whose 2026 timeline is a target, not a result. Direct listed exposure is scarce for either — this is a theme to watch through the ecosystem and the conglomerate level, not one with an obvious single stock to buy.