The dwarf satellites of `giant' Milky Way (MW)-mass galaxies are our primary probes of low-mass dark matter halos. The number and velocities of the satellite galaxies of the MW and M31 initially puzzled galaxy formation theorists, but are now reproduced well by many models. Yet, are the MW's and M31's satellites representative? Were galaxy formation models `overfit'? These questions motivate deep searches for satellite galaxies outside the Local Group. We present a deep survey of the `classical' satellites ($M_{\star}$$\geqslant$4$\times$10$^5 M_{\odot}$) of the MW-mass galaxy M94 out to 150 kpc projected distance. We find $only\ two$ satellites, each with $M_{\star}{\sim}10^6 M_{\odot}$, compared with 6-12 such satellites in the four other MW-mass systems with comparable data (MW, M31, M81, M101). Using a `standard' prescription for occupying dark matter halos (taken from the fully hydrodynamical EAGLE simulation) with galaxies, we find that such a sparse satellite population occurs in $<$0.2% of MW-mass systems $-$ a $<$1% probability among a sample of five (known systems + M94). In order to produce an M94-like system more frequently we make satellite galaxy formation much more stochastic than is currently predicted by dramatically increasing the slope and scatter of the stellar mass-halo mass (SMHM) relation. Surprisingly, the SMHM relation must be altered even for halos masses up to 10$^{11}M_{\odot} -$ significantly above the mass scales predicted to have increased scatter from current hydrodynamical models. The sparse satellite population of this `lonely giant' thus advocates for an important modification to ideas of how the satellites around MW-mass galaxies form.