Nova — AI supply-chain exposure

Nova makes process-control metrology tools, the measurement systems that let a chip fab see whether the microscopic structures it is etching are actually correct, layer by layer. It sells into leading-edge logic foundries and memory fabs, the exact places where AI silicon and high-bandwidth memory are made. Its exposure is to yield and foundry capacity: without accurate metrology, a fab cannot tune its process or catch defects, and the fraction of good chips per wafer suffers.

The structural hook is that metrology is invisible but load-bearing. As transistors shrink and memory stacks grow taller, the margin for error narrows, and fabs lean harder on measurement to hold yield, which is what ultimately converts wafer capacity into shippable accelerators and memory. The pool of suppliers with credible leading-edge process control is small, giving Nova quiet pricing power tied to every capacity expansion. The model places it upstream, where usable output is won or lost before a chip ever leaves the fab.