Phase-change materials (PCMs) are important photonic materials that have the advantages of a rapid and reversible phase change, a great difference in the optical properties between the crystalline and amorphous states, scalability, and nonvolatility. With the constant development in the PCM platform and integration of multiple material platforms, more and more reconfigurable photonic devices and their dynamic regulation have been theoretically proposed and experimentally demonstrated, showing the great potential of PCMs in integrated photonic chips. Here, we review the recent developments in PCMs and discuss their potential for photonic devices. A universal overview of the mechanism of the phase transition and models of PCMs is presented. PCMs have injected new life into on-chip photonic integrated circuits, which generally contain an optical switch, an optical logical gate, and an optical modulator. Photonic neural networks based on PCMs are another interesting application of PCMs. Finally, the future development prospects and problems that need to be solved are discussed. PCMs are likely to have wide applications in future intelligent photonic systems.

All-optical switches are among the most important parts of integrated photonics. Ultrahigh speed and ultralow energy consumption are two necessary indexes of all-optical switches. Traditionally, all-optical switches are based on concepts such as micro-ring resonators, surface plasmon polaritons, photonic crystals, and metamaterials. However, such platforms cannot satisfy the demand for high performance of all-optical switches. To overcome the limited response time and energy consumption, recent studies have introduced new applications of such physics as parity–time symmetry, exceptional points, topological insulators, and bound states in a continuum. Such physical concepts not only provide promising research avenues for the all-optical switch but also broaden the design channel. This is expected to achieve ultracompact, ultrafast, and high-capacity all-optical information processing.