Over a thousand of warm debris disks were detected as infrared excess at mid-infrared wavelength,
and their frequencies were obtained for various spectral types of stars. However, the dependence
of frequencies on spectral types is still under debate, since the number of stars showing significant
and detectable infrared excess is limited. Here, we present the largest systematic search for infrared
excess using Gaia, WISE, and Spizter. We identify 372, 559, and 270 reliable infrared excess in mid-
infrared archival data at wavelengths of 12, 22, and 24 μm for WISE/W3, W4, and Spitzer/MIPS ch1,
respectively. Several factors can cause differences between observed raw excess rate and a frequency of debris disks, i.e., contamination due to background objects. We develop methods to mitigate those factors, and obtain frequencies of warm debris disks as a function of stellar effective temperature. While we confirm the tendency of higher frequencies of debris disks for more massive stars, these disk frequencies are relatively flat for both low-mass and intermediate-mass stars at a border of around 7000 K for all the three wavelengths. Assuming bright warm debris disks have lifetime of a few to several hundred million years, the disk frequency can be understood by the ratio between the timescale and upper limits of the sample age. We also find that intermediate-mass stars with infrared excess tend to have bluer colors on color-magnitude diagrams, implying that massive stars hosting debris disks are relatively young, isochronal age of 200?800 Myr on average. These tendencies are reasonably explained by a standard scenario, in which debris disks are likely to be detected due to collisions of planetesimals at early phases of stellar evolution such as the giant impact.