In this talk, we will explore the challenges faced by compiler developers using LLVM to mitigate hardware vulnerabilities and the potential need to evolve the LLVM compiler to effectively integrate security-focused transformations alongside its powerful optimization capabilities. As microarchitectural side-channel attacks like Spectre continue to evolve, it is crucial to address the challenges faced by mitigation developers at various levels of LLVM. These compiler-based defenses provide a flexible and timely response to emerging threats. They also enhance the overall security posture by layering defenses, protecting systems even when hardware-based solutions fall short or are not feasible. We will examine common defense patterns, the complexities of implementing mitigations, and the critical differences between these and traditional compiler transformations. Furthermore, we will discuss potential enhancements to LLVM that could increase the reliability of security measures and argue why compilers are the optimal platform for these defenses. We aim to underscore the necessity of a two-step mitigation infrastructure and highlight how tools like Bolt could help in addressing current limitations, paving the way for a more secure code transformation landscape.