Using improved estimates for the IR luminosities of 24~$\mu$m detected sources in the GOODS-North field having deep 70~$\mu$m data taken as part of the FIDEL survey, we search for, and do not find, significant evidence for evolution in the FIR-radio correlation out to $z\sim2$.

While evolution in the FIR/radio ratios with redshift is not observed for the brightest star-forming galaxies currently detectable out to $z\sim2$, we do expect the FIR-radio correlation to deviate from the canonical ratio as Inverse Compton (IC) losses to cosmic ray electrons due to the CMB, whose energy density goes as $U_{\rm CMB} \sim(1+z)^4$, may become important relative to synchrotron cooling by a galaxy's internal magnetic field.

To combat IC losses off of the CMB at $z \geq 3.5$, magnetic field strengths need to reach values in excess of $\sim$100~$\mu$G.

The existence of strong magnetic fields in galaxies at such early epochs relative to the field strengths in galaxies today is somewhat counterintuitive; magnetic fields are thought to build up over time and not decay quickly.

However, radio quiet QSOs at $z\sim6$ appear to lie on the canonical FIR-radio relationship.

It may also be possible that the FIR emission per unit star formation is depressed for galaxies at these early epochs due to lower metallicities or other changing conditions which could lead to a FIR/radio ratio which is fortuitously consistent with the low-$z$ value.

I will discuss how the combination of deep radio continuum surveys using the SKA with data from future FIR/submm/mm facilities such as $Herschel$, ALMA, CCAT, SPICA, and eventually CALSITO/SAFIR, will help to characterize better the star-forming and magnetic field properties of IR bright ($L_{\rm IR} \geq 10^{11.5}~L_{\odot};~{\rm SFR} \geq 50~M_{\odot}~{\rm yr}^{-1}$) galaxies out to redshifts of $z\sim6$, when the Universe was only $\sim$1~Gyr old.