<p>Although the applications of artificial intelligence especially deep learning have greatly improved various aspects of intelligent manufacturing, they still face challenges for broader adoption due to poor generalization ability, difficulties in establishing high-quality training datasets, and unsatisfactory performance of deep learning methods. The emergence of large scale foundation models (LSFMs) has transformed applications of deep learning models from single task, single-modal, limited data patterns to a new paradigm encompassing diverse tasks, multi-modal, and pre-training on massive datasets. Although LSFMs have demonstrated powerful generalization capabilities, efficient data utilization and superior performance in various domains, applications of LSFMs in intelligent manufacturing are still in their nascent stage. A systematic overview of this topic is lacking, especially regarding challenges of utilizing deep learning in intelligent manufacturing and how these challenges can be systematically tackled by LSFMs. To fill this gap, this survey provides a systematic overview of the current status of LSFMs and their advantages in the context of intelligent manufacturing. It also presents comprehensive comparisons between current deep learning models and LSFMs in various intelligent manufacturing applications. Subsequently, roadmaps for utilizing LSFMs to address these challenges are also outlined. Case studies of employing LSFMs in real-world intelligent manufacturing scenarios are also presented highlighting how LSFMs can enhance industry efficiency. Finally, the challenges currently faced by LSFMs in intelligent manufacturing are discussed, along with future research directions. The survey highlights that the application of LSFMs in intelligent manufacturing is rapidly gaining widespread attention and demonstrating potential across various tasks.</p>