Spectre is a class of CPU side-channel attacks that abuses speculative execution. Modern processors predict which instructions they will need next and run them early to stay fast. Even when the prediction is wrong, traces can remain in caches, branch predictors, or timing behavior. An attacker who can measure those traces may infer data that should have stayed private, such as secrets from another process, sandbox, or virtual machine.
Spectre matters because it targets hardware behavior, not just software bugs. Traditional validation and access controls may still be bypassed if the processor leaks information through microarchitectural state. Defenses focus on reducing what can be observed or speculated: processor and firmware mitigations, compiler hardening, kernel isolation changes, and sometimes disabling or limiting high-risk features like JIT-heavy code paths. In practice, security teams treat Spectre as a performance-versus-isolation tradeoff that must be managed across CPUs, operating systems, browsers, and runtimes.