We investigate the impact of multiple distributed intelligent reflecting surfaces (IRSs), which are deployed and optimized by a mobile operator (MO), on the performance of user equipments (UEs) served by other co-existing out-of-band (OOB) MOs that do not control the IRSs. We show that, under round-robin scheduling, in mmWave frequencies, the ergodic sum spectral efficiency (SE) of an OOB MO is monotonic in the total number of IRS elements with a pre-log factor that depends on the channel properties of the OOB UE. We further show that the maximum achievable SE of OOB MO scales log-linearly in IRS elements. Then, by specifying the minimum number of IRSs as a function of the channel parameters, we design a distributed IRS system in which an OOB MO almost surely obtains the maximum SE. Finally, we prove that the outage probability at an OOB UE decreases exponentially in the number of IRSs, even though they are randomly configured from the UE's viewpoint. We numerically verify our theory and conclude that distributed IRSs always help every MO, but the MO controlling the IRSs benefits the most.