Navitas Semiconductor — AI supply-chain exposure

Navitas is a fabless designer of power semiconductors built on wide-bandgap materials — gallium nitride (GaN) and silicon carbide (SiC) — rather than ordinary silicon. These switch electricity faster and waste less of it as heat, which matters as the AI data center becomes a power problem as much as a compute one. It sits in the power-delivery chain as a supplier, and in a recent tie-up with Nvidia it is helping design a high-voltage DC architecture meant to move electricity to next-generation AI racks with fewer conversion losses.

Its structural hook is exactly that transition: the industry is shifting from feeding racks at lower voltage to distributing at much higher voltage, and wide-bandgap devices are what make that efficient. That places Navitas near a genuine bottleneck if the shift happens. But the model gives it modest centrality — it is one of several players in power silicon, its designs still ride on foundry capacity it does not own, and the payoff depends on an architecture change that is only beginning.