Why doesn't India build its own GaN RF semiconductor devices?

Only a handful of nations can build the semiconductor that sits at the heart of a modern radar. As of 2026, India is the seventh — one of just seven countries with demonstrated indigenous gallium nitride RF chip capability.

Gallium nitride handles far higher energy than silicon, which makes it essential for 5G and 6G base stations, advanced radar, electronic warfare, and satellite communications. That performance is exactly why it is hard to make. Native GaN wafers are extremely difficult to produce, so the material is instead grown as a thin film on silicon carbide (prized for thermal conductivity) or plain silicon (prized for cost) — both routes carrying significant lattice-mismatch problems. Unresolved global challenges persist in p-type doping, substrate compatibility, thermal management, and defect-free epitaxy at scale. At DRDO's facilities, a single fabrication and testing cycle takes around 80 days and runs through hundreds of process steps, each requiring precise execution.

India has climbed steadily. DRDO's Solid State Physics Laboratory built prototype GaN chips around March 2023, and by early 2026 had fabricated GaN MMICs — monolithic microwave integrated circuits — at the GAETEC facility in Hyderabad that cleared all functionality tests. It has demonstrated indigenous methods for 4-inch silicon carbide wafers, GaN HEMT transistors up to 150W, and MMICs delivering 40W at X-band frequencies. Agnit Semiconductors, founded on more than 18 years of R&D at the Indian Institute of Science, shipped its first GaN wafers in 2022 and now offers 5W to 60W devices with a roadmap toward 200-300W parts for telecom, defence, and jamming systems; it holds an MoU with the Ministry of Defence under iDEX. On 5 May 2026, the Union Cabinet approved Crystal Matrix Limited to establish GaN foundry services on 6-inch wafers in Dholera, Gujarat.

The gap between demonstrated and dominant persists structurally. The global RF GaN market is concentrated among five vendors — Wolfspeed, Qorvo, Sumitomo, NXP, and MACOM — controlling roughly 60% of revenue. India's wider semiconductor sector remains 90-95% import-dependent as of 2026, and the foundational layer beneath GaN is thin: commercial-scale compound-semiconductor foundry capacity is only emerging, and indigenous substrate sourcing for silicon carbide and sapphire is effectively absent.

What it would take is visible in the economics. A GaN fab needs roughly $500 million against $20 billion for an advanced CMOS fab — a reachable figure. NITI Aayog's May 2026 roadmap names compound semiconductors a strategic priority within a targeted $120-150 billion value chain by 2035. The lab work is done; the remaining climb is volume, substrates, and foundries.