Capabilities for light sources in the future (but not necessarily from any single source) include: 1) the ability to produce x-ray pulses with Fourier-transform-limit time structure from the picosecond to attosecond regime, integrated and synchronized with conventional lasers, and control of longitudinal pulse shape, amplitude and phase; 2) high or full transverse coherence; 3) high average flux and brightness; 4) energy tunability in soft and hard x-ray regimes, extending to 20 keV; 5) polarization control with up to kHz switching speeds; and 6) high stability. Metrics characterizing source properties include not only average and peak spectral brightness but also the photons per pulse and repetition rate as a function of pulse length, the proximity to transform-limited dimensions in six dimensional phase space, and the less specific but over-arching requirement of maximizing the number of usable photons for experiments of the future, some of which do not require cutting-edge performance capabilities. Sources that will provide these capabilities include linac-based x-ray FELs, very low-emittance storage rings, perhaps ERLs and potentially novel sources that employ advanced acceleration techniques. We conclude in broad terms that two nominal source types will be required in the future: those having the coherence, high peak brightness and pulse length and shape control, represented by linac-based FEL sources, and those having the high repetition rate, stability, and high average but low peak brightness, represented by very low emittance storage rings and ERLs.