Radar systems do not merely need processors. They need specialised RF silicon that can survive extreme frequency, voltage and power demands. DRDO’s Solid State Physics Laboratory has developed indigenous GaN-on-SiC technology, including 4-inch SiC wafers, GaN HEMTs and a 3.5 × 3.5 mm GaN MMIC capable of 30 watts. Higher-power devices have reached 150 watts, with limited production beginning at GAETEC, Hyderabad. The significance extends beyond one chip. AESA radars, electronic warfare systems, seekers, satellite communications and counter-drone platforms all depend on such components. Semiconductor sovereignty, therefore, is increasingly about owning the specialised silicon beneath the battlefield’s visible machinery.

Construction has always resisted the factory model. Every site is different, terrain shifts, structures rise, and workers constantly reposition. K-nest Construction Tech is betting that robotics can finally make the industry programmable, combining 3D concrete printing, autonomous inspection robots, construction exoskeletons, precision sensors, self-climbing safety screens and automated high-rise formwork. The interesting part is not any single machine, but the stack: robotics, computer vision, materials science and industrial automation converging around one of the world’s least automated industries. As India enters an infrastructure-heavy decade, even incremental gains in construction speed, safety, precision and material efficiency could compound into substantial productivity gains.

AI’s next generation may depend on a machine that never runs a model. ASML’s High-NA EUV lithography systems, costing roughly $400 million each, are becoming central to future chip roadmaps. Samsung and SK Hynix target adoption from 2028, Intel is already working with the technology, and TSMC plans production use from 2030. High-NA can print features roughly 40% smaller than conventional EUV, potentially extending transistor scaling while simplifying some process steps. ASML already dominates advanced lithography and effectively monopolises EUV. The strategic lesson extends beyond AI: semiconductor power increasingly belongs to whoever controls the machines that manufacture the chips.