The first round of India's semiconductor PLI was dominated by the political symbolism of "Indian fabs." The economic reality was always going to disappoint that framing — building leading-edge fabs in India in the 2020s is not a credible proposition. The second-round PLI redesign acknowledges this by shifting weight decisively toward OSAT (outsourced semiconductor assembly and test) and advanced packaging. That is the right answer to the strategic question.
Key takeaways
- Advanced packaging is now the principal bottleneck in AI accelerator supply chains.
- India has labor cost, talent depth and downstream electronics scale that fit packaging well.
- Taiwan and Korea hold packaging dominance — India can plausibly take third place.
- The PLI redesign is structured to fund this directly.
Why packaging is the constraint
Modern AI compute depends on CoWoS, SoIC and HBM stacking. Capacity is concentrated in a small number of TSMC, ASE and Amkor sites. Demand is structurally above supply, with allocation pressure that has lasted multiple years. Adding capacity outside Taiwan is one of the highest-leverage industrial-policy moves available to any government.
- CoWoS. TSMC capacity supply is the binding constraint on every AI accelerator program.
- HBM. Memory packaging concentrated in two Korean and one US site.
- Substrates. Japanese supplier base near full utilization.
What India brings
India has a deep pool of test and assembly talent through legacy electronics manufacturing, low operating costs versus Taiwan and Korea, and a downstream electronics base that absorbs domestic capacity. The combination is not unique but it is sufficient.
Where the partners come from
Micron, Tata-PSMC, Amkor and Korean memory suppliers are already committed. The second-round PLI sweetens conditions further.
What it does for the AI compute supply chain
It is genuinely additive. The global packaging bottleneck eases marginally by 2027 and meaningfully by 2029 if Indian capacity executes.
Capacity allocation
PLI-2 redesign by subsidy allocation.
| Segment | PLI-1 share | PLI-2 share |
|---|---|---|
| Greenfield fabs | ~65% | ~30% |
| OSAT/packaging | ~20% | ~50% |
| Design/IP | ~10% | ~15% |
| Materials | ~5% | ~5% |
The PLI redesign moves money toward the segment where India can actually win.
Frequently asked questions
Does India still get a leading-edge fab?
Eventually, probably. Tata-PSMC is the most credible near-term path, on trailing nodes.
Where does the talent come from?
Domestic engineering schools plus return migration of senior engineers from Singapore, Malaysia and the US.
What about Chinese competition in OSAT?
China's OSAT players are mid-tier; India's path is largely additive in the advanced segment.
The bottom line
India is correctly pivoting from the politically attractive but commercially difficult "Indian fabs" framing to the harder-but-real packaging opportunity. If the second-round PLI is executed, India becomes a meaningful node in the global semiconductor supply chain by 2029.
