Phase-change memories (PCMs) utilize both fast phase transitions between the amorphous (a-) and crystalline (c-) states of PC materials, and accompanying large electrical- resistance contrast, offering feasible intriguing applications in fast/nonvolatile/nanoscale built-in information storage, particularly for their ability to scale down to nanometer length scales.[1-5] Recently, PCM technology has further found applications in the area of neuromorphic computing by virtue of the accumulative nature of the amorphous-to-crystalline phase transition (ACT) in PC materials, which allows biological-like read/write operations in PCMs.[6-9] However, the overall operation speeds of all these PCM-based devices, along with their corresponding power consumption, are intrinsically limited owing to the ACT being much slower than the amorphization process. Current efforts to overcome such limitations are focused on employing material-optimization or device-miniaturization approaches. Nevertheless, much still remains to be done due to the contradictory nature of increasing the ACT speed while extending the data-retention properties of PC materials. This means that the existing PCM-based devices will soon reach their ultimate performance limit under the current operating paradigm, insurmountable merely by adopting conventional approaches for improving the performance of PCMs.