universal gate set
Think about how, in ordinary digital electronics, you don't need a different piece of hardware for every possible logical operation. A handful of basic gates — AND, OR, NOT, or even just NAND on its own — can be wired together to compute any function at all. A universal gate set in quantum computing is the same idea, one layer up: it's a small collection of quantum gates from which you can build any quantum computation you want, just by applying them in the right sequence.
One common universal set is {H, T, CNOT} — the Hadamard gate, the T gate, and the controlled-NOT. With only these three, you can approximate any quantum operation you need. The word approximate matters and is not a weakness: quantum gates rotate states by continuous angles, so a finite toolbox can't hit every possible operation exactly. Instead, you compose these gates to land as close to the target as you like, and a result called the Solovay-Kitaev theorem promises you can get there efficiently, without needing an absurd number of gates for each extra digit of precision.
So when engineers design a quantum processor, they don't try to build hardware for thousands of distinct operations. They build a few high-quality gates that together are universal, then express every algorithm as a sequence drawn from that set — much like a compiler turning your code into a stream of a few machine instructions. Which exact gates are chosen depends on the hardware, but the promise of universality is the same: nothing computable on a quantum computer is left out of reach.
Universal means any quantum operation can be built from the set — it says nothing about speed; most problems get no quantum speedup, and the gains that exist range from quadratic (Grover) to exponential only for specific structured problems like Shor's factoring.