Large Language Models (LLMs) are part of Artificial Intelligence (AI) systems. LLMs are highly capable of generating human-like text and conversation. The use of LLMs became prevalent in educational spaces among students, teachers, curriculum developers, and policymakers. One of the educational areas where LLMs are of great assistance is in the field of computing. There has been extensive research on how LLMs can have a positive influence in programming education. However, the capacity of LLMs to handle tasks similar to compilers has not been adequately investigated. This study has been triggered by how accurately LLMs mimic tasks performed by compilers. In this paper, we present a comparative case study between the capabilities of LLMs and compilers. The comparative experimental results (obtained through ChatGPT and Google Gemini) show that LLMs are capable of mimicking compilers and can serve as additional support to compilers. These findings will help programming students and teachers, especially in open-distance education, to be aware of what LLMs are capable of, and their usage as additional support to compilers.