Wi-Fi vulnerabilities are often tied to specific chipsets, yet identifying the chipset behind an observed MAC address from over-the-air transmission alone is an open problem. We present NICPrint, an identification system that recovers the manufacturer and chipset model of a Wi-Fi network interface (NIC) by remotely probing its digital reception logic. The IEEE 802.11 standard specifies what the Wi-Fi preamble must accomplish but not how, leaving vendors free to make implementation choices in their PHY baseband designs. Without requiring network credentials, I/Q access, CSI, or target cooperation beyond ACK transmission, NICPrint elicits these implementation differences by transmitting frames with controlled preamble corruption and observing the polite-ACK responses to construct a fingerprint. We evaluate NICPrint on 25 devices spanning five major Wi-Fi vendors and four Wi-Fi generations. Same- session cross-validation provides an upper bound of 99% accuracy for both manufacturer and chipset identification. On previously unseen devices with chipsets represented in training, NICPrint averages 95% manufacturer and 96% chipset accuracy across the tested classifiers. Under through-wall, non-line-of-sight, and varying distance conditions measured more than one month after training, mean classifier accuracy is 83% for manufacturer and 80% for chipset identification; channel augmentation increases these to 98% and 93%, respectively. These results establish NICPrint as a practical Wi-Fi reconnaissance tool.