Setup violation
A setup violation means the data signal arrives too late — it shows up at a flip-flop's input after the clock edge has already tried to grab it, like a passenger reaching the gate after the plane has pushed back. The flip-flop needs the data to be stable for a small window (the setup time) before the clock edge; if the path is too slow to make that window, the captured value is wrong or metastable. This is the classic max-delay failure, and it's the violation timing engineers fight hardest, because it directly caps how fast the chip's clock can run.
Because it depends on path delay, a setup violation gets worse in the slow corner — low voltage, high temperature, slow process — and you fix it by making the path faster or the clock slower. Real-world cures include upsizing weak gates, restructuring or pipelining the logic, reducing wire load, applying useful skew to borrow time, or, as a last resort, dropping the target frequency. Speed-binning a CPU is essentially shipping the parts whose worst setup paths still close at a higher clock.
A negative setup slack of 70 ps caps this path's max clock at a longer period.
Setup is checked between the launch edge and the next capture edge (one cycle later by default), so unlike hold it is fixable by slowing the clock — which is exactly why hold violations, immune to clock period, are considered the scarier of the two.